From mdehaan at redhat.com Fri Feb 2 23:46:24 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 02 Feb 2007 18:46:24 -0500 Subject: [et-mgmt-tools] [Cobbler] New feature -- Eliminating the need for that annoying "sync" command Message-ID: <45C3CD50.3080802@redhat.com> Previously cobbler was set up such that to apply any changes to the cobbler configuration to the server (i.e. to make it actually do anything) one had to run "cobbler sync" after any commands. In upstream code (mercurial, you can grab it now), the need for running sync is no more -- for most use cases. For each "distro add", "profile add", or "system add" command -- plus the corresponding delete commands as well -- cobbler will now update the filesystem as they are run. This enables changes to be made to one's provisioning environment without taking it offline for the usual 6-60 seconds (depending on how many distro/profiles there are) it takes to run a sync. Unless you are making changes to the kickstart files or replacing kernel/initrd files, sync now only has to be run right after first install -- and never again. This feature is enabled by default -- though it can be turned off by setting minimize_syncs to 0 in /var/lib/cobbler/settings. I don't know why anyone would want to turn it off, but I figured I'd make it possible. Check it out if you like -- Feedback is welcomed. This is slated for inclusion in 0.4.0. Thanks, Michael From David.Mackintosh at xdroop.com Wed Feb 7 15:58:26 2007 From: David.Mackintosh at xdroop.com (David Mackintosh) Date: Wed, 7 Feb 2007 10:58:26 -0500 Subject: [et-mgmt-tools] Stupid Cobbler Trick: PXE Booting SuSE-family distributions Message-ID: <20070207155826.GA6585@xdroop.com> In the spirit of using tools beyond their original intent, I figured out how to use cobbler to help automate setting up PXE install environments for SuSE-family distributions. http://wiki.xdroop.com/space/Linux/SuSE/Using+Cobbler+with+SuSE Presumably the principles here could be extended to any Linux distribution. I think it a huge complement to Mr. DeHaan that his tool can be bent to such uses beyond his original intent without any modification whatsoever. -- /\oo/\ / /()\ \ David Mackintosh | dave at xdroop.com | http://www.xdroop.com -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From mdehaan at redhat.com Wed Feb 7 17:03:18 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 07 Feb 2007 12:03:18 -0500 Subject: [et-mgmt-tools] re: Stupid Cobbler Trick: PXE Booting SuSE-family distributions Message-ID: <45CA0656.3050805@redhat.com> David Mackintosh wrot > In the spirit of using tools beyond their original intent, I figured > out how to use cobbler to help automate setting up PXE install > environments for SuSE-family distributions. > http://wiki.xdroop.com/space/Linux/SuSE/Using+Cobbler+with+SuSE > Presumably the principles here could be extended to any Linux > distribution. David, Neat stuff! Looking over the above example it seems that if you're going interactive you could just leave off the kickstart parameter rather than just passing in a filler argument. However, I have a suggestion that might be useful. If you pass the address of an autoyast file to --kickstart, cobbler will write (behind the scenes) a "ks=blah" line to the kernel command line parameters. I would expect SuSE would ignore this. However, adding a --kopts parameter that referenced the file (URL) given in --kickstart would seemingly allow passing in a SuSE answer file. Cobbler really doesn't know what a kickstart is, technically, it's just a file to which some templating voodoo can be applied, so that might be possible. From http://www.suse.com/~ug/AutoYaST_FAQ.html it appears this parameter is just "autoyast=" instead of "ks=", so that's pretty simple. This would work today. Unfortunately I have no experience generating AutoYAST files, though it appears there are GUI tools to help with this. One thing that comes to mind (for a new feature) is having a parameter on "distro" that would accomodate various breeds of distros, such that minor distribution tweaks could be made. Example: cobbler distro add --name=foo --distro=blah --kernel=blah --breed=suse Obviously breed would default to kickstartable RH/Fedora-based distros. I'd need someone to volunteer to do some SuSE testing though. Don't have any machines here :) Thoughts? Thanks for pointing out that this is doable. I really like the idea of making cobbler a bit more versatile. --Michael From mdehaan at redhat.com Wed Feb 7 17:07:53 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 07 Feb 2007 12:07:53 -0500 Subject: [et-mgmt-tools] Cobbler and zeroconf Message-ID: <45CA0769.5000502@redhat.com> I'm thinking about adding Avahi (zeroconf) support to Cobbler, such that koan can optionally auto-discover cobbler servers if they do not specify a --server parameter. Zeroconf support would need to be added in such a way that cobbler still worked on systems where Avahi wasn't easily installable -- both cobbler-side and koan-side. This could also, in the future, allow other software to discover the locations of cobbler servers that can be used as an install base. Anyone have any thoughts on this or other use cases for discovery? --Michael From mdehaan at redhat.com Wed Feb 7 17:15:17 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 07 Feb 2007 12:15:17 -0500 Subject: [et-mgmt-tools] Cobbler: having Xen images use a PXE boot loader Message-ID: <45CA0925.6060909@redhat.com> Some new work has been done on making Xen machines be able to use PXE as a boot loader. You can see more here: https://www.cs.tcd.ie/Stephen.Childs/pypxeboot/ Stay tuned, but this should make management of virt via koan a lot more interesting. Right now koan doesn't have any hooks for this, but I plan to add some. If you're capable enough with Xen, you should be able to try it now. --Michael From berrange at redhat.com Wed Feb 7 18:11:00 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 7 Feb 2007 18:11:00 +0000 Subject: [et-mgmt-tools] Stupid Cobbler Trick: PXE Booting SuSE-family distributions In-Reply-To: <20070207155826.GA6585@xdroop.com> References: <20070207155826.GA6585@xdroop.com> Message-ID: <20070207181100.GA10502@redhat.com> On Wed, Feb 07, 2007 at 10:58:26AM -0500, David Mackintosh wrote: > In the spirit of using tools beyond their original intent, I figured > out how to use cobbler to help automate setting up PXE install > environments for SuSE-family distributions. > > http://wiki.xdroop.com/space/Linux/SuSE/Using+Cobbler+with+SuSE > > Presumably the principles here could be extended to any Linux > distribution. On a similar theme. I have actually got virtinst capable of downloading and booting Xen guests off SuSE paravirtualized kernels. So once koan is redone to use virtinst underneath, it'll Suse capable for baremetal and DomU full & paravirt. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From mdehaan at redhat.com Wed Feb 7 18:17:53 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 07 Feb 2007 13:17:53 -0500 Subject: [et-mgmt-tools] Stupid Cobbler Trick: PXE Booting SuSE-family distributions In-Reply-To: <20070207181100.GA10502@redhat.com> References: <20070207155826.GA6585@xdroop.com> <20070207181100.GA10502@redhat.com> Message-ID: <45CA17D1.2040303@redhat.com> Daniel P. Berrange wrote: > On Wed, Feb 07, 2007 at 10:58:26AM -0500, David Mackintosh wrote: > >> In the spirit of using tools beyond their original intent, I figured >> out how to use cobbler to help automate setting up PXE install >> environments for SuSE-family distributions. >> >> http://wiki.xdroop.com/space/Linux/SuSE/Using+Cobbler+with+SuSE >> >> Presumably the principles here could be extended to any Linux >> distribution. >> > > On a similar theme. I have actually got virtinst capable of downloading > and booting Xen guests off SuSE paravirtualized kernels. So once koan > is redone to use virtinst underneath, it'll Suse capable for baremetal > and DomU full & paravirt. > > Dan. > Planning to do that soon (next week, possibly?) given that virt-inst is now compatible between FC-6 and FC-7. This should be goodness. --Michael From alikins at redhat.com Wed Feb 7 19:13:37 2007 From: alikins at redhat.com (Adrian Likins) Date: Wed, 07 Feb 2007 14:13:37 -0500 Subject: [et-mgmt-tools] Cobbler and zeroconf In-Reply-To: <45CA0769.5000502@redhat.com> References: <45CA0769.5000502@redhat.com> Message-ID: <45CA24E1.3080907@redhat.com> Michael DeHaan wrote: > This could also, in the future, allow other software to discover the > locations of cobbler servers that can be used as an install base. > > Anyone have any thoughts on this or other use cases for discovery? At the very least, I think driving stuff towards expecting this to at least maybe work is a good idea. Aka, get people used to the idea of actually using SRV records and the like. Adrian From John.Sanabria at ece.uprm.edu Wed Feb 7 22:18:22 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Wed, 7 Feb 2007 18:18:22 -0400 (AST) Subject: [et-mgmt-tools] cobbler server's port Message-ID: <48511.136.145.116.78.1170886702.squirrel@ece.uprm.edu> Hello, i want to access remote cobbler server accessing the profile list. The server is 136.146.116.114. The command locally runs ok, however, when i try to access from remote host, i got the next message: [root at dhcp-crl-116-78 ~]# koan --server=136.145.116.114 --list-profiles listing defined profiles... couldn't access listing information i know is a problem of firewall, because i disabled it, and it works. What's the cobbler's port? thanks a lot. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From John.Sanabria at ece.uprm.edu Thu Feb 8 04:36:01 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Thu, 8 Feb 2007 00:36:01 -0400 (AST) Subject: [et-mgmt-tools] problems with koan... Message-ID: <55179.136.145.116.78.1170909361.squirrel@ece.uprm.edu> ok, after has been resolved the problems with fedora core 6 in my laptop (compaq r3000), running xen and SELinux disabled, i try to run the koan in my laptop. The scenario as follows: my cobbler server is 136.145.116.114. I set up the next on it: # cobbler distro add --name="test1" --kernel=/boot/vmlinuz-2.6.18... --initrd=/boot/initrd-2.6.18... # cobbler profile add --name="test1-profile" --distro="test1" --kickstart=/root/anaconda-ks.cfg.virtual --virt-name="cobblertest" --virt-file-size=3 --virt-ram=250 when run koan in my laptop, as follows: # koan --virt --profile=test1-profile --server=136.145.116.114 i got the next output: - processing profile: test1-profile - fetching configuration for profile: test1-profile - url=http://136.145.116.114/cobbler/profiles/test1-profile - {'kickstart': 'http://136.145.116.114/cobbler_track/kickstarts/test1-profile/ks.cfg', 'name': 'test1-profile', 'virt_ram': 250, 'repos': '', 'kernel_options': 'lang ksdevice=eth0 text syslog=136.145.116.114:25150 devfs=nomount ramdisk_size=16438 syslog=136.145.116.114:25150', 'virt_name': 'cobblertest', 'virt_file_size': 3, 'distro': 'test1', 'virt_paravirt': 'True', 'ks_meta': ''} - fetching configuration for distro: test1 - url=http://136.145.116.114/cobbler/distros/test1 - {'kernel': '/boot/vmlinuz-2.6.18-1.2798.fc6', 'ks_meta': '', 'kernel_options': 'lang ksdevice=eth0 text devfs=nomount ramdisk_size=16438 syslog=136.145.116.114:25150', 'initrd': '/boot/initrd-2.6.18-1.2798.fc6.img', 'arch': 'x86', 'name': 'test1'} - downloading initrd initrd-2.6.18-1.2798.fc6.img to /tmp/initrd-2.6.18-1.2798.fc6.img - url=http://136.145.116.114/cobbler/images/test1/initrd-2.6.18-1.2798.fc6.img - downloading kernel vmlinuz-2.6.18-1.2798.fc6 to /tmp/vmlinuz-2.6.18-1.2798.fc6 - url=http://136.145.116.114/cobbler/images/test1/vmlinuz-2.6.18-1.2798.fc6 - kernel saved = /tmp/vmlinuz-2.6.18-1.2798.fc6 - initrd saved = /tmp/initrd-2.6.18-1.2798.fc6.img - invalid RAM size specified, defaulting to 256 MB libvir: Xen Daemon error : POST operation failed: (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') Failed to create domain cobblertest Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main k.run() File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run self.do_virt() File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in do_virt return self.do_net_install("/tmp",after_download) File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in do_net_install after_download(self, distro_data, profile_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in after_download self.do_virt_net_install(profile_data, distro_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in do_virt_net_install extra=kextra File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in start_paravirt_install dom = conn.createLinux(cfgxml, 0) File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self) libvirtError: virDomainCreateLinux() failed POST operation failed: (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') ok, looks the parameter "--virt-ram=250" is a bad idea, i remove the profile and create another one with no "--virt-ram" because default value is 512, so, according to message looks enough. I run again the koan command again and i got the next output: - processing profile: test1-profile - fetching configuration for profile: test1-profile - url=http://136.145.116.114/cobbler/profiles/test1-profile - {'kickstart': 'http://136.145.116.114/cobbler_track/kickstarts/test1-profile/ks.cfg', 'name': 'test1-profile', 'virt_ram': 250, 'repos': '', 'kernel_options': 'lang ksdevice=eth0 text syslog=136.145.116.114:25150 devfs=nomount ramdisk_size=16438 syslog=136.145.116.114:25150', 'virt_name': 'cobblertest', 'virt_file_size': 3, 'distro': 'test1', 'virt_paravirt': 'True', 'ks_meta': ''} - fetching configuration for distro: test1 - url=http://136.145.116.114/cobbler/distros/test1 - {'kernel': '/boot/vmlinuz-2.6.18-1.2798.fc6', 'ks_meta': '', 'kernel_options': 'lang ksdevice=eth0 text devfs=nomount ramdisk_size=16438 syslog=136.145.116.114:25150', 'initrd': '/boot/initrd-2.6.18-1.2798.fc6.img', 'arch': 'x86', 'name': 'test1'} - downloading initrd initrd-2.6.18-1.2798.fc6.img to /tmp/initrd-2.6.18-1.2798.fc6.img - url=http://136.145.116.114/cobbler/images/test1/initrd-2.6.18-1.2798.fc6.img - downloading kernel vmlinuz-2.6.18-1.2798.fc6 to /tmp/vmlinuz-2.6.18-1.2798.fc6 - url=http://136.145.116.114/cobbler/images/test1/vmlinuz-2.6.18-1.2798.fc6 - kernel saved = /tmp/vmlinuz-2.6.18-1.2798.fc6 - initrd saved = /tmp/initrd-2.6.18-1.2798.fc6.img - invalid RAM size specified, defaulting to 256 MB libvir: Xen Daemon error : POST operation failed: (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') Failed to create domain cobblertest Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main k.run() File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run self.do_virt() File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in do_virt return self.do_net_install("/tmp",after_download) File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in do_net_install after_download(self, distro_data, profile_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in after_download self.do_virt_net_install(profile_data, distro_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in do_virt_net_install extra=kextra File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in start_paravirt_install dom = conn.createLinux(cfgxml, 0) File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self) libvirtError: virDomainCreateLinux() failed POST operation failed: (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') Any suggestions? thanks a lot! Configurations: Server: 136.145.116.114 RAM=1Gbyte SO=Fedora Core 6 Laptop: RAM=512Mbytes SO=Fedora Core 6 both systems has disable the firewall and SELinux. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From mdehaan at redhat.com Thu Feb 8 14:22:37 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 09:22:37 -0500 Subject: [et-mgmt-tools] cobbler server's port In-Reply-To: <48511.136.145.116.78.1170886702.squirrel@ece.uprm.edu> References: <48511.136.145.116.78.1170886702.squirrel@ece.uprm.edu> Message-ID: <45CB322D.4020502@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hello, > > i want to access remote cobbler server accessing the profile list. > > The server is 136.146.116.114. The command locally runs ok, however, when > i try to access from remote host, i got the next message: > > [root at dhcp-crl-116-78 ~]# koan --server=136.145.116.114 --list-profiles > listing defined profiles... > couldn't access listing information > > i know is a problem of firewall, because i disabled it, and it works. > What's the cobbler's port? > > thanks a lot. > > > > (`'`'`'`'`) > | | > | | > | | > -----.. (()---- | > | || (_ | > | || | | > | || | | > | || /\ ..-- > '--------'' /\ ||-'' \ > / \ \ \// ,, \---. > .---------. \./ |~| /__\ \ | > ___|_________|__|""-.___ / || | | > | | .-----' || | | > | |CC.-----. | | > | | '-----' | |-ABG > | | > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > koan accesses content from Apache over port 80. --Michael From mdehaan at redhat.com Thu Feb 8 14:27:21 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 09:27:21 -0500 Subject: [et-mgmt-tools] Re: problems with koan... In-Reply-To: <55179.136.145.116.78.1170909361.squirrel@ece.uprm.edu> References: <55179.136.145.116.78.1170909361.squirrel@ece.uprm.edu> Message-ID: <45CB3349.50502@redhat.com> John.Sanabria at ece.uprm.edu wrote: > ok, > > after has been resolved the problems with fedora core 6 in my laptop > (compaq r3000), running xen and SELinux disabled, i try to run the koan in > my laptop. > > The scenario as follows: > > my cobbler server is 136.145.116.114. I set up the next on it: > > # cobbler distro add --name="test1" --kernel=/boot/vmlinuz-2.6.18... > --initrd=/boot/initrd-2.6.18... > > # cobbler profile add --name="test1-profile" --distro="test1" > --kickstart=/root/anaconda-ks.cfg.virtual --virt-name="cobblertest" > --virt-file-size=3 --virt-ram=250 > > when run koan in my laptop, as follows: > # koan --virt --profile=test1-profile --server=136.145.116.114 > > i got the next output: > > - processing profile: test1-profile > - fetching configuration for profile: test1-profile > - url=http://136.145.116.114/cobbler/profiles/test1-profile > - {'kickstart': > 'http://136.145.116.114/cobbler_track/kickstarts/test1-profile/ks.cfg', > 'name': 'test1-profile', 'virt_ram': 250, 'repos': '', 'kernel_options': > 'lang ksdevice=eth0 text syslog=136.145.116.114:25150 devfs=nomount > ramdisk_size=16438 syslog=136.145.116.114:25150', 'virt_name': > 'cobblertest', 'virt_file_size': 3, 'distro': 'test1', 'virt_paravirt': > 'True', 'ks_meta': ''} > - fetching configuration for distro: test1 > - url=http://136.145.116.114/cobbler/distros/test1 > - {'kernel': '/boot/vmlinuz-2.6.18-1.2798.fc6', 'ks_meta': '', > 'kernel_options': 'lang ksdevice=eth0 text devfs=nomount > ramdisk_size=16438 syslog=136.145.116.114:25150', 'initrd': > '/boot/initrd-2.6.18-1.2798.fc6.img', 'arch': 'x86', 'name': 'test1'} - > downloading initrd initrd-2.6.18-1.2798.fc6.img to > /tmp/initrd-2.6.18-1.2798.fc6.img > - > url=http://136.145.116.114/cobbler/images/test1/initrd-2.6.18-1.2798.fc6.img > - downloading kernel vmlinuz-2.6.18-1.2798.fc6 to > /tmp/vmlinuz-2.6.18-1.2798.fc6 > - > url=http://136.145.116.114/cobbler/images/test1/vmlinuz-2.6.18-1.2798.fc6 > - kernel saved = /tmp/vmlinuz-2.6.18-1.2798.fc6 > - initrd saved = /tmp/initrd-2.6.18-1.2798.fc6.img > - invalid RAM size specified, defaulting to 256 MB > libvir: Xen Daemon error : POST operation failed: (xend.err 'Error > creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and > shrinking to 262144 KiB would leave only 244156 KiB free.') > Failed to create domain cobblertest > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main > k.run() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run > self.do_virt() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in > do_virt > return self.do_net_install("/tmp",after_download) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in > do_net_install > after_download(self, distro_data, profile_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in > after_download > self.do_virt_net_install(profile_data, distro_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in > do_virt_net_install > extra=kextra > File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in > start_paravirt_install > dom = conn.createLinux(cfgxml, 0) > File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in > createLinux > if ret is None:raise libvirtError('virDomainCreateLinux() failed', > conn=self) > libvirtError: virDomainCreateLinux() failed POST operation failed: > (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is > 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') > > ok, looks the parameter "--virt-ram=250" is a bad idea, i remove the > profile and create another one with no "--virt-ram" because default value > is 512, so, according to message looks enough. > > I run again the koan command again and i got the next output: > - processing profile: test1-profile > - fetching configuration for profile: test1-profile > - url=http://136.145.116.114/cobbler/profiles/test1-profile > - {'kickstart': > 'http://136.145.116.114/cobbler_track/kickstarts/test1-profile/ks.cfg', > 'name': 'test1-profile', 'virt_ram': 250, 'repos': '', 'kernel_options': > 'lang ksdevice=eth0 text syslog=136.145.116.114:25150 devfs=nomount > ramdisk_size=16438 syslog=136.145.116.114:25150', 'virt_name': > 'cobblertest', 'virt_file_size': 3, 'distro': 'test1', 'virt_paravirt': > 'True', 'ks_meta': ''} > - fetching configuration for distro: test1 > - url=http://136.145.116.114/cobbler/distros/test1 > - {'kernel': '/boot/vmlinuz-2.6.18-1.2798.fc6', 'ks_meta': '', > 'kernel_options': 'lang ksdevice=eth0 text devfs=nomount > ramdisk_size=16438 syslog=136.145.116.114:25150', 'initrd': > '/boot/initrd-2.6.18-1.2798.fc6.img', 'arch': 'x86', 'name': 'test1'} - > downloading initrd initrd-2.6.18-1.2798.fc6.img to > /tmp/initrd-2.6.18-1.2798.fc6.img > - > url=http://136.145.116.114/cobbler/images/test1/initrd-2.6.18-1.2798.fc6.img > - downloading kernel vmlinuz-2.6.18-1.2798.fc6 to > /tmp/vmlinuz-2.6.18-1.2798.fc6 > - > url=http://136.145.116.114/cobbler/images/test1/vmlinuz-2.6.18-1.2798.fc6 > - kernel saved = /tmp/vmlinuz-2.6.18-1.2798.fc6 > - initrd saved = /tmp/initrd-2.6.18-1.2798.fc6.img > - invalid RAM size specified, defaulting to 256 MB > libvir: Xen Daemon error : POST operation failed: (xend.err 'Error > creating domain: I need 262144 KiB, but dom0_min_mem is 262144 and > shrinking to 262144 KiB would leave only 244156 KiB free.') > Failed to create domain cobblertest > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main > k.run() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run > self.do_virt() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in > do_virt > return self.do_net_install("/tmp",after_download) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in > do_net_install > after_download(self, distro_data, profile_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in > after_download > self.do_virt_net_install(profile_data, distro_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in > do_virt_net_install > extra=kextra > File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in > start_paravirt_install > dom = conn.createLinux(cfgxml, 0) > File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in > createLinux > if ret is None:raise libvirtError('virDomainCreateLinux() failed', > conn=self) > libvirtError: virDomainCreateLinux() failed POST operation failed: > (xend.err 'Error creating domain: I need 262144 KiB, but dom0_min_mem is > 262144 and shrinking to 262144 KiB would leave only 244156 KiB free.') > > Any suggestions? thanks a lot! > > Configurations: > > Server: 136.145.116.114 > RAM=1Gbyte > SO=Fedora Core 6 > > Laptop: > RAM=512Mbytes > SO=Fedora Core 6 > > both systems has disable the firewall and SELinux. > > > (`'`'`'`'`) > | | > | | > | | > -----.. (()---- | > | || (_ | > | || | | > | || | | > | || /\ ..-- > '--------'' /\ ||-'' \ > / \ \ \// ,, \---. > .---------. \./ |~| /__\ \ | > ___|_________|__|""-.___ / || | | > | | .-----' || | | > | |CC.-----. | | > | | '-----' | |-ABG > | | > > > > There was a bug based on koan ignoring the RAM setting from cobbler, though that was fixed a few releases ago. Can you give me the output of "cobbler profile report" as well as "rpm -q cobbler" and "rpm -q koan" ? Thanks, Michael From John.Sanabria at ece.uprm.edu Thu Feb 8 15:21:51 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Thu, 8 Feb 2007 11:21:51 -0400 (AST) Subject: [et-mgmt-tools] Re: problems with koan... In-Reply-To: <45CB3349.50502@redhat.com> References: <55179.136.145.116.78.1170909361.squirrel@ece.uprm.edu> <45CB3349.50502@redhat.com> Message-ID: <57763.136.145.116.78.1170948111.squirrel@ece.uprm.edu> > > There was a bug based on koan ignoring the RAM setting from cobbler, > though that was fixed a few releases ago. > Can you give me the output of "cobbler profile report" as well as "rpm > -q cobbler" and "rpm -q koan" ? > At 136.145.116.114 and remote site the output is: [root at dhcp-crl-116-78 ~]# rpm -q cobbler cobbler-0.3.9-1.fc6 [root at dhcp-crl-116-78 ~]# rpm -q koan koan-0.2.5-1.fc6 > Thanks, > you're welcome (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From mdehaan at redhat.com Thu Feb 8 15:35:29 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 10:35:29 -0500 Subject: [et-mgmt-tools] Re: problems with koan... In-Reply-To: <57763.136.145.116.78.1170948111.squirrel@ece.uprm.edu> References: <55179.136.145.116.78.1170909361.squirrel@ece.uprm.edu> <45CB3349.50502@redhat.com> <57763.136.145.116.78.1170948111.squirrel@ece.uprm.edu> Message-ID: <45CB4341.3010406@redhat.com> John.Sanabria at ece.uprm.edu wrote: >> There was a bug based on koan ignoring the RAM setting from cobbler, >> though that was fixed a few releases ago. >> Can you give me the output of "cobbler profile report" as well as "rpm >> -q cobbler" and "rpm -q koan" ? >> >> > > At 136.145.116.114 and remote site the output is: > > [root at dhcp-crl-116-78 ~]# rpm -q cobbler > cobbler-0.3.9-1.fc6 > [root at dhcp-crl-116-78 ~]# rpm -q koan > koan-0.2.5-1.fc6 John, Those are up to date, which is good. What may have happened, and this is very likely, is that you did not run "cobbler sync" after making the profile change on the boot server, so koan is requesting the old configuration still. I would bet that if you run "cobbler sync" on the server and then run koan again on the remote it should work fine. (Note that the new version of cobbler - 0.4.0, not yet released, will not require running sync after these operations). --Michael From ivanhhn at yahoo.com Thu Feb 8 17:08:54 2007 From: ivanhhn at yahoo.com (javier ivan mendoza) Date: Thu, 8 Feb 2007 11:08:54 -0600 (CST) Subject: [et-mgmt-tools] Program installation Message-ID: <833197.20386.qm@web30702.mail.mud.yahoo.com> Hi, my name is Javier Ivan Mendoza, from Choluteca, Honduras, Central America. I have downloaded fedora core linux 6 and installed it in a 32-bit x86 intel platform computer. I got recently a software from Xilinx, inc. the WebPack 9i for linux. I tried to install it running the script setup included in the package, but fedora comes with a utility named setup to configure some system features. I changed the name of the script but the bash shell tell that the file does not exist. What is missing? I do not know if this the correct channel for help, but I will appreciate your recommendations. Best Regards Ivan Mendoza Get Firefox! http://www.spreadfirefox.com/?q=user/register&r=50951 -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdehaan at redhat.com Thu Feb 8 18:00:43 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 13:00:43 -0500 Subject: [et-mgmt-tools] Re: Program Installation In-Reply-To: References: Message-ID: <45CB654B.6030403@redhat.com> > > ------------------------------------------------------------------------ > > Subject: > Program installation > From: > javier ivan mendoza > Date: > Thu, 8 Feb 2007 11:08:54 -0600 (CST) > To: > et-mgmt-tools at redhat.com > > To: > et-mgmt-tools at redhat.com > > > Hi, my name is Javier Ivan Mendoza, from Choluteca, Honduras, Central > America. > I have downloaded fedora core linux 6 and installed it in a 32-bit x86 > intel platform computer. I got recently a software from Xilinx, inc. > the WebPack 9i for linux. I tried to install it running the script > setup included in the package, but fedora comes with a utility named > setup to configure some system features. I changed the name of the > script but the bash shell tell that the file does not exist. > What is missing? > I do not know if this the correct channel for help, but I will > appreciate your recommendations. > > Best Regards > > Ivan Mendoza > > > > Get Firefox! > http://www.spreadfirefox.com/?q=user/register&r=50951 Get Firefox! > > ------------------------------------------------------------------------ > > Subject: > confirm cd5249208d1b8ef875302b76eb854285323f3494 > From: > et-mgmt-tools-request at redhat.com > > > If you reply to this message, keeping the Subject: header intact, > Mailman will discard the held message. Do this if the message is > spam. If you reply to this message and include an Approved: header > with the list password in it, the message will be approved for posting > to the list. The Approved: header can also appear in the first line > of the body of the reply. > This email list is for systems management software hosted on http://et.redhat.com. We do not support Xilinux. You should find a Xilinux mailing list. --Michael From John.Sanabria at ece.uprm.edu Thu Feb 8 19:47:57 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Thu, 8 Feb 2007 15:47:57 -0400 (AST) Subject: [et-mgmt-tools] xend.err "Error creating domain: (22, 'Invalid argument')" Message-ID: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> Hi, i did a littler changes respect previous messages. Now, the cobbler server is 136.145.116.78 and the machine that will be use koan is 136.145.116.114. So, i execute the steps in order to create a distro and profiles in .78 machine, and execute "cobbler sync". When i run the command "cobbler report" i got it: [root at dhcp-crl-116-78 ~]# cobbler report distro : test1 kernel : /boot/vmlinuz-2.6.19-1.2895.fc6 initrd : /boot/initrd-2.6.19-1.2895.fc6.img kernel options : architecture : x86 ks metadata : profile : test1-profile distro : test1 kickstart : /root/anaconda-ks.cfg kernel options : ks metadata : virt name : cobblertest virt file size : 3 virt ram : 512 virt paravirt : True repos : commands executed, previously: cobbler distro add --name="test1" --kernel=/boot/vmlinuz-2.6.19-1.2895.fc6 --initrd=/boot/initrd-2.6.19-1.2895.fc6.img cobbler profile add --name="test1-profile" --distro="test1" --kickstart=/root/anaconda-ks.cfg --virt-name="cobblertest" --virt-file-size=3 cobbler sync then, when i try to run the next koan command: koan --virt --server=136.145.116.78 --profile="test1-profile" at host 136.145.116.114, i got it: - processing profile: test1-profile - fetching configuration for profile: test1-profile - url=http://136.145.116.78/cobbler/profiles/test1-profile - {'kickstart': 'http://136.145.116.78/cobbler_track/kickstarts/test1-profile/ks .cfg', 'name': 'test1-profile', 'virt_ram': 512, 'repos': '', 'kernel_options': 'lang ksdevice=eth0 text syslog=136.145.116.78:25150 devfs=nomount ramdisk_size =16438 syslog=136.145.116.78:25150', 'virt_name': 'cobblertest', 'virt_file_size ': 3, 'distro': 'test1', 'virt_paravirt': 'True', 'ks_meta': ''} - fetching configuration for distro: test1 - url=http://136.145.116.78/cobbler/distros/test1 - {'kernel': '/boot/vmlinuz-2.6.19-1.2895.fc6', 'ks_meta': '', 'kernel_options': 'lang ksdevice=eth0 text devfs=nomount ramdisk_size=16438 syslog=136.145.116.7 8:25150', 'initrd': '/boot/initrd-2.6.19-1.2895.fc6.img', 'arch': 'x86', 'name': 'test1'} - downloading initrd initrd-2.6.19-1.2895.fc6.img to /tmp/initrd-2.6.19-1.2895.f c6.img - url=http://136.145.116.78/cobbler/images/test1/initrd-2.6.19-1.2895.fc6.img - downloading kernel vmlinuz-2.6.19-1.2895.fc6 to /tmp/vmlinuz-2.6.19-1.2895.fc6 - url=http://136.145.116.78/cobbler/images/test1/vmlinuz-2.6.19-1.2895.fc6 - kernel saved = /tmp/vmlinuz-2.6.19-1.2895.fc6 - initrd saved = /tmp/initrd-2.6.19-1.2895.fc6.imglibvir: Xen Daemon error : POST operation failed: (xend.err "Error creating doma in: (22, 'Invalid argument')") Failed to create domain cobblertest_1 Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main k.run() File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run self.do_virt() File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in do_virt return self.do_net_install("/tmp",after_download) File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in do_net_insta ll after_download(self, distro_data, profile_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in after_downlo ad self.do_virt_net_install(profile_data, distro_data) File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in do_virt_net_ install extra=kextra File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in start _paravirt_install dom = conn.createLinux(cfgxml, 0) File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self ) libvirtError: virDomainCreateLinux() failed POST operation failed: (xend.err "Er ror creating domain: (22, 'Invalid argument')") Thanks a lot, for any suggestion. regards. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From berrange at redhat.com Thu Feb 8 19:51:28 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Thu, 8 Feb 2007 19:51:28 +0000 Subject: [et-mgmt-tools] xend.err "Error creating domain: (22, 'Invalid argument')" In-Reply-To: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> References: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> Message-ID: <20070208195128.GB4071@redhat.com> On Thu, Feb 08, 2007 at 03:47:57PM -0400, John.Sanabria at ece.uprm.edu wrote: > Hi, > > i did a littler changes respect previous messages. > > Now, the cobbler server is 136.145.116.78 and the machine that will be use > koan is 136.145.116.114. > > So, i execute the steps in order to create a distro and profiles in .78 > machine, and execute "cobbler sync". When i run the command "cobbler > report" i got it: > > > [root at dhcp-crl-116-78 ~]# cobbler report > distro : test1 > kernel : /boot/vmlinuz-2.6.19-1.2895.fc6 > initrd : /boot/initrd-2.6.19-1.2895.fc6.img > cobbler distro add --name="test1" --kernel=/boot/vmlinuz-2.6.19-1.2895.fc6 > --initrd=/boot/initrd-2.6.19-1.2895.fc6.img This is a bare metal kernel. > then, when i try to run the next koan command: > > koan --virt --server=136.145.116.78 --profile="test1-profile" > > at host 136.145.116.114, i got it: > libvirtError: virDomainCreateLinux() failed POST operation failed: > (xend.err "Error creating domain: (22, 'Invalid argument')") But you're trying to install a paravirt guest OS. Try using a Xen kernel instead. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From mdehaan at redhat.com Thu Feb 8 19:54:39 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 14:54:39 -0500 Subject: [et-mgmt-tools] Re: xend.err "Error creating domain: (22, 'Invalid argument')" In-Reply-To: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> References: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> Message-ID: <45CB7FFF.7000302@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hi, > > i did a littler changes respect previous messages. > > Now, the cobbler server is 136.145.116.78 and the machine that will be use > koan is 136.145.116.114. > > So, i execute the steps in order to create a distro and profiles in .78 > machine, and execute "cobbler sync". When i run the command "cobbler > report" i got it: > > > [root at dhcp-crl-116-78 ~]# cobbler report > distro : test1 > kernel : /boot/vmlinuz-2.6.19-1.2895.fc6 > initrd : /boot/initrd-2.6.19-1.2895.fc6.img > kernel options : > architecture : x86 > ks metadata : > > profile : test1-profile > distro : test1 > kickstart : /root/anaconda-ks.cfg > kernel options : > ks metadata : > virt name : cobblertest > virt file size : 3 > virt ram : 512 > virt paravirt : True > repos : > > commands executed, previously: > > cobbler distro add --name="test1" --kernel=/boot/vmlinuz-2.6.19-1.2895.fc6 > --initrd=/boot/initrd-2.6.19-1.2895.fc6.img > cobbler profile add --name="test1-profile" --distro="test1" > --kickstart=/root/anaconda-ks.cfg --virt-name="cobblertest" > --virt-file-size=3 > cobbler sync > > then, when i try to run the next koan command: > > koan --virt --server=136.145.116.78 --profile="test1-profile" > > at host 136.145.116.114, i got it: > > > - processing profile: test1-profile > - fetching configuration for profile: test1-profile > - url=http://136.145.116.78/cobbler/profiles/test1-profile > - {'kickstart': > 'http://136.145.116.78/cobbler_track/kickstarts/test1-profile/ks > .cfg', 'name': 'test1-profile', 'virt_ram': 512, 'repos': '', > 'kernel_options': > 'lang ksdevice=eth0 text syslog=136.145.116.78:25150 devfs=nomount > ramdisk_size > =16438 syslog=136.145.116.78:25150', 'virt_name': 'cobblertest', > 'virt_file_size > ': 3, 'distro': 'test1', 'virt_paravirt': 'True', 'ks_meta': ''} > - fetching configuration for distro: test1 > - url=http://136.145.116.78/cobbler/distros/test1 > - {'kernel': '/boot/vmlinuz-2.6.19-1.2895.fc6', 'ks_meta': '', > 'kernel_options': > 'lang ksdevice=eth0 text devfs=nomount ramdisk_size=16438 > syslog=136.145.116.7 > 8:25150', 'initrd': '/boot/initrd-2.6.19-1.2895.fc6.img', 'arch': 'x86', > 'name': > 'test1'} > - downloading initrd initrd-2.6.19-1.2895.fc6.img to > /tmp/initrd-2.6.19-1.2895.f > c6.img > - url=http://136.145.116.78/cobbler/images/test1/initrd-2.6.19-1.2895.fc6.img > - downloading kernel vmlinuz-2.6.19-1.2895.fc6 to > /tmp/vmlinuz-2.6.19-1.2895.fc6 > - url=http://136.145.116.78/cobbler/images/test1/vmlinuz-2.6.19-1.2895.fc6 > - kernel saved = /tmp/vmlinuz-2.6.19-1.2895.fc6 > - initrd saved = /tmp/initrd-2.6.19-1.2895.fc6.imglibvir: Xen Daemon error > : POST operation failed: (xend.err "Error creating doma > in: (22, 'Invalid argument')") > Failed to create domain cobblertest_1 > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/koan/app.py", line 97, in main > k.run() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 155, in run > self.do_virt() > File "/usr/lib/python2.4/site-packages/koan/app.py", line 296, in do_virt > return self.do_net_install("/tmp",after_download) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 254, in > do_net_insta > ll > after_download(self, distro_data, profile_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 295, in > after_downlo > ad > self.do_virt_net_install(profile_data, distro_data) > File "/usr/lib/python2.4/site-packages/koan/app.py", line 595, in > do_virt_net_ > install > extra=kextra > File "/usr/lib/python2.4/site-packages/koan/virtcreate.py", line 196, in > start > _paravirt_install > dom = conn.createLinux(cfgxml, 0) > File "/usr/lib/python2.4/site-packages/libvirt.py", line 329, in > createLinux > if ret is None:raise libvirtError('virDomainCreateLinux() failed', > conn=self > ) > libvirtError: virDomainCreateLinux() failed POST operation failed: > (xend.err "Er > ror creating domain: (22, 'Invalid argument')") > > Thanks a lot, for any suggestion. regards. > > Good job, the koan issue is resolved. You now have a straight up Xen issue, which should be directly reproducible by using virt-install, so here are some initial thoughts: Initial thoughts: - Make sure you're using a Xen kernel and initrd, rather than the normal one. This comes out of the "xen" directory on a CD, not the "pxeboot" one. - Check the logfiles (/var/log/xen/*) for anything suspicious - If all else fails, ask fedora-xen at redhat.com From mdehaan at redhat.com Thu Feb 8 19:55:12 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 08 Feb 2007 14:55:12 -0500 Subject: [et-mgmt-tools] xend.err "Error creating domain: (22, 'Invalid argument')" In-Reply-To: <20070208195128.GB4071@redhat.com> References: <40402.136.145.116.78.1170964077.squirrel@ece.uprm.edu> <20070208195128.GB4071@redhat.com> Message-ID: <45CB8020.8030709@redhat.com> > > But you're trying to install a paravirt guest OS. > > Try using a Xen kernel instead. > > Regards, > Dan. > Beat me to it :) From John.Sanabria at ece.uprm.edu Thu Feb 8 20:56:43 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Thu, 8 Feb 2007 16:56:43 -0400 (AST) Subject: [et-mgmt-tools] [Out of Topic] wrong device... Message-ID: <51073.136.145.116.78.1170968203.squirrel@ece.uprm.edu> Hi, many thanks to Michael and Daniel, for your pertinent comments. Next question is nothing related to cobbler or koan, but perhaps you can help me about it. when i run the koan command (koan --virt --server=136.145.116.78 --profile="..."), at the end i got the next message: Creating root device. Mounting root filesystem. mount: could not find filesystem '/dev/root' Setting up other filesystems. Setting up new root fs setuproot: moving /dev failed: No such file or directory no fstab.sys, mounting internal defaults setuproot: error mounting /proc: No such file or directory setuproot: error mounting /sys: No such file or directory Switching to new root and running init. unmounting old /dev unmounting old /proc unmounting old /sys switchroot: mount failed: No such file or directory Kernel panic - not syncing: Attempted to kill init! It appears the installation has crashed The problem at this time, is that kernel image try to mount the root partition to device that is not available. I know the command "rdev" allows me change the root partition that kernel image expected. What's the suitable partition in order to follow with the installation? thanks a lot, regards. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From mdehaan at redhat.com Fri Feb 9 14:06:37 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 09 Feb 2007 09:06:37 -0500 Subject: [et-mgmt-tools] [Out of Topic] wrong device... In-Reply-To: <51073.136.145.116.78.1170968203.squirrel@ece.uprm.edu> References: <51073.136.145.116.78.1170968203.squirrel@ece.uprm.edu> Message-ID: <45CC7FED.80304@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hi, > > many thanks to Michael and Daniel, for your pertinent comments. > > Next question is nothing related to cobbler or koan, but perhaps you can > help me about it. > > when i run the koan command (koan --virt --server=136.145.116.78 > --profile="..."), at the end i got the next message: > > Creating root device. > Mounting root filesystem. > mount: could not find filesystem '/dev/root' > Setting up other filesystems. > Setting up new root fs > setuproot: moving /dev failed: No such file or directory > no fstab.sys, mounting internal defaults > setuproot: error mounting /proc: No such file or directory > setuproot: error mounting /sys: No such file or directory > Switching to new root and running init. > unmounting old /dev > unmounting old /proc > unmounting old /sys > switchroot: mount failed: No such file or directory > Kernel panic - not syncing: Attempted to kill init! > It appears the installation has crashed > > The problem at this time, is that kernel image try to mount the root > partition to device that is not available. I know the command "rdev" > allows me change the root partition that kernel image expected. What's the > suitable partition in order to follow with the installation? > > thanks a lot, regards. > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Offhand, no idea. Ask fedora-xen list? From mdehaan at redhat.com Fri Feb 9 15:12:03 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 09 Feb 2007 10:12:03 -0500 Subject: [et-mgmt-tools] re: Stupid Cobbler Trick: PXE Booting SuSE-family distributions In-Reply-To: <45CA0656.3050805@redhat.com> References: <45CA0656.3050805@redhat.com> Message-ID: <45CC8F43.2030908@redhat.com> Michael DeHaan wrote: > David Mackintosh wrot > >> In the spirit of using tools beyond their original intent, I figured >> out how to use cobbler to help automate setting up PXE install >> environments for SuSE-family distributions. > >> http://wiki.xdroop.com/space/Linux/SuSE/Using+Cobbler+with+SuSE > >> Presumably the principles here could be extended to any Linux >> distribution. > > > David, > > Neat stuff! > > Looking over the above example it seems that if you're going > interactive you could just leave off the kickstart parameter rather > than just passing in a filler argument. However, I have a suggestion > that might be useful. > If you pass the address of an autoyast file to --kickstart, cobbler > will write (behind the scenes) a "ks=blah" line to the kernel command > line parameters. I would expect SuSE would ignore this. However, > adding a --kopts parameter that referenced the file (URL) given in > --kickstart would seemingly allow passing in a SuSE answer file. > Cobbler really doesn't know what a kickstart is, technically, it's > just a file to which some templating voodoo can be applied, so that > might be possible. From http://www.suse.com/~ug/AutoYaST_FAQ.html > it appears this parameter is just "autoyast=" instead of "ks=", so > that's pretty simple. This would work today. > > Unfortunately I have no experience generating AutoYAST files, though > it appears there are GUI tools to help with this. > One thing that comes to mind (for a new feature) is having a parameter > on "distro" that would accomodate various breeds of distros, such that > minor distribution tweaks could be made. > > Example: > > cobbler distro add --name=foo --distro=blah --kernel=blah --breed=suse > > Obviously breed would default to kickstartable RH/Fedora-based distros. > > I'd need someone to volunteer to do some SuSE testing though. Don't > have any machines here :) > > Thoughts? > > Thanks for pointing out that this is doable. I really like the idea > of making cobbler a bit more versatile. > > --Michael > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools I've pushed the above --breed option to the Mercurial repo. > hg clone http://hg.et.redhat.com/hg/emd/applications/cobbler cobbler Testing from SuSE users would be great. Note that this only covers PXE at the moment. I do not forsee koan --replace-self ever supporting alternative breeds (the code is rather RH specific and I don't have much time to keep those other distros working), though this may come into play for use with Xen and other things that libvirt will come to support. Of course if someone is interested in getting that running, patches are accepted. The same goes for porting koan to other distros. --Michael From hbrock at redhat.com Fri Feb 9 16:05:22 2007 From: hbrock at redhat.com (Hugh Brock) Date: Fri, 09 Feb 2007 11:05:22 -0500 Subject: [et-mgmt-tools] virt-manager features and roadmap tentative draft Message-ID: <45CC9BC2.50405@redhat.com> Red Hat team members had a meeting this week to look at the work ahead for virt-manager. Based on our own thoughts and a list of outside requests, we came up with this tentative roadmap for Fedora 7 and Fedora 8. We'd like to hear from the community on additional features and modifications to these as well. Features when Issues --------------------------------------------------------------------------- Qemu and KVM backends Fedora 7 Will need libvirt support OS profiles Fedora 7 Should be able to select an OS type when installing a full-virt guest, so we can pass appropriate boot parameters (acpi on/off for example). Mouse pointer in console for PV Fedora 7 Already works in rawhide for RHEL5 and Fedora 6 guests, with configuration. Drop-down lists for PV installs Fedora 7 We would like to prepopulate the PV kernel source lists with some sensible URLs. Problem: mirror selection? Group operations Fedora 7 We would like to be able to select multiple guests at once for certain operations (shutdown, for example). UI for block/net devs Fedora 7 We need UI for adding vbds and vnics to guests. Support for this is already available in libvirt XML output Fedora 7 We would like to be able to write the libvirt guest config XML to a file, for export to other tools Connect help button Fedora 7 We need help text for all of the virt-manager screens Crash-dump button Fedora 7 Would like to have UI for crashing a guest and storing the crash dump somewhere. Support for this is already available in libvirt Managing multiple Requires libvirt remote mgmt remote hosts support * connect dialog Fedora 8 Needs some UI for remote authentication/authorization * manager window Fedora 8 Should display hosts/guests in a tree or something similar, rather than the current simple list * VNC widget performance Fedora 8 The Python vnc code runs too slowly to be usable when connecting over a network pipe. We need to replace it with a C widget. * VNC widget auth and encrypt Fedora 8 The connection from the VNC widget to the remote console needs to be secured and authenticated Dogtail tests Fedora 8 We need a full set of UI tests now that the UI is somewhat stabilized. Of course the schedule above is tentative, and we may well try to do more. There is also some effort to be put into fixing bugs. If you have any comments or requests for either the Fedora 7 or Fedora 8 time frame, please share them here! Many thanks, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From hbrock at redhat.com Fri Feb 9 16:48:52 2007 From: hbrock at redhat.com (Hugh Brock) Date: Fri, 09 Feb 2007 11:48:52 -0500 Subject: [et-mgmt-tools] virt-manager features and roadmap tentative draft In-Reply-To: <45CC9BC2.50405@redhat.com> References: <45CC9BC2.50405@redhat.com> Message-ID: <45CCA5F4.10803@redhat.com> Hugh Brock wrote: > Red Hat team members had a meeting this week to look at the work ahead > for virt-manager. Based on our own thoughts and a list of outside > requests, we came up with this tentative roadmap for Fedora 7 and Fedora > 8. We'd like to hear from the community on additional features and > modifications to these as well. > Apologies for the bad formatting of the last post. This one should be better: Features When Issues --------------------------------------------------------------------------- Qemu and KVM backends Fedora 7 Will need libvirt support OS profiles Fedora 7 Should be able to select an OS type when installing a full-virt guest, so we can pass appropriate boot parameters (acpi on/off for example). Mouse pointer in console for PV Fedora 7 Already works in rawhide for RHEL5 and Fedora 6 guests, with configuration. Drop-down lists for PV installs Fedora 7 We would like to prepopulate the PV kernel source lists with some sensible URLs. Problem: mirror selection? Group operations Fedora 7 We would like to be able to select multiple guests at once for certain operations (shutdown, for example). UI for block/net devs Fedora 7 We need UI for adding vbds and vnics to guests. Support for this is already available in libvirt XML output Fedora 7 We would like to be able to write the libvirt guest config XML to a file, for export to other tools Connect help button Fedora 7 We need help text for all of the virt-manager screens Crash-dump button Fedora 7 Would like to have UI for crashing a guest and storing the crash dump somewhere. Support for this is already available in libvirt Managing multiple Requires libvirt remote mgmt remote hosts support * connect dialog Fedora 8 Needs some UI for remote authentication/authorization * manager window Fedora 8 Should display hosts/guests in a tree or something similar, rather than the current simple list * VNC widget performance Fedora 8 The Python vnc code runs too slowly to be usable when connecting over a network pipe. We need to replace it with a C widget. * VNC widget auth and encrypt Fedora 8 The connection from the VNC widget to the remote console needs to be secured and authenticated Dogtail tests Fedora 8 We need a full set of UI tests now that the UI is somewhat stabilized. -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From hbrock at redhat.com Fri Feb 9 17:06:35 2007 From: hbrock at redhat.com (Hugh Brock) Date: Fri, 09 Feb 2007 12:06:35 -0500 Subject: [et-mgmt-tools] virt-manager features and roadmap tentative draft In-Reply-To: <45CCA5F4.10803@redhat.com> References: <45CC9BC2.50405@redhat.com> <45CCA5F4.10803@redhat.com> Message-ID: <45CCAA1B.9030804@redhat.com> Hugh Brock wrote: > Hugh Brock wrote: >> Red Hat team members had a meeting this week to look at the work ahead >> for virt-manager. Based on our own thoughts and a list of outside >> requests, we came up with this tentative roadmap for Fedora 7 and >> Fedora 8. We'd like to hear from the community on additional features >> and modifications to these as well. >> > Apologies for the bad formatting of the last two posts (Thunderbird and tabs are not behaving well for me). This one should be more readable... Features When Issues --------------------------------------------------------------------------- Qemu and KVM backends Fedora 7 Will need libvirt support OS profiles Fedora 7 Should be able to select an OS type when installing a full-virt guest, so we can pass appropriate boot parameters (acpi on/off for example). Mouse pointer in console for PV Fedora 7 Already works in rawhide for RHEL5 and Fedora 6 guests, with configuration. Drop-down lists for PV installs Fedora 7 We would like to prepopulate the PV kernel source lists with some sensible URLs. Problem: mirror selection? Group operations Fedora 7 We would like to be able to select multiple guests at once for certain operations (shutdown, for example). UI for block/net devs Fedora 7 We need UI for adding vbds and vnics to guests. Support for this is already available in libvirt XML output Fedora 7 We would like to be able to write the libvirt guest config XML to a file, for export to other tools Connect help button Fedora 7 We need help text for all of the virt-manager screens Crash-dump button Fedora 7 Would like to have UI for crashing a guest and storing the crash dump somewhere. Support for this is already available in libvirt Managing multiple Requires libvirt remote mgmt remote hosts support * connect dialog Fedora 8 Needs some UI for remote authentication/authorization * manager window Fedora 8 Should display hosts/guests in a tree or something similar, rather than the current simple list * VNC widget performance Fedora 8 The Python vnc code runs too slowly to be usable when connecting over a network pipe. We need to replace it with a C widget. * VNC widget auth and encrypt Fedora 8 The connection from the VNC widget to the remote console needs to be secured and authenticated Dogtail tests Fedora 8 We need a full set of UI tests now that the UI is somewhat stabilized. --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From John.Sanabria at ece.uprm.edu Mon Feb 12 14:51:13 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Mon, 12 Feb 2007 10:51:13 -0400 (AST) Subject: [et-mgmt-tools] virt-install question... Message-ID: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> Hi, i'm using virt-install and it runs successfully with that command: virt-install -n xengrid2 --vcpus=1 -r 256 -s 4 -f /media/disk/iso/xengrid/xengrid2 -l ftp://mirrors.hpcf.upr.edu/pub/Mirrors/redhat/download.fedora.redhat.com/6/i386/os -p --nographics now, i have a question. What's the kernel image and initrd file used for virt-install in order to carry out the installation process? I need that information because i want use the cobbler/koan tools in order to provision virtual machines remotely, but when i submit the command koan --virt --server=136.145.116.114 --profile="xentemplate-profile" i got that: XENBUS: Device with no driver: device/vbd/51712 XENBUS: Device with no driver: device/vif/0 Freeing unused kernel memory: 184k freed Write protecting the kernel read-only data: 387k Red Hat nash version 5.1.19.0.2 starting Mounting proc filesystem Mounting sysfs filesystem Creating /dev Creating initial device nodes Setting up hotplug. Creating block device nodes. Loading uhci-hcd.ko module USB Universal Host Controller Interface driver v3.0 Loading ohci-hcd.ko module Loading ehci-hcd.ko module Loading jbd.ko module Loading ext3.ko module Loading scsi_mod.ko module SCSI subsystem initialized Loading sd_mod.ko module Loading libata.ko module Loading ata_piix.ko module Creating root device. Mounting root filesystem. mount: could not find filesystem '/dev/root' Setting up other filesystems. Setting up new root fs setuproot: moving /dev failed: No such file or directory no fstab.sys, mounting internal defaults setuproot: error mounting /proc: No such file or directory setuproot: error mounting /sys: No such file or directory Switching to new root and running init. unmounting old /dev unmounting old /proc unmounting old /sys switchroot: mount failed: No such file or directory Kernel panic - not syncing: Attempted to kill init! It appears the installation has crashed i'm using the kernel used by domain0 in my machine. thanks a lot for your comments. regards. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From mdehaan at redhat.com Mon Feb 12 14:58:03 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 12 Feb 2007 09:58:03 -0500 Subject: [et-mgmt-tools] virt-install question... In-Reply-To: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> Message-ID: <45D0807B.7060207@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hi, > > i'm using virt-install and it runs successfully with that command: > > virt-install -n xengrid2 --vcpus=1 -r 256 -s 4 -f > /media/disk/iso/xengrid/xengrid2 -l > ftp://mirrors.hpcf.upr.edu/pub/Mirrors/redhat/download.fedora.redhat.com/6/i386/os > -p --nographics > > now, i have a question. What's the kernel image and initrd file used for > virt-install in order to carry out the installation process? I need that > information because i want use the cobbler/koan tools in order to > provision virtual machines remotely, but when i submit the command > > koan --virt --server=136.145.116.114 --profile="xentemplate-profile" > The kernel and initrd can be found in ftp://mirrors.hpcf.upr.edu/pub/Mirrors/redhat/download.fedora.redhat.com/6/i386/os/images/xen/ From berrange at redhat.com Mon Feb 12 14:59:58 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Mon, 12 Feb 2007 14:59:58 +0000 Subject: [et-mgmt-tools] virt-install question... In-Reply-To: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> Message-ID: <20070212145958.GD21671@redhat.com> On Mon, Feb 12, 2007 at 10:51:13AM -0400, John.Sanabria at ece.uprm.edu wrote: > Hi, > > i'm using virt-install and it runs successfully with that command: > > virt-install -n xengrid2 --vcpus=1 -r 256 -s 4 -f > /media/disk/iso/xengrid/xengrid2 -l > ftp://mirrors.hpcf.upr.edu/pub/Mirrors/redhat/download.fedora.redhat.com/6/i386/os > -p --nographics > > now, i have a question. What's the kernel image and initrd file used for > virt-install in order to carry out the installation process? When you supply a NFS/HTTP/FTP url to virt-install, it will take that URL and try to fetch the 2 files $URL/images/xen/vmlinuz and $URL/images/xen/initrd.img for the remote site. It saves those two files to /var/lib/xen/images and then boots the VM - the files are removed once the VM has started. > I need that > information because i want use the cobbler/koan tools in order to > provision virtual machines remotely, but when i submit the command You'll want to grab the vmlinuz & initrd.img files out of $URL/images/xen and give them to cobbler when setting up the profile. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From hbrock at redhat.com Mon Feb 12 15:19:27 2007 From: hbrock at redhat.com (Hugh Brock) Date: Mon, 12 Feb 2007 10:19:27 -0500 Subject: [et-mgmt-tools] virt-manager choose-your-os screen Message-ID: <45D0857F.9050308@redhat.com> Hi all. I am thinking about adding a bit to the full-virt page of the create wizard that will allow users to define which OS+variant they are installing. This will let us do useful things in the install (like set acpi off for Windows, for example). What I'm wondering is what other things we might want to be able to do on a full-virt install or boot based on OS type. Things that came to mind were: 1. Set APIC and ACPI boot flags as described above 2. Suggest what kind of mouse driver to use (to the extent we can) Any other os-specific things we might want to to at boot-time? Take care, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From John.Sanabria at ece.uprm.edu Mon Feb 12 15:46:53 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Mon, 12 Feb 2007 11:46:53 -0400 (AST) Subject: [et-mgmt-tools] virt-install question... In-Reply-To: <20070212145958.GD21671@redhat.com> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> <20070212145958.GD21671@redhat.com> Message-ID: <33595.136.145.116.114.1171295213.squirrel@ece.uprm.edu> Thanks a lot, now it works! regards. > On Mon, Feb 12, 2007 at 10:51:13AM -0400, John.Sanabria at ece.uprm.edu > wrote: >> Hi, >> >> i'm using virt-install and it runs successfully with that command: >> >> virt-install -n xengrid2 --vcpus=1 -r 256 -s 4 -f >> /media/disk/iso/xengrid/xengrid2 -l >> ftp://mirrors.hpcf.upr.edu/pub/Mirrors/redhat/download.fedora.redhat.com/6/i386/os >> -p --nographics >> >> now, i have a question. What's the kernel image and initrd file used for >> virt-install in order to carry out the installation process? > > When you supply a NFS/HTTP/FTP url to virt-install, it will take that URL > and try to fetch the 2 files $URL/images/xen/vmlinuz and > $URL/images/xen/initrd.img > for the remote site. It saves those two files to /var/lib/xen/images and > then > boots the VM - the files are removed once the VM has started. > >> I need that >> information because i want use the cobbler/koan tools in order to >> provision virtual machines remotely, but when i submit the command > > You'll want to grab the vmlinuz & initrd.img files out of $URL/images/xen > and give them to cobbler when setting up the profile. > > Dan. > -- > |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 > -=| > |=- Perl modules: http://search.cpan.org/~danberr/ > -=| > |=- Projects: http://freshmeat.net/~danielpb/ > -=| > |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 > -=| > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From John.Sanabria at ece.uprm.edu Mon Feb 12 15:50:10 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Mon, 12 Feb 2007 11:50:10 -0400 (AST) Subject: [et-mgmt-tools] kickstart file issue In-Reply-To: <45D0807B.7060207@redhat.com> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> <45D0807B.7060207@redhat.com> Message-ID: <58396.136.145.116.114.1171295410.squirrel@ece.uprm.edu> Hi, i had created a kickstart file, with kickstart tool. When i create the profile to consider it i submit next command, and corresponding output: [root at reboot xentemplate]# cobbler profile add --name="xentemplate-profile" --distro="xentemplate" --kickstart=/root/another.cfg --virt-name="xentemplate" --virt-file-size=3 --virt-ram=256 [root at reboot xentemplate]# cobbler sync del /var/www/cobbler/images del /var/www/cobbler/profiles del /var/www/cobbler/systems del /var/www/cobbler/kickstarts del /var/www/cobbler/kickstarts_sys del /var/www/cobbler/distros del /tftpboot/pxelinux.cfg del /tftpboot/images koan path = loader path = /var/lib/cobbler/elilo-3.6-ia64.efi loader new name = elilo-3.6-ia64.efi destpath = /tftpboot/elilo-3.6-ia64.efi /var/lib/cobbler/elilo-3.6-ia64.efi -> /tftpboot/elilo-3.6-ia64.efi loader path = /usr/lib/syslinux/pxelinux.0 loader new name = pxelinux.0 destpath = /tftpboot/pxelinux.0 /usr/lib/syslinux/pxelinux.0 -> /tftpboot/pxelinux.0 /root/vmlinuz -> /tftpboot/images/xentemplate/vmlinuz /root/initrd.img -> /tftpboot/images/xentemplate/initrd.img /root/vmlinuz -> /var/www/cobbler/images/xentemplate/vmlinuz /root/initrd.img -> /var/www/cobbler/images/xentemplate/initrd.img Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 753, in mkdir return os.makedirs(path,mode) File "/usr/lib/python2.4/os.py", line 159, in makedirs mkdir(name, mode) OSError: [Errno 2] No such file or directory: 'ks.cfg' 2 Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 350, in validate_kickstarts_per_profile self.apply_template(kickstart_path, meta, dest) File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 440, in apply_template self.mkdir(os.path.basename(out_path)) File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 758, in mkdir raise cexceptions.CobblerException("no_create", path) CobblerException: 'cobbler could not create: ks.cfg' Error while mirroring kickstart file (/root/another.cfg) to (/var/www/cobbler/kickstarts/xentemplate-profile/ks.cfg) what am i doing wrong? thanks a lot. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From mdehaan at redhat.com Mon Feb 12 16:25:19 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 12 Feb 2007 11:25:19 -0500 Subject: [et-mgmt-tools] kickstart file issue In-Reply-To: <58396.136.145.116.114.1171295410.squirrel@ece.uprm.edu> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> <45D0807B.7060207@redhat.com> <58396.136.145.116.114.1171295410.squirrel@ece.uprm.edu> Message-ID: <45D094EF.3040309@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hi, > > i had created a kickstart file, with kickstart tool. When i create the > profile to consider it i submit next command, and corresponding output: > > [root at reboot xentemplate]# cobbler profile add > --name="xentemplate-profile" --distro="xentemplate" > --kickstart=/root/another.cfg --virt-name="xentemplate" > --virt-file-size=3 --virt-ram=256 > [root at reboot xentemplate]# cobbler sync > del /var/www/cobbler/images > del /var/www/cobbler/profiles > del /var/www/cobbler/systems > del /var/www/cobbler/kickstarts > del /var/www/cobbler/kickstarts_sys > del /var/www/cobbler/distros > del /tftpboot/pxelinux.cfg > del /tftpboot/images > koan path = > loader path = /var/lib/cobbler/elilo-3.6-ia64.efi > loader new name = elilo-3.6-ia64.efi > destpath = /tftpboot/elilo-3.6-ia64.efi > /var/lib/cobbler/elilo-3.6-ia64.efi -> /tftpboot/elilo-3.6-ia64.efi > loader path = /usr/lib/syslinux/pxelinux.0 > loader new name = pxelinux.0 > destpath = /tftpboot/pxelinux.0 > /usr/lib/syslinux/pxelinux.0 -> /tftpboot/pxelinux.0 > /root/vmlinuz -> /tftpboot/images/xentemplate/vmlinuz > /root/initrd.img -> /tftpboot/images/xentemplate/initrd.img > /root/vmlinuz -> /var/www/cobbler/images/xentemplate/vmlinuz > /root/initrd.img -> /var/www/cobbler/images/xentemplate/initrd.img > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line > 753, in mkdir > return os.makedirs(path,mode) > File "/usr/lib/python2.4/os.py", line 159, in makedirs > mkdir(name, mode) > OSError: [Errno 2] No such file or directory: 'ks.cfg' > 2 > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line > 350, in validate_kickstarts_per_profile > self.apply_template(kickstart_path, meta, dest) > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line > 440, in apply_template > self.mkdir(os.path.basename(out_path)) > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line > 758, in mkdir > raise cexceptions.CobblerException("no_create", path) > CobblerException: 'cobbler could not create: ks.cfg' > Error while mirroring kickstart file (/root/another.cfg) to > (/var/www/cobbler/kickstarts/xentemplate-profile/ks.cfg) > > what am i doing wrong? thanks a lot. > > (`'`'`'`'`) > | | > | | > | | > -----.. (()---- | > | || (_ | > | || | | > | || | | > | || /\ ..-- > '--------'' /\ ||-'' \ > / \ \ \// ,, \---. > .---------. \./ |~| /__\ \ | > ___|_________|__|""-.___ / || | | > | | .-----' || | | > | |CC.-----. | | > | | '-----' | |-ABG > | | > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > The tracebacks look a bit odd. Can you verify that /root/anaconda.ks is a file and that /var/www/cobbler/kickstarts and /var/www/cobbler/kickstarts/xentemplate-profile are directories? If you hand edit the configuration (/var/lib/cobbler/profiles) it is possible to change the kickstart parameter to a file that is no longer present, and that might cause an error. This is usually the cause of such problems. Also, it's a longshot -- but is SELinux enabled? Cobbler generally is supposed to play nice with SELinux, but that can occasionally be a cause of permissions errors if your configuration is a bit strange. If I had to bet on anything, it's that the file was edited between the first command and the second -- though you don't seem to indicate this. If you're competent with python, adding some extra output to the functions mentioned in the traceback may help you pin it down. I'd be interested to see what you might come up with. --Michael From mdehaan at redhat.com Mon Feb 12 16:29:53 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 12 Feb 2007 11:29:53 -0500 Subject: [et-mgmt-tools] kickstart file issue In-Reply-To: <45D094EF.3040309@redhat.com> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> <45D0807B.7060207@redhat.com> <58396.136.145.116.114.1171295410.squirrel@ece.uprm.edu> <45D094EF.3040309@redhat.com> Message-ID: <45D09601.8080406@redhat.com> Michael DeHaan wrote: > John.Sanabria at ece.uprm.edu wrote: >> Hi, >> >> i had created a kickstart file, with kickstart tool. When i create the >> profile to consider it i submit next command, and corresponding output: >> >> [root at reboot xentemplate]# cobbler profile add >> --name="xentemplate-profile" --distro="xentemplate" >> --kickstart=/root/another.cfg --virt-name="xentemplate" >> --virt-file-size=3 --virt-ram=256 >> [root at reboot xentemplate]# cobbler sync >> del /var/www/cobbler/images >> del /var/www/cobbler/profiles >> del /var/www/cobbler/systems >> del /var/www/cobbler/kickstarts >> del /var/www/cobbler/kickstarts_sys >> del /var/www/cobbler/distros >> del /tftpboot/pxelinux.cfg >> del /tftpboot/images >> koan path = >> loader path = /var/lib/cobbler/elilo-3.6-ia64.efi >> loader new name = elilo-3.6-ia64.efi >> destpath = /tftpboot/elilo-3.6-ia64.efi >> /var/lib/cobbler/elilo-3.6-ia64.efi -> /tftpboot/elilo-3.6-ia64.efi >> loader path = /usr/lib/syslinux/pxelinux.0 >> loader new name = pxelinux.0 >> destpath = /tftpboot/pxelinux.0 >> /usr/lib/syslinux/pxelinux.0 -> /tftpboot/pxelinux.0 >> /root/vmlinuz -> /tftpboot/images/xentemplate/vmlinuz >> /root/initrd.img -> /tftpboot/images/xentemplate/initrd.img >> /root/vmlinuz -> /var/www/cobbler/images/xentemplate/vmlinuz >> /root/initrd.img -> /var/www/cobbler/images/xentemplate/initrd.img >> Traceback (most recent call last): >> File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line >> 753, in mkdir >> return os.makedirs(path,mode) >> File "/usr/lib/python2.4/os.py", line 159, in makedirs >> mkdir(name, mode) >> OSError: [Errno 2] No such file or directory: 'ks.cfg' >> 2 >> Traceback (most recent call last): >> File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line >> 350, in validate_kickstarts_per_profile >> self.apply_template(kickstart_path, meta, dest) >> File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line >> 440, in apply_template >> self.mkdir(os.path.basename(out_path)) >> File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line >> 758, in mkdir >> raise cexceptions.CobblerException("no_create", path) >> CobblerException: 'cobbler could not create: ks.cfg' >> Error while mirroring kickstart file (/root/another.cfg) to >> (/var/www/cobbler/kickstarts/xentemplate-profile/ks.cfg) >> >> what am i doing wrong? thanks a lot. >> >> (`'`'`'`'`) >> | | >> | | >> | | >> -----.. (()---- | >> | || (_ | >> | || | | >> | || | | >> | || /\ ..-- >> '--------'' /\ ||-'' \ >> / \ \ \// ,, \---. >> .---------. \./ |~| /__\ \ | >> ___|_________|__|""-.___ / || | | >> | | .-----' || | | >> | |CC.-----. | | >> | | '-----' | |-ABG >> | | >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > > The tracebacks look a bit odd. Can you verify that /root/anaconda.ks > is a file and that /var/www/cobbler/kickstarts and > /var/www/cobbler/kickstarts/xentemplate-profile are directories? I mean /root/another.cfg here -- the file you specified on the cobbler command line. cobbler should reject invalid input to --kickstart in either case though. > > If you hand edit the configuration (/var/lib/cobbler/profiles) it is > possible to change the kickstart parameter to a file that > is no longer present, and that might cause an error. This is usually > the cause of such problems. > Also, it's a longshot -- but is SELinux enabled? Cobbler generally > is supposed to play nice with SELinux, but that can occasionally be a > cause of permissions errors if your configuration is a bit strange. > > If I had to bet on anything, it's that the file was edited between the > first command and the second -- though you don't seem to indicate this. > If you're competent with python, adding some extra output to the > functions mentioned in the traceback may help you pin it down. I'd > be interested to see what you might come up with. > > --Michael > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools From John.Sanabria at ece.uprm.edu Mon Feb 12 20:43:40 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Mon, 12 Feb 2007 16:43:40 -0400 (AST) Subject: [et-mgmt-tools] kickstart file issue In-Reply-To: <45D094EF.3040309@redhat.com> References: <36596.136.145.116.114.1171291873.squirrel@ece.uprm.edu> <45D0807B.7060207@redhat.com><58396.136.145.116.114.1171295410.squirrel@ece.uprm.edu> <45D094EF.3040309@redhat.com> Message-ID: <1730.136.145.116.78.1171313020.squirrel@ece.uprm.edu> I sit again and i can't reproduce the error again, however, i put the new .cfg file in /etc/cobbler directoy, create again the profile, and run the koan command and now is creating a new virtual machine :-), thanks a lot for your support and tool. regards. > > The tracebacks look a bit odd. Can you verify that /root/anaconda.ks > is a file and that /var/www/cobbler/kickstarts and > /var/www/cobbler/kickstarts/xentemplate-profile are directories? > > If you hand edit the configuration (/var/lib/cobbler/profiles) it is > possible to change the kickstart parameter to a file that > is no longer present, and that might cause an error. This is usually > the cause of such problems. > > Also, it's a longshot -- but is SELinux enabled? Cobbler generally is > supposed to play nice with SELinux, but that can occasionally be a cause > of permissions errors if your configuration is a bit strange. > > If I had to bet on anything, it's that the file was edited between the > first command and the second -- though you don't seem to indicate this. > If you're competent with python, adding some extra output to the > functions mentioned in the traceback may help you pin it down. I'd be > interested to see what you might come up with. > > --Michael > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From John.Sanabria at ece.uprm.edu Tue Feb 13 20:18:19 2007 From: John.Sanabria at ece.uprm.edu (John.Sanabria at ece.uprm.edu) Date: Tue, 13 Feb 2007 16:18:19 -0400 (AST) Subject: [et-mgmt-tools] where's defined the output directory? Message-ID: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> Hi, i already built virtual boxes with koan. Question, there exist a parameter where can specify the directory where virtual image will be located? I assume by default is /var/lib/virtimages thanks, PD: Sorry, i'm not python programmer, but i did a grep in directory /usr/lib/python2.4/site-packages/cobbler for "virtimages" word and i can't found anything. (`'`'`'`'`) | | | | | | -----.. (()---- | | || (_ | | || | | | || | | | || /\ ..-- '--------'' /\ ||-'' \ / \ \ \// ,, \---. .---------. \./ |~| /__\ \ | ___|_________|__|""-.___ / || | | | | .-----' || | | | |CC.-----. | | | | '-----' | |-ABG | | From hbrock at redhat.com Tue Feb 13 20:28:48 2007 From: hbrock at redhat.com (Hugh Brock) Date: Tue, 13 Feb 2007 15:28:48 -0500 Subject: [et-mgmt-tools] specifying os-type and os-variant on the command line Message-ID: <45D21F80.9090107@redhat.com> I have just added code that allows callers to specify os-type and os-variant from the virt-install command line (the legal types are in the header of FullVirtGuest.py). If the type is unspecified it will default to Other/Other. The primary purpose of this code at the moment is to allow users to control the acpi and apic "feature" settings for fully virtualized guests, although I can imagine we might add other os-specific settings later. The changeset also allows callers to pass --noapic and --noacpi directly from the virt-install command line, overriding the os-specific setting. See http://hg.et.redhat.com/virt/applications/virtinst--devel?cs=23382c1d1495 for the code... Take care, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mdehaan at redhat.com Tue Feb 13 20:40:53 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 13 Feb 2007 15:40:53 -0500 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> Message-ID: <45D22255.8010409@redhat.com> John.Sanabria at ece.uprm.edu wrote: > Hi, > > i already built virtual boxes with koan. Question, there exist a parameter > where can specify the directory where virtual image will be located? I > assume by default is /var/lib/virtimages > > thanks, > > PD: Sorry, i'm not python programmer, but i did a grep in directory > /usr/lib/python2.4/site-packages/cobbler for "virtimages" word and i can't > found anything. > > > (`'`'`'`'`) > | | > | | > | | > -----.. (()---- | > | || (_ | > | || | | > | || | | > | || /\ ..-- > '--------'' /\ ||-'' \ > / \ \ \// ,, \---. > .---------. \./ |~| /__\ \ | > ___|_________|__|""-.___ / || | | > | | .-----' || | | > | |CC.-----. | | > | | '-----' | |-ABG > | | > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Currently this is hardcoded in koan's source to /var/lib/virtimages. Adding a /var/lib/virtimages directory symlink in your kickstart's %post when provisioning the dom0's might be a good idea, if you wanted to use storage not on /var. Or virtimages could be a mount point... either way. I'll look into adding some options for this. --Michael From hbrock at redhat.com Tue Feb 13 20:51:40 2007 From: hbrock at redhat.com (Hugh Brock) Date: Tue, 13 Feb 2007 15:51:40 -0500 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <45D22255.8010409@redhat.com> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> <45D22255.8010409@redhat.com> Message-ID: <45D224DC.5090604@redhat.com> Michael DeHaan wrote: > John.Sanabria at ece.uprm.edu wrote: >> Hi, >> >> i already built virtual boxes with koan. Question, there exist a >> parameter >> where can specify the directory where virtual image will be located? I >> assume by default is /var/lib/virtimages >> >> thanks, >> >> PD: Sorry, i'm not python programmer, but i did a grep in directory >> /usr/lib/python2.4/site-packages/cobbler for "virtimages" word and i >> can't >> found anything. >> >> >> (`'`'`'`'`) >> | | >> | | >> | | >> -----.. (()---- | >> | || (_ | >> | || | | >> | || | | >> | || /\ ..-- >> '--------'' /\ ||-'' \ >> / \ \ \// ,, \---. >> .---------. \./ |~| /__\ \ | >> ___|_________|__|""-.___ / || | | >> | | .-----' || | | >> | |CC.-----. | | >> | | '-----' | |-ABG >> | | >> >> >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > > Currently this is hardcoded in koan's source to /var/lib/virtimages. > Adding a /var/lib/virtimages directory symlink in your kickstart's %post > when provisioning the dom0's might be a good idea, if you wanted to use > storage not on /var. Or virtimages could be a mount point... either way. > > I'll look into adding some options for this. > --Michael If you're referring to Xen filesystem images, they should really go in /var/lib/xen/images to keep selinux from complaining. Does anyone know if there's a consensus directory for virtual filesystem images for for other hypervisors (qemu, kvm)? --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mdehaan at redhat.com Tue Feb 13 21:09:54 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 13 Feb 2007 16:09:54 -0500 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <45D224DC.5090604@redhat.com> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> <45D22255.8010409@redhat.com> <45D224DC.5090604@redhat.com> Message-ID: <45D22922.40400@redhat.com> Hugh Brock wrote: > Michael DeHaan wrote: >> John.Sanabria at ece.uprm.edu wrote: >>> Hi, >>> >>> i already built virtual boxes with koan. Question, there exist a >>> parameter >>> where can specify the directory where virtual image will be located? I >>> assume by default is /var/lib/virtimages >>> >>> thanks, >>> >>> PD: Sorry, i'm not python programmer, but i did a grep in directory >>> /usr/lib/python2.4/site-packages/cobbler for "virtimages" word and i >>> can't >>> found anything. >>> >>> >>> (`'`'`'`'`) >>> | | >>> | | >>> | | >>> -----.. (()---- | >>> | || (_ | >>> | || | | >>> | || | | >>> | || /\ ..-- >>> '--------'' /\ ||-'' \ >>> / \ \ \// ,, \---. >>> .---------. \./ |~| /__\ \ | >>> ___|_________|__|""-.___ / || | | >>> | | .-----' || | | >>> | |CC.-----. | | >>> | | '-----' | |-ABG >>> | | >>> >>> >>> >>> _______________________________________________ >>> et-mgmt-tools mailing list >>> et-mgmt-tools at redhat.com >>> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >>> >> >> Currently this is hardcoded in koan's source to /var/lib/virtimages. >> Adding a /var/lib/virtimages directory symlink in your kickstart's >> %post when provisioning the dom0's might be a good idea, if you >> wanted to use storage not on /var. Or virtimages could be a mount >> point... either way. >> >> I'll look into adding some options for this. >> --Michael > > If you're referring to Xen filesystem images, they should really go in > /var/lib/xen/images to keep selinux from complaining. Does anyone know > if there's a consensus directory for virtual filesystem images for for > other hypervisors (qemu, kvm)? > > --Hugh > Trueness, I will make it so. I think I had /var/lib/xen[/|_]images at one point also but decided against it given that we thought xen itself owned that path. It would be nice if SELinux just cared about, say, context instead of paths in this case :) --Michael From mdehaan at redhat.com Tue Feb 13 21:20:44 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 13 Feb 2007 16:20:44 -0500 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <45D22922.40400@redhat.com> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> <45D22255.8010409@redhat.com> <45D224DC.5090604@redhat.com> <45D22922.40400@redhat.com> Message-ID: <45D22BAC.6010103@redhat.com> Michael DeHaan wrote: > Hugh Brock wrote: >> Michael DeHaan wrote: >>> John.Sanabria at ece.uprm.edu wrote: >>>> Hi, >>>> >>>> i already built virtual boxes with koan. Question, there exist a >>>> parameter >>>> where can specify the directory where virtual image will be located? I >>>> assume by default is /var/lib/virtimages >>>> >>>> thanks, >>>> >>>> PD: Sorry, i'm not python programmer, but i did a grep in directory >>>> /usr/lib/python2.4/site-packages/cobbler for "virtimages" word and >>>> i can't >>>> found anything. >>>> >>>> >>>> (`'`'`'`'`) >>>> | | >>>> | | >>>> | | >>>> -----.. (()---- | >>>> | || (_ | >>>> | || | | >>>> | || | | >>>> | || /\ ..-- >>>> '--------'' /\ ||-'' \ >>>> / \ \ \// ,, \---. >>>> .---------. \./ |~| /__\ \ | >>>> ___|_________|__|""-.___ / || | | >>>> | | .-----' || | | >>>> | |CC.-----. | | >>>> | | '-----' | |-ABG >>>> | | >>>> >>>> >>>> >>>> _______________________________________________ >>>> et-mgmt-tools mailing list >>>> et-mgmt-tools at redhat.com >>>> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >>>> >>> >>> Currently this is hardcoded in koan's source to /var/lib/virtimages. >>> Adding a /var/lib/virtimages directory symlink in your kickstart's >>> %post when provisioning the dom0's might be a good idea, if you >>> wanted to use storage not on /var. Or virtimages could be a mount >>> point... either way. >>> >>> I'll look into adding some options for this. >>> --Michael >> >> If you're referring to Xen filesystem images, they should really go >> in /var/lib/xen/images to keep selinux from complaining. Does anyone >> know if there's a consensus directory for virtual filesystem images >> for for other hypervisors (qemu, kvm)? >> >> --Hugh >> > > Trueness, I will make it so. I think I had /var/lib/xen[/|_]images > at one point also but decided against it given that we thought xen > itself owned that path. > > It would be nice if SELinux just cared about, say, context instead of > paths in this case :) Actually you can mount with different SELinux context's, so that's my fault. I'll probably just add a flag for koan to do an override on the save location. > > --Michael > > > > From berrange at redhat.com Tue Feb 13 21:23:53 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 13 Feb 2007 21:23:53 +0000 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <45D22922.40400@redhat.com> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> <45D22255.8010409@redhat.com> <45D224DC.5090604@redhat.com> <45D22922.40400@redhat.com> Message-ID: <20070213212353.GE13577@redhat.com> On Tue, Feb 13, 2007 at 04:09:54PM -0500, Michael DeHaan wrote: > Hugh Brock wrote: > >Michael DeHaan wrote: > >>John.Sanabria at ece.uprm.edu wrote: > >>>Hi, > >>> > >>>i already built virtual boxes with koan. Question, there exist a > >>>parameter > >>>where can specify the directory where virtual image will be located? I > >>>assume by default is /var/lib/virtimages > >>> > >>>thanks, > >>> > >>>PD: Sorry, i'm not python programmer, but i did a grep in directory > >>>/usr/lib/python2.4/site-packages/cobbler for "virtimages" word and i > >>>can't > >>>found anything. > >>> [snip stupidly long signature ascii art crap] > >>Currently this is hardcoded in koan's source to /var/lib/virtimages. > >>Adding a /var/lib/virtimages directory symlink in your kickstart's > >>%post when provisioning the dom0's might be a good idea, if you > >>wanted to use storage not on /var. Or virtimages could be a mount > >>point... either way. > >> > >>I'll look into adding some options for this. > >>--Michael > > > >If you're referring to Xen filesystem images, they should really go in > >/var/lib/xen/images to keep selinux from complaining. Does anyone know > >if there's a consensus directory for virtual filesystem images for for > >other hypervisors (qemu, kvm)? No defined location for other hypervisors yet. > Trueness, I will make it so. I think I had /var/lib/xen[/|_]images at > one point also but decided against it given that we thought xen itself > owned that path. The Xen RPM does own this directory. > It would be nice if SELinux just cared about, say, context instead of > paths in this case :) SELinux does only care about context. It is the policy which says which paths map to which contexts, and in current policy we have the mapping from xen_image_t going to /var/lib/xen/images. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From kmacmill at redhat.com Tue Feb 13 21:36:34 2007 From: kmacmill at redhat.com (Karl MacMillan) Date: Tue, 13 Feb 2007 16:36:34 -0500 Subject: [et-mgmt-tools] where's defined the output directory? In-Reply-To: <20070213212353.GE13577@redhat.com> References: <2796.136.145.116.78.1171397899.squirrel@ece.uprm.edu> <45D22255.8010409@redhat.com> <45D224DC.5090604@redhat.com> <45D22922.40400@redhat.com> <20070213212353.GE13577@redhat.com> Message-ID: <45D22F62.9080001@redhat.com> Daniel P. Berrange wrote: > On Tue, Feb 13, 2007 at 04:09:54PM -0500, Michael DeHaan wrote: >> It would be nice if SELinux just cared about, say, context instead of >> paths in this case :) > > SELinux does only care about context. It is the policy which says which > paths map to which contexts, and in current policy we have the mapping > from xen_image_t going to /var/lib/xen/images. > Though it is easy to add additional paths via a bz against policy or (just locally) the semanage / restorecon commands. Karl From mdehaan at redhat.com Tue Feb 13 23:41:00 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 13 Feb 2007 18:41:00 -0500 Subject: [et-mgmt-tools] Cobbler and kickstart templating... Message-ID: <45D24C8C.507@redhat.com> Friends, Romans, Cobbler users, Seeing I personally have a few uses for a better templating system than what I have now, I'm moving Cobbler back to using Cheetah for kickstart templating in 0.4.0. Originally the move away from Cheetah was done to accomdate RHEL3 based cobbler servers, but I can't continue to justify not having a good templating engine, so I'm cutting those lose. Kickstart files you have now will be unaffected (and will still work as is), though now "TEMPLATE::foo" just becomes an alias for the much simpler $foo. Variables that aren't found in --ksmeta arguments (say $thisisnotavariable) just don't get expanded. I'll probably bundle Cheetah in the cobbler RPM (as I did before) to make RHEL4 happy (good for Centos for having python-cheetah in the repository). This will be in the 0.4.0 release and is already pushed out to the public hg repo if you want to get a jump start on testing it out. You can learn to use the full power of Cheetah for your kickstart templates here: http://www.cheetahtemplate.org/learn.html --Michael From michael.gorlier at univ-lille3.fr Wed Feb 14 12:50:41 2007 From: michael.gorlier at univ-lille3.fr (GORLIER Michael) Date: Wed, 14 Feb 2007 13:50:41 +0100 Subject: [et-mgmt-tools] virt-install debian etch Message-ID: <45D305A1.6000903@univ-lille3.fr> Bonjour, le script virt-install sous debian etch me retourne : Traceback (most recent call last): File "./virt-install", line 25, in ? import libvirt File "/usr/lib/python2.4/site-packages/libvirt.py", line 7, in ? import libvirtmod ImportError: libxenstore.so.3.0: cannot open shared object file: No such file or directory Merci d'avance From satyaakam at gmail.com Wed Feb 14 15:44:01 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 21:14:01 +0530 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest Message-ID: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> hi, i am trying to create a paravirtualized guest with virt-manager, i am trying to use http method to install, i am using 0.3.0 version of virt-manager , the errors in the terminal window are as follows ./virt-manager Traceback (most recent call last): File "/usr/local/share/virt-manager/virtManager/create.py", line 415, in finish d = virtinst.VirtualDisk(self.get_config_disk_image(), filesize, sparse = self.is_sparse_file()) AttributeError: 'module' object has no attribute 'VirtualDisk' satya From berrange at redhat.com Wed Feb 14 15:58:35 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 14 Feb 2007 15:58:35 +0000 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> Message-ID: <20070214155835.GC29628@redhat.com> On Wed, Feb 14, 2007 at 09:14:01PM +0530, satyaakam goswami wrote: > hi, > i am trying to create a paravirtualized guest with virt-manager, > i am trying to use http method to install, i am using 0.3.0 version > of virt-manager , the errors in the terminal window are as follows > > ./virt-manager > Traceback (most recent call last): > File "/usr/local/share/virt-manager/virtManager/create.py", line > 415, in finish > d = virtinst.VirtualDisk(self.get_config_disk_image(), filesize, > sparse = self.is_sparse_file()) > AttributeError: 'module' object has no attribute 'VirtualDisk' What version of python-virtinst do you have ? It sounds like you need to have 0.100.0 IIRC Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From james at cloud9.co.uk Wed Feb 14 16:52:44 2007 From: james at cloud9.co.uk (James Fidell) Date: Wed, 14 Feb 2007 16:52:44 +0000 Subject: [et-mgmt-tools] cobbler import --mirror error Message-ID: <45D33E5C.3050401@cloud9.co.uk> I'm new to cobbler, so it's possible I've missed something obvious. Please point me in the right direction if that's the case. I've rebuild the source RPM for 0.3.9-1 on a Centos 4.4 server and I'm trying to import Fedora Core 6 from a mirror using rsync: cobbler import --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386/os/ --name=fedoracore6 Once the rsync is complete (fairly quick, as I populated most of it from RPMs I'd already downloaded), I get the following messages: /var/www/cobbler/ks_mirror/fedoracore6 /var/www/cobbler/ks_mirror/fedoracore6/images /var/www/cobbler/ks_mirror/fedoracore6/images/xen (distro added) (profile added) - looking for comps in /var/www/cobbler/ks_mirror/fedoracore6 Error: groupfile /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml cannot be found. Followed by a usage message from createrepo. /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml is present however, and readable. It looks possible to me that cobbler is running createrepo --groupfile /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml /var/www/cobbler/ks_mirror/fedoracore6 but then createrepo is prepending the path to fedoracore6 and trying to open /var/www/cobbler/ks_mirror/fedoracore6//var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml Any ideas on how to get this working? I'm using createrepo-0.4.3-1 if that makes a difference. James From satyaakam at gmail.com Wed Feb 14 17:15:55 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 22:45:55 +0530 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <20070214155835.GC29628@redhat.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> Message-ID: <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> > What version of python-virtinst do you have ? It sounds like you need to > have 0.100.0 IIRC i copied virtinst version 0.100.0 to /usr/sbin/virt-install, i still get the same error . Satya From hbrock at redhat.com Wed Feb 14 17:17:51 2007 From: hbrock at redhat.com (Hugh Brock) Date: Wed, 14 Feb 2007 12:17:51 -0500 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> Message-ID: <45D3443F.1040003@redhat.com> satyaakam goswami wrote: >> What version of python-virtinst do you have ? It sounds like you need to >> have 0.100.0 IIRC > i copied virtinst version 0.100.0 to /usr/sbin/virt-install, i still > get the same error . > > Satya You'll need to install the whole python-virtinst package. Virt-manager calls modules in the underlying python-virtinst package that have had significant changes in the last month. Take care, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From satyaakam at gmail.com Wed Feb 14 17:22:23 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 22:52:23 +0530 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <45D3443F.1040003@redhat.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> <45D3443F.1040003@redhat.com> Message-ID: <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> > You'll need to install the whole python-virtinst package. Virt-manager > calls modules in the underlying python-virtinst package that have had > significant changes in the last month. i picked up virtinst from there there are no install instructions in README http://virt-manager.et.redhat.com/download/sources/virtinst/virtinst-0.100.0.tar.gz how to i install the whole package. Satya ps: do you guys hang around on freenode From hbrock at redhat.com Wed Feb 14 17:29:53 2007 From: hbrock at redhat.com (Hugh Brock) Date: Wed, 14 Feb 2007 12:29:53 -0500 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> <45D3443F.1040003@redhat.com> <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> Message-ID: <45D34711.3010006@redhat.com> satyaakam goswami wrote: >> You'll need to install the whole python-virtinst package. Virt-manager >> calls modules in the underlying python-virtinst package that have had >> significant changes in the last month. > i picked up virtinst from there there are no install instructions in README > http://virt-manager.et.redhat.com/download/sources/virtinst/virtinst-0.100.0.tar.gz > > how to i install the whole package. > > Satya > ps: do you guys hang around on freenode Hi. Re freenode, normally not although I probably should drop in every so often. If you can't install the RPM, you'll need to put the virt-install script in /usr/sbin as you did, and the rest of the python (everything under the virtinst/ subdirectory) in your python library directory. For Fedora that's generally /usr/lib/python2.4/site-packages/... your distro may vary of course. Most important thing is to replace the existing virtinst code on your machine with the new stuff. --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mark.cave-ayland at ilande.co.uk Wed Feb 14 17:32:15 2007 From: mark.cave-ayland at ilande.co.uk (Mark Cave-Ayland) Date: Wed, 14 Feb 2007 17:32:15 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? Message-ID: <1171474335.15655.17.camel@mca-desktop> Hi everyone, Slightly off-topic since I am a Ubuntu user, but I was really keen to test the new virt-manager support for activating inactive virtual machines but unfortunately it doesn't seem to work for me in that inactive virtual machines are not listed in the virt-manager window. So far I've performed the following steps: i) Upgrade to Xen 3.0.4 from an earlier 3.0 Hg version ii) Install libvirt 0.1.11 iii) Compile & install virt-inst 0.100.0 iv) Compile & install virt-manager 0.3.0 Using virt-manager I can now view statistics from Dom0, but I don't get anything listed under inactive virtual machines. If it helps, my existing VM configurations are stored under /etc/xen (with the drive images themselves under /var/xen/images) and I can manually start inactive virtual machines using "xm create" from the command line, at which point they appear in virt-manager as expected. Can anyone explain what I have done wrong to prevent my inactive virtual machines from appearing in virt-manager? Many thanks, Mark. I From hbrock at redhat.com Wed Feb 14 17:33:59 2007 From: hbrock at redhat.com (Hugh Brock) Date: Wed, 14 Feb 2007 12:33:59 -0500 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <45D34711.3010006@redhat.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> <45D3443F.1040003@redhat.com> <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> <45D34711.3010006@redhat.com> Message-ID: <45D34807.7080805@redhat.com> Hugh Brock wrote: > satyaakam goswami wrote: >>> You'll need to install the whole python-virtinst package. Virt-manager >>> calls modules in the underlying python-virtinst package that have had >>> significant changes in the last month. >> i picked up virtinst from there there are no install instructions in >> README >> http://virt-manager.et.redhat.com/download/sources/virtinst/virtinst-0.100.0.tar.gz >> >> how to i install the whole package. >> >> Satya >> ps: do you guys hang around on freenode > > Hi. Re freenode, normally not although I probably should drop in every > so often. > > If you can't install the RPM, you'll need to put the virt-install script > in /usr/sbin as you did, and the rest of the python (everything under > the virtinst/ subdirectory) in your python library directory. For Fedora > that's generally /usr/lib/python2.4/site-packages/... your > distro may vary of course. Most important thing is to replace the > existing virtinst code on your machine with the new stuff. > Oh, and, you will also need to update libvirt to 0.1.11 or whatever the current release is. --H -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From David.Mackintosh at xdroop.com Wed Feb 14 17:34:58 2007 From: David.Mackintosh at xdroop.com (David Mackintosh) Date: Wed, 14 Feb 2007 12:34:58 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu Message-ID: <20070214173458.GL6585@xdroop.com> I was about to write some wrappers to do this, but it occurs to me that it might be easier to do inside cobbler itself. How about instead of just having pxelinux.cfg/default kick straight into a hard disk boot (which is a nice, safe default) it pops up a list of the profiles it knows about, and will accept those names at the prompt? This would let you do one-time installs without having to generate a cobbler section for it. The pxelinux.cfg/default file would then look like default local display pxelinux-default.msg timeout 100 prompt 1 label local localboot 0 label SLES-10-std kernel /images/SLES-10/vmlinuz append lang ksdevice=eth0 text devfs=nomount $OTHER-THINGS [...] I would suspect for most sites there are only going to be a few profiles, but it probably wouldn't be too hard to write the piece which generates pxelinux-default.msg to put the profiles into columns for easier display. It would probably be simplest to do this inside cobbler rather than writing a bolt-on which figures all this out. -- /\oo/\ / /()\ \ David Mackintosh | dave at xdroop.com | http://www.xdroop.com -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From mdehaan at redhat.com Wed Feb 14 17:46:03 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 14 Feb 2007 12:46:03 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D33E5C.3050401@cloud9.co.uk> References: <45D33E5C.3050401@cloud9.co.uk> Message-ID: <45D34ADB.40601@redhat.com> James Fidell wrote: > I'm new to cobbler, so it's possible I've missed something obvious. > Please point me in the right direction if that's the case. > > I've rebuild the source RPM for 0.3.9-1 on a Centos 4.4 server and > I'm trying to import Fedora Core 6 from a mirror using rsync: > > cobbler import > --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386/os/ > --name=fedoracore6 > > Once the rsync is complete (fairly quick, as I populated most of it from > RPMs I'd already downloaded), I get the following messages: > > /var/www/cobbler/ks_mirror/fedoracore6 > /var/www/cobbler/ks_mirror/fedoracore6/images > /var/www/cobbler/ks_mirror/fedoracore6/images/xen > (distro added) > (profile added) > - looking for comps in /var/www/cobbler/ks_mirror/fedoracore6 > Error: groupfile > /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml cannot be found. > > Followed by a usage message from createrepo. > > /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml is present > however, and readable. > > It looks possible to me that cobbler is running > > createrepo --groupfile > /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml > /var/www/cobbler/ks_mirror/fedoracore6 > > but then createrepo is prepending the path to fedoracore6 and > trying to open > > > /var/www/cobbler/ks_mirror/fedoracore6//var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml > > Any ideas on how to get this working? I'm using createrepo-0.4.3-1 > if that makes a difference. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > I've noticed this recently as well... I'm wondering if createrepo might have changed somewhat. I'll look into this and get back to you. --MPD From berrange at redhat.com Wed Feb 14 17:48:28 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 14 Feb 2007 17:48:28 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <1171474335.15655.17.camel@mca-desktop> References: <1171474335.15655.17.camel@mca-desktop> Message-ID: <20070214174827.GD29628@redhat.com> On Wed, Feb 14, 2007 at 05:32:15PM +0000, Mark Cave-Ayland wrote: > Hi everyone, > > Slightly off-topic since I am a Ubuntu user, but I was really keen to > test the new virt-manager support for activating inactive virtual > machines but unfortunately it doesn't seem to work for me in that > inactive virtual machines are not listed in the virt-manager window. > > So far I've performed the following steps: > > i) Upgrade to Xen 3.0.4 from an earlier 3.0 Hg version > ii) Install libvirt 0.1.11 > iii) Compile & install virt-inst 0.100.0 > iv) Compile & install virt-manager 0.3.0 > > Using virt-manager I can now view statistics from Dom0, but I don't get > anything listed under inactive virtual machines. If it helps, my > existing VM configurations are stored under /etc/xen (with the drive > images themselves under /var/xen/images) and I can manually start > inactive virtual machines using "xm create" from the command line, at > which point they appear in virt-manager as expected. Can anyone explain > what I have done wrong to prevent my inactive virtual machines from > appearing in virt-manager? Xen 3.0.4 has built in domain management. Unfortunately in doing this upstream Xen moved the config dir from /etc/xen to /var/lib/xen. So to make your inactive domains appear to Xen / virt-manager you need to load them into Xen by using 'xm new ' for each config in the /etc/xen directory. After doing this they should appear when doing xm list virsh list --all And in virt-manager Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From mdehaan at redhat.com Wed Feb 14 17:54:55 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 14 Feb 2007 12:54:55 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <20070214173458.GL6585@xdroop.com> References: <20070214173458.GL6585@xdroop.com> Message-ID: <45D34CEF.9020705@redhat.com> David Mackintosh wrote: > I was about to write some wrappers to do this, but it occurs to me that > it might be easier to do inside cobbler itself. > > How about instead of just having pxelinux.cfg/default kick straight into > a hard disk boot (which is a nice, safe default) it pops up a list of the > profiles it knows about, and will accept those names at the prompt? > > This would let you do one-time installs without having to generate > a cobbler section for it. > > The pxelinux.cfg/default file would then look like > > default local > display pxelinux-default.msg > timeout 100 > prompt 1 > > label local > localboot 0 > > label SLES-10-std > kernel /images/SLES-10/vmlinuz > append lang ksdevice=eth0 text devfs=nomount $OTHER-THINGS > > [...] > > I would suspect for most sites there are only going to be a few profiles, > but it probably wouldn't be too hard to write the piece which generates > pxelinux-default.msg to put the profiles into columns for easier display. > > It would probably be simplest to do this inside cobbler rather than > writing a bolt-on which figures all this out. > > > ------------------------------------------------------------------------ > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools This has come up before, and it's a good idea. One of the key issues here is that the pxelinux menus don't seem like they would scale very well with a very large number of profiles, so I've been hesitant to make that default behavior. Honestly I'm not entirely aware of what happens when there are 100 profiles -- if you could figure this out, that would be great, and I'll implement it :) Patches also accepted, of course. If it's going to be done outside of Cobbler, it would basically require understanding /var/lib/cobbler/profiles and /var/lib/cobbler/distros, merging the options as appropriate, and then regenerating the file when sync is run. Not too bad, really... (note that the formats of those files are subject to change though, so yes, you're right in that it would be better if part of the cobbler tree). --MPD From james at cloud9.co.uk Wed Feb 14 17:59:55 2007 From: james at cloud9.co.uk (James Fidell) Date: Wed, 14 Feb 2007 17:59:55 +0000 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <20070214173458.GL6585@xdroop.com> References: <20070214173458.GL6585@xdroop.com> Message-ID: <45D34E1B.60504@cloud9.co.uk> David Mackintosh wrote: > I was about to write some wrappers to do this, but it occurs to me that > it might be easier to do inside cobbler itself. > > How about instead of just having pxelinux.cfg/default kick straight into > a hard disk boot (which is a nice, safe default) it pops up a list of the > profiles it knows about, and will accept those names at the prompt? FWIW, I've used a system like this with one hosting company. Their PXE boot takes you into a menu that allows you to do a network-based installation of various Linux or *BSD distributions. It's quite cool. James From satyaakam at gmail.com Wed Feb 14 17:57:33 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 23:27:33 +0530 Subject: [et-mgmt-tools] http install gives 404 Message-ID: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> Hi, can i install rhel3 as paravirtualized machine using virt-manager. Satya From berrange at redhat.com Wed Feb 14 18:01:14 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 14 Feb 2007 18:01:14 +0000 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> <45D3443F.1040003@redhat.com> <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> Message-ID: <20070214180114.GE29628@redhat.com> On Wed, Feb 14, 2007 at 10:52:23PM +0530, satyaakam goswami wrote: > >You'll need to install the whole python-virtinst package. Virt-manager > >calls modules in the underlying python-virtinst package that have had > >significant changes in the last month. > i picked up virtinst from there there are no install instructions in README > http://virt-manager.et.redhat.com/download/sources/virtinst/virtinst-0.100.0.tar.gz > how to i install the whole package. Its the standard python install process, which is: python setup.py install Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Wed Feb 14 18:02:39 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 14 Feb 2007 18:02:39 +0000 Subject: [et-mgmt-tools] http install gives 404 In-Reply-To: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> References: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> Message-ID: <20070214180239.GF29628@redhat.com> On Wed, Feb 14, 2007 at 11:27:33PM +0530, satyaakam goswami wrote: > Hi, > can i install rhel3 as paravirtualized machine using virt-manager. No. Only RHEL-5 or Fedora Core 5 & later are currently supported. In the next release we'll also support Suse 10 based distros. And when RHEL-4.5 comes out, we'll support that too. There isn't any supported RHEL-3 paravirt kernel, so you'll have to do a full virt install for that. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From satyaakam at gmail.com Wed Feb 14 18:06:51 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 23:36:51 +0530 Subject: [et-mgmt-tools] virt-manger unable to create paravirtualized guest In-Reply-To: <20070214180114.GE29628@redhat.com> References: <6491e1350702140744g59ed1d39sbb8e4a16d14e7997@mail.gmail.com> <20070214155835.GC29628@redhat.com> <6491e1350702140915q5a96ca72y46d20bd562c7eb3b@mail.gmail.com> <45D3443F.1040003@redhat.com> <6491e1350702140922g6e5d594cnc7cd9bb4ac677b2c@mail.gmail.com> <20070214180114.GE29628@redhat.com> Message-ID: <6491e1350702141006j9bfb5c8q88f61fc1d32033c5@mail.gmail.com> > Its the standard python install process, which is: > > python setup.py install thanks , did a copy as suggested by Hugh. Now comign to the original issue of http install giving me 404 , i am trying to install RHEL3 is it supported. Satya From satyaakam at gmail.com Wed Feb 14 17:57:33 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 23:27:33 +0530 Subject: [et-mgmt-tools] http install gives 404 Message-ID: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> Hi, can i install rhel3 as paravirtualized machine using virt-manager. Satya From matts at hostingsupport.com Wed Feb 14 18:11:47 2007 From: matts at hostingsupport.com (Matt S Unix Administrator) Date: Wed, 14 Feb 2007 12:11:47 -0600 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D34CEF.9020705@redhat.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> Message-ID: <45D350E3.5080908@hostingsupport.com> Michael DeHaan wrote: > David Mackintosh wrote: >> I was about to write some wrappers to do this, but it occurs to me that >> it might be easier to do inside cobbler itself. >> >> How about instead of just having pxelinux.cfg/default kick straight into >> a hard disk boot (which is a nice, safe default) it pops up a list of >> the >> profiles it knows about, and will accept those names at the prompt? >> >> This would let you do one-time installs without having to generate >> a cobbler section for it. >> >> The pxelinux.cfg/default file would then look like >> >> default local >> display pxelinux-default.msg >> timeout 100 >> prompt 1 >> >> label local >> localboot 0 >> >> label SLES-10-std >> kernel /images/SLES-10/vmlinuz >> append lang ksdevice=eth0 text devfs=nomount $OTHER-THINGS >> >> [...] >> >> I would suspect for most sites there are only going to be a few >> profiles, >> but it probably wouldn't be too hard to write the piece which generates >> pxelinux-default.msg to put the profiles into columns for easier >> display. >> >> It would probably be simplest to do this inside cobbler rather than >> writing a bolt-on which figures all this out. >> >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools > > This has come up before, and it's a good idea. > > One of the key issues here is that the pxelinux menus don't seem like > they would scale very well with a very large number of profiles, so > I've been hesitant to make that default behavior. Honestly I'm not > entirely aware of what happens when there are 100 profiles -- if you > could figure this out, that would be great, and I'll implement it :) > > Patches also accepted, of course. > > If it's going to be done outside of Cobbler, it would basically > require understanding /var/lib/cobbler/profiles and > /var/lib/cobbler/distros, merging the options as appropriate, and then > regenerating the file when sync is run. Not too bad, really... > (note that the formats of those files are subject to change though, so > yes, you're right in that it would be better if part of the cobbler > tree). > > --MPD > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools You can create submenus which contain whatever options perhaps distros or some delimiter could define what should constitute a menu item. The default could contain: menu label Windows kernel menu.c32 append windows.conf menu label Linux kernel menu.c32 append linux.conf From satyaakam at gmail.com Wed Feb 14 18:22:13 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Wed, 14 Feb 2007 23:52:13 +0530 Subject: [et-mgmt-tools] http install gives 404 In-Reply-To: <20070214180239.GF29628@redhat.com> References: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> <20070214180239.GF29628@redhat.com> Message-ID: <6491e1350702141022p67e579e9n5350e88e56f12d1f@mail.gmail.com> > No. Only RHEL-5 or Fedora Core 5 & later are currently supported. In the > next release we'll also support Suse 10 based distros. And when RHEL-4.5 > comes out, we'll support that too. RHEL4.5 ?? Satya From matts at hostingsupport.com Wed Feb 14 18:13:42 2007 From: matts at hostingsupport.com (Matt S Unix Administrator) Date: Wed, 14 Feb 2007 12:13:42 -0600 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D34E1B.60504@cloud9.co.uk> References: <20070214173458.GL6585@xdroop.com> <45D34E1B.60504@cloud9.co.uk> Message-ID: <45D35156.5020804@hostingsupport.com> James Fidell wrote: > David Mackintosh wrote: > >> I was about to write some wrappers to do this, but it occurs to me that >> it might be easier to do inside cobbler itself. >> >> How about instead of just having pxelinux.cfg/default kick straight into >> a hard disk boot (which is a nice, safe default) it pops up a list of the >> profiles it knows about, and will accept those names at the prompt? >> > > FWIW, I've used a system like this with one hosting company. Their PXE > boot takes you into a menu that allows you to do a network-based > installation of various Linux or *BSD distributions. It's quite cool. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > I work for a hosting company and we employ pxe/cobbler in that way. Most of our installs are one time and sparse in frequency. From mdehaan at redhat.com Wed Feb 14 18:23:28 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 14 Feb 2007 13:23:28 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D350E3.5080908@hostingsupport.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> Message-ID: <45D353A0.7080004@redhat.com> Matt S Unix Administrator wrote: > Michael DeHaan wrote: > >> David Mackintosh wrote: >> >>> I was about to write some wrappers to do this, but it occurs to me that >>> it might be easier to do inside cobbler itself. >>> >>> How about instead of just having pxelinux.cfg/default kick straight into >>> a hard disk boot (which is a nice, safe default) it pops up a list of >>> the >>> profiles it knows about, and will accept those names at the prompt? >>> >>> This would let you do one-time installs without having to generate >>> a cobbler section for it. >>> >>> The pxelinux.cfg/default file would then look like >>> >>> default local >>> display pxelinux-default.msg >>> timeout 100 >>> prompt 1 >>> >>> label local >>> localboot 0 >>> >>> label SLES-10-std >>> kernel /images/SLES-10/vmlinuz >>> append lang ksdevice=eth0 text devfs=nomount $OTHER-THINGS >>> >>> [...] >>> >>> I would suspect for most sites there are only going to be a few >>> profiles, >>> but it probably wouldn't be too hard to write the piece which generates >>> pxelinux-default.msg to put the profiles into columns for easier >>> display. >>> >>> It would probably be simplest to do this inside cobbler rather than >>> writing a bolt-on which figures all this out. >>> >>> >>> ------------------------------------------------------------------------ >>> >>> _______________________________________________ >>> et-mgmt-tools mailing list >>> et-mgmt-tools at redhat.com >>> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >>> >> This has come up before, and it's a good idea. >> >> One of the key issues here is that the pxelinux menus don't seem like >> they would scale very well with a very large number of profiles, so >> I've been hesitant to make that default behavior. Honestly I'm not >> entirely aware of what happens when there are 100 profiles -- if you >> could figure this out, that would be great, and I'll implement it :) >> >> Patches also accepted, of course. >> >> If it's going to be done outside of Cobbler, it would basically >> require understanding /var/lib/cobbler/profiles and >> /var/lib/cobbler/distros, merging the options as appropriate, and then >> regenerating the file when sync is run. Not too bad, really... >> (note that the formats of those files are subject to change though, so >> yes, you're right in that it would be better if part of the cobbler >> tree). >> >> --MPD >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > You can create submenus which contain whatever options perhaps distros > or some delimiter could define what should constitute a menu item. > > The default could contain: > > menu label Windows > kernel menu.c32 > append windows.conf > > menu label Linux > kernel menu.c32 > append linux.conf > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Excellent! I'll check this out... From hbrock at redhat.com Wed Feb 14 18:24:12 2007 From: hbrock at redhat.com (Hugh Brock) Date: Wed, 14 Feb 2007 13:24:12 -0500 Subject: [et-mgmt-tools] http install gives 404 In-Reply-To: <6491e1350702141022p67e579e9n5350e88e56f12d1f@mail.gmail.com> References: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> <20070214180239.GF29628@redhat.com> <6491e1350702141022p67e579e9n5350e88e56f12d1f@mail.gmail.com> Message-ID: <45D353CC.2000408@redhat.com> satyaakam goswami wrote: >> No. Only RHEL-5 or Fedora Core 5 & later are currently supported. In the >> next release we'll also support Suse 10 based distros. And when RHEL-4.5 >> comes out, we'll support that too. > RHEL4.5 ?? > That's the next update of RHEL 4, coming soon. It will have a paravirt kernel in addition to the regular kernel. --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From satyaakam at gmail.com Wed Feb 14 19:00:42 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Thu, 15 Feb 2007 00:30:42 +0530 Subject: [et-mgmt-tools] http install gives 404 In-Reply-To: <45D353CC.2000408@redhat.com> References: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> <20070214180239.GF29628@redhat.com> <6491e1350702141022p67e579e9n5350e88e56f12d1f@mail.gmail.com> <45D353CC.2000408@redhat.com> Message-ID: <6491e1350702141100h630dfb39s7d2c6362cd3be738@mail.gmail.com> > That's the next update of RHEL 4, coming soon. It will have a paravirt > kernel in addition to the regular kernel. well what i am trying to do is install RHEL 3.0 as guest not as Host .Can i do that using virt-manager Satya From hbrock at redhat.com Wed Feb 14 19:03:37 2007 From: hbrock at redhat.com (Hugh Brock) Date: Wed, 14 Feb 2007 14:03:37 -0500 Subject: [et-mgmt-tools] http install gives 404 In-Reply-To: <6491e1350702141100h630dfb39s7d2c6362cd3be738@mail.gmail.com> References: <6491e1350702140957q7e376ea0m3ded2c418bdcc2ab@mail.gmail.com> <20070214180239.GF29628@redhat.com> <6491e1350702141022p67e579e9n5350e88e56f12d1f@mail.gmail.com> <45D353CC.2000408@redhat.com> <6491e1350702141100h630dfb39s7d2c6362cd3be738@mail.gmail.com> Message-ID: <45D35D09.2040006@redhat.com> satyaakam goswami wrote: >> That's the next update of RHEL 4, coming soon. It will have a paravirt >> kernel in addition to the regular kernel. > well what i am trying to do is install RHEL 3.0 as guest not as Host > .Can i do that using virt-manager > Sure, but you will have to install it fully virtualized. This means you'll need install media or a rhel3 boot.iso. It also means rhel3 performance will suffer somewhat since all its I/O will be fully virtualized. However it should work reasonably well. Oh, you may also need to turn off acpi and apic in the rhel3 guest's config file... but try it and see. --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From dhawal at netmagicsolutions.com Wed Feb 14 19:07:39 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Thu, 15 Feb 2007 00:37:39 +0530 Subject: [et-mgmt-tools] multi-os howto? Message-ID: <45D35DFB.8040105@netmagicsolutions.com> Hello Michael and List, i recently started testing cobbler for my setup and quite like the ease of provisioning. Here is what i have done so far.. cobbler distro add --name=rhel4 --kernel=/path/to/vmlinuz --initrd=/path/to/initrd.img cobbler profile add --name=rhel4_ni --distro=rhel4 --kickstart=http://192.168.1.5/ks.cfg cobbler profile add --name=rhel4_interactive --distro=rhel4 --kickstart=http://192.168.1.5/ks_custom_part.cfg .... .... same for centos However each of these creates a /tftpboot/pxelinux.cfg/distro and not /tftpboot/pxelinux.cfg/default. In the production setup i have the 'default' file being updated manually with different labels to support multiple OSes.. and a Display file to present the options at the boot: prompt. My question is how do it do this using cobbler? - dhawal From dhawal at netmagicsolutions.com Wed Feb 14 19:10:56 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Thu, 15 Feb 2007 00:40:56 +0530 Subject: [et-mgmt-tools] multi-os howto? In-Reply-To: <45D35DFB.8040105@netmagicsolutions.com> References: <45D35DFB.8040105@netmagicsolutions.com> Message-ID: <45D35EC0.8020900@netmagicsolutions.com> Top posting, oops, and i just read the other thread.. "Cobbler idea: PXE boot profile menu". What we do is more/less detailed here if you want a link: http://www.howtoforge.com/ubuntu_pxe_install_server - dhawal Dhawal Doshy wrote: > Hello Michael and List, > > i recently started testing cobbler for my setup and quite like the ease > of provisioning. > > Here is what i have done so far.. > cobbler distro add --name=rhel4 --kernel=/path/to/vmlinuz > --initrd=/path/to/initrd.img > > cobbler profile add --name=rhel4_ni --distro=rhel4 > --kickstart=http://192.168.1.5/ks.cfg > > cobbler profile add --name=rhel4_interactive --distro=rhel4 > --kickstart=http://192.168.1.5/ks_custom_part.cfg > .... > .... > same for centos > > However each of these creates a /tftpboot/pxelinux.cfg/distro and not > /tftpboot/pxelinux.cfg/default. In the production setup i have the > 'default' file being updated manually with different labels to support > multiple OSes.. and a Display file to present the options at the boot: > prompt. > > My question is how do it do this using cobbler? > > - dhawal From mdehaan at redhat.com Wed Feb 14 19:30:27 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 14 Feb 2007 14:30:27 -0500 Subject: [et-mgmt-tools] multi-os howto? In-Reply-To: <45D35EC0.8020900@netmagicsolutions.com> References: <45D35DFB.8040105@netmagicsolutions.com> <45D35EC0.8020900@netmagicsolutions.com> Message-ID: <45D36353.3050903@redhat.com> Dhawal Doshy wrote: > Top posting, oops, and i just read the other thread.. "Cobbler idea: > PXE boot profile menu". > > What we do is more/less detailed here if you want a link: > http://www.howtoforge.com/ubuntu_pxe_install_server > > - dhawal > > Dhawal Doshy wrote: >> Hello Michael and List, >> >> i recently started testing cobbler for my setup and quite like the ease >> of provisioning. >> >> Here is what i have done so far.. >> cobbler distro add --name=rhel4 --kernel=/path/to/vmlinuz >> --initrd=/path/to/initrd.img >> >> cobbler profile add --name=rhel4_ni --distro=rhel4 >> --kickstart=http://192.168.1.5/ks.cfg >> >> cobbler profile add --name=rhel4_interactive --distro=rhel4 >> --kickstart=http://192.168.1.5/ks_custom_part.cfg >> .... >> .... >> same for centos >> >> However each of these creates a /tftpboot/pxelinux.cfg/distro and not >> /tftpboot/pxelinux.cfg/default. In the production setup i have the >> 'default' file being updated manually with different labels to support >> multiple OSes.. and a Display file to present the options at the boot: >> prompt. >> >> My question is how do it do this using cobbler? >> >> - dhawal > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools If you are interested, cobbler can keep a MAC address database in cobbler so the systems will PXE boot to what they are supposed to be when they do not have an OS on them, based on the MAC address. This makes the defaults file unneccessary. cobbler system add --name=AA:BB:CC:DD:EE:EE --profile=fc5i386dbserver cobbler system add --name=AA:BB:CC:DD:EE:FF --profile=fc5i386webserver I will definitely add the menus though as that appears to be a very common hosting use case. OT, but I would also suggest starting to reference your kickstart files on the local system (i.e. --kickstart=/path/to/file) as that way you can start taking advantage of cobbler's kickstart templating features should you want to use them later. However what you are doing is fine if you don't need that just yet. As for Ubuntu, it's pretty cool to see this: https://help.ubuntu.com/community/KickstartCompatibility, which seems to imply you can do fully automatic installs from Cobbler without changing anything. Awesome. --MPD From dhawal at netmagicsolutions.com Wed Feb 14 19:54:29 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Thu, 15 Feb 2007 01:24:29 +0530 Subject: [et-mgmt-tools] multi-os howto? In-Reply-To: <45D36353.3050903@redhat.com> References: <45D35DFB.8040105@netmagicsolutions.com> <45D35EC0.8020900@netmagicsolutions.com> <45D36353.3050903@redhat.com> Message-ID: <45D368F5.1070201@netmagicsolutions.com> Michael DeHaan wrote: >> Dhawal Doshy wrote: >>> Hello Michael and List, >>> >>> i recently started testing cobbler for my setup and quite like the ease >>> of provisioning. [SNIP] >>> However each of these creates a /tftpboot/pxelinux.cfg/distro and not >>> /tftpboot/pxelinux.cfg/default. In the production setup i have the >>> 'default' file being updated manually with different labels to support >>> multiple OSes.. and a Display file to present the options at the boot: >>> prompt. >>> >>> My question is how do it do this using cobbler? > > If you are interested, cobbler can keep a MAC address database in > cobbler so the systems will PXE boot to what they are supposed to be > when they do not have an OS on them, based on the MAC address. This > makes the defaults file unneccessary. > > cobbler system add --name=AA:BB:CC:DD:EE:EE --profile=fc5i386dbserver > cobbler system add --name=AA:BB:CC:DD:EE:FF --profile=fc5i386webserver I tried that for a test case.. but in most cases we are not aware of the MAC address till the system boots up (when its already too late and time consuming to note it down and create a system tailored for it).. hence our need for the 'default' thingy in pxelinux.cfg > I will definitely add the menus though as that appears to be a very > common hosting use case. That'll be nice.. > OT, but I would also suggest starting to reference your kickstart files > on the local system (i.e. --kickstart=/path/to/file) as that way you can > start taking advantage of cobbler's kickstart templating features should > you want to use them later. However what you are doing is fine if you > don't need that just yet. We might need templating of the kickstart file (if i understand the concept correctly) since we offer a host of pre-configured servers (mx, smtp, delivery, web, database etc..) with standard and custom partitioning. > As for Ubuntu, it's pretty cool to see this: > https://help.ubuntu.com/community/KickstartCompatibility, which seems to > imply you can do fully automatic installs from Cobbler without changing > anything. Awesome. cool.. once we are done with a functional setup of cobbler, we intend on taking up koan as well (just for kicks).. though we rarely re-provision a server and virtualization is still not explored completely. - dhawal From mdehaan at redhat.com Wed Feb 14 21:59:31 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 14 Feb 2007 16:59:31 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D34ADB.40601@redhat.com> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> Message-ID: <45D38643.10309@redhat.com> Michael DeHaan wrote: > James Fidell wrote: >> I'm new to cobbler, so it's possible I've missed something obvious. >> Please point me in the right direction if that's the case. >> >> I've rebuild the source RPM for 0.3.9-1 on a Centos 4.4 server and >> I'm trying to import Fedora Core 6 from a mirror using rsync: >> >> cobbler import >> --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386/os/ >> >> --name=fedoracore6 >> >> Once the rsync is complete (fairly quick, as I populated most of it from >> RPMs I'd already downloaded), I get the following messages: >> >> /var/www/cobbler/ks_mirror/fedoracore6 >> /var/www/cobbler/ks_mirror/fedoracore6/images >> /var/www/cobbler/ks_mirror/fedoracore6/images/xen >> (distro added) >> (profile added) >> - looking for comps in /var/www/cobbler/ks_mirror/fedoracore6 >> Error: groupfile >> /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml cannot be >> found. >> >> Followed by a usage message from createrepo. >> >> /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml is present >> however, and readable. >> >> It looks possible to me that cobbler is running >> >> createrepo --groupfile >> /var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml >> /var/www/cobbler/ks_mirror/fedoracore6 >> >> but then createrepo is prepending the path to fedoracore6 and >> trying to open >> >> >> /var/www/cobbler/ks_mirror/fedoracore6//var/www/cobbler/ks_mirror/fedoracore6/repodata/comps.xml >> >> >> Any ideas on how to get this working? I'm using createrepo-0.4.3-1 >> if that makes a difference. >> >> James >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > I've noticed this recently as well... I'm wondering if createrepo > might have changed somewhat. > > I'll look into this and get back to you. > > --MPD > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools Apologies on being lazy here, though per the manpage... mirror The addresss of the mirror. This needs to be either an rsync:// url or an ssh location usable with rsync. The mirror address should specify an exact repository to mirror -- just one architecture and just one distribution. If you have a seperate repo to mirror for a different arch, add that repo seperately. Here's an example of what looks like a good URL: rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync protocol) user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) What I thought was a related problem is just something I'm seeing during the unit tests, so that was a bit of wild goose chase. Your problem is that you didn't rsync enough data, and don't have the comps file. Do an import using the URL format specified in the manpage and you should be good to go. I would suggest cleaning up /var/www/cobbler/ks_mirror manually before doing that, just to free up the extra space on your drives. Hope this helps, if not, let me know... --Michael From mark.cave-ayland at ilande.co.uk Thu Feb 15 09:06:02 2007 From: mark.cave-ayland at ilande.co.uk (Mark Cave-Ayland) Date: Thu, 15 Feb 2007 09:06:02 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <20070214174827.GD29628@redhat.com> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> Message-ID: <1171530362.7896.8.camel@mca-desktop> On Wed, 2007-02-14 at 17:48 +0000, Daniel P. Berrange wrote: > Xen 3.0.4 has built in domain management. Unfortunately in doing this > upstream Xen moved the config dir from /etc/xen to /var/lib/xen. So to > make your inactive domains appear to Xen / virt-manager you need to > load them into Xen by using 'xm new ' for each config in the > /etc/xen directory. > > After doing this they should appear when doing > > xm list > virsh list --all > > And in virt-manager > > Regards, > Dan. Hi Dan, Yes that was it - thanks very much for your help! I searched the FAQ and man pages and couldn't find any reference to this, although a google on "xm new" did show a similar thread on xen-users. Another quick couple of questions: firstly is there a way to bypass "Open Connection" dialog on startup so that it assumes that the default "Local Xen Host" is selected? I can see the -C option for a URI but can't see a way to specify the equivalent of "Local Xen Host" on the command line? Secondly, I'm using a HVM WinXP installation and I appear to be suffering badly from the misaligned mouse pointer problem as mentioned in the virt-manager FAQ. Are there any updates from the Xen team as to when this is likely to be fixed? Many thanks, Mark. From berrange at redhat.com Thu Feb 15 11:58:06 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Thu, 15 Feb 2007 11:58:06 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <1171530362.7896.8.camel@mca-desktop> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> <1171530362.7896.8.camel@mca-desktop> Message-ID: <20070215115806.GA16485@redhat.com> On Thu, Feb 15, 2007 at 09:06:02AM +0000, Mark Cave-Ayland wrote: > On Wed, 2007-02-14 at 17:48 +0000, Daniel P. Berrange wrote: > > > Xen 3.0.4 has built in domain management. Unfortunately in doing this > > upstream Xen moved the config dir from /etc/xen to /var/lib/xen. So to > > make your inactive domains appear to Xen / virt-manager you need to > > load them into Xen by using 'xm new ' for each config in the > > /etc/xen directory. > > > > After doing this they should appear when doing > > > > xm list > > virsh list --all > > > > And in virt-manager > > > > Yes that was it - thanks very much for your help! I searched the FAQ and > man pages and couldn't find any reference to this, although a google on > "xm new" did show a similar thread on xen-users. Yes, the documentation leaves alot of be desired :-( > Another quick couple of questions: firstly is there a way to bypass > "Open Connection" dialog on startup so that it assumes that the default > "Local Xen Host" is selected? I can see the -C option for a URI but > can't see a way to specify the equivalent of "Local Xen Host" on the > command line? virt-manager --connect Xen Will do the trick IIRC. > Secondly, I'm using a HVM WinXP installation and I appear to be > suffering badly from the misaligned mouse pointer problem as mentioned > in the virt-manager FAQ. Are there any updates from the Xen team as to > when this is likely to be fixed? The next release of virt-manager will implement a workaround - when moving the mouse cursor over the virtual console, we will grab the cursor and lock it to be within the console window - we can then hide the local cursor so you only see one. The trouble is to release the grab, you need a magic key sequence, eg hold down 'Ctrl + Alt', but I think this will still be better than the current situation which just sucks :-) Further on, we plan to add support for turning on the USB graphics tablet emulation for Windows based OS which ought to be give saner cursor movement without need for the mouse grab. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From hbrock at redhat.com Thu Feb 15 14:37:18 2007 From: hbrock at redhat.com (Hugh Brock) Date: Thu, 15 Feb 2007 09:37:18 -0500 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <20070215115806.GA16485@redhat.com> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> <1171530362.7896.8.camel@mca-desktop> <20070215115806.GA16485@redhat.com> Message-ID: <45D4701E.60304@redhat.com> Daniel P. Berrange wrote: > On Thu, Feb 15, 2007 at 09:06:02AM +0000, Mark Cave-Ayland wrote: >> On Wed, 2007-02-14 at 17:48 +0000, Daniel P. Berrange wrote: > >> Secondly, I'm using a HVM WinXP installation and I appear to be >> suffering badly from the misaligned mouse pointer problem as mentioned >> in the virt-manager FAQ. Are there any updates from the Xen team as to >> when this is likely to be fixed? > > The next release of virt-manager will implement a workaround - when moving > the mouse cursor over the virtual console, we will grab the cursor and > lock it to be within the console window - we can then hide the local cursor > so you only see one. The trouble is to release the grab, you need a magic > key sequence, eg hold down 'Ctrl + Alt', but I think this will still be > better than the current situation which just sucks :-) > > Further on, we plan to add support for turning on the USB graphics tablet > emulation for Windows based OS which ought to be give saner cursor movement > without need for the mouse grab. > > Dan. Another way around the problem, until we get a real fix, is to run a vnc server on your guest and connect to it from dom0 in the usual way. The mouse will work normally just as in any other regular vnc session. Drawback is that this doesn't give you real "console" access... --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mdehaan at redhat.com Thu Feb 15 17:19:29 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 15 Feb 2007 12:19:29 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D353A0.7080004@redhat.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> <45D353A0.7080004@redhat.com> Message-ID: <45D49621.9090303@redhat.com> Michael DeHaan wrote: > Matt S Unix Administrator wrote: .. snip ... Allright, so I've got a pretty good prototype of pxelinux menus being generated from Cobbler now. One small problem -- timeouts don't appear to count down at all -- or else, they just don't work -- documentation is fairly lacking. The following configuration file (which has been generated from cobbler), shows that it will "Automatic boot in X seconds", though after this time expires, nothing happens.... what I really /need/ to happen is for it fall through to local boot rather than hanging around on the menu screen. Does anyone have any idea how to get this last bit working? This might be my test hardware specifically, though I kind of doubt it. Here's the file... DEFAULT menu.c32 PROMPT 0 MENU TITLE Cobbler | http://cobbler.et.redhat.com TIMEOUT 200 TOTALTIMEOUT 600 ONTIMEOUT local LABEL local MENU LABEL (local) MENU DEFAULT LOCALBOOT 0 LABEL fc62895 kernel /images/fc62895/vmlinuz MENU LABEL fc62895 append ksdevice=eth0 lang= text syslog=192.168.1.99:25150 devfs=nomount ramdisk_size=16438 append initrd=/images/fc62895/initrd.img ks=http://192.168.1.99/cobbler_track/kickstarts/fc62895/ks.cfg I've also tried "ONTIMEOUT LOCALBOOT 0" which also doesn't work, as well as leaving off the "ONTIMEOUT" bit and assuming the MENU DEFAULT would do the right thing. Provided I can't get this working, there's another decent way to make this happen, that is slightly less usable. That's to set the DEFAULT to "local" in the first few lines of the file, and we enable "PROMPT 1". Then the user has 20 or so seconds to type "menu" at the pxelinux "boot:" prompt, otherwise it would boot locally. If they type "menu", they would get the menu. It works, but it's not quite obvious. Anyone have any ideas on how to utilize timeouts and menu.c32 at the same time? Thanks, --Michael From matts at hostingsupport.com Thu Feb 15 17:23:50 2007 From: matts at hostingsupport.com (Matt S Unix Administrator) Date: Thu, 15 Feb 2007 11:23:50 -0600 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D49621.9090303@redhat.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> <45D353A0.7080004@redhat.com> <45D49621.9090303@redhat.com> Message-ID: <45D49726.2030407@hostingsupport.com> Michael DeHaan wrote: > Michael DeHaan wrote: >> Matt S Unix Administrator wrote: > .. snip ... > > Allright, so I've got a pretty good prototype of pxelinux menus being > generated from Cobbler now. One small problem -- timeouts don't > appear to count down at all -- or else, they just don't work -- > documentation is fairly lacking. The following configuration file > (which has been generated from cobbler), shows that it will "Automatic > boot in X seconds", though after this time expires, nothing > happens.... what I really /need/ to happen is for it fall through to > local boot rather than hanging around on the menu screen. Does > anyone have any idea how to get this last bit working? This might > be my test hardware specifically, though I kind of doubt it. > > Here's the file... > > DEFAULT menu.c32 > PROMPT 0 > MENU TITLE Cobbler | http://cobbler.et.redhat.com > TIMEOUT 200 > TOTALTIMEOUT 600 > ONTIMEOUT local > > LABEL local > MENU LABEL (local) > MENU DEFAULT > LOCALBOOT 0 > > LABEL fc62895 > kernel /images/fc62895/vmlinuz > MENU LABEL fc62895 > append ksdevice=eth0 lang= text syslog=192.168.1.99:25150 > devfs=nomount ramdisk_size=16438 append > initrd=/images/fc62895/initrd.img > ks=http://192.168.1.99/cobbler_track/kickstarts/fc62895/ks.cfg > > > I've also tried "ONTIMEOUT LOCALBOOT 0" which also doesn't work, as > well as leaving off the "ONTIMEOUT" bit and assuming the MENU DEFAULT > would do the right thing. > > Provided I can't get this working, there's another decent way to make > this happen, that is slightly less usable. That's to set the DEFAULT > to "local" in the first few lines of the file, and we enable "PROMPT > 1". Then the user has 20 or so seconds to type "menu" at the > pxelinux "boot:" prompt, otherwise it would boot locally. If they > type "menu", they would get the menu. It works, but it's not quite > obvious. > > Anyone have any ideas on how to utilize timeouts and menu.c32 at the > same time? > > Thanks, > > --Michael > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools I've had this same issue with the timeout, it never counts down. From mdehaan at redhat.com Thu Feb 15 17:37:15 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 15 Feb 2007 12:37:15 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D49726.2030407@hostingsupport.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> <45D353A0.7080004@redhat.com> <45D49621.9090303@redhat.com> <45D49726.2030407@hostingsupport.com> Message-ID: <45D49A4B.7000509@redhat.com> Matt S Unix Administrator wrote: > Michael DeHaan wrote: > >> Michael DeHaan wrote: >> >>> Matt S Unix Administrator wrote: >>> >> .. snip ... >> >> Allright, so I've got a pretty good prototype of pxelinux menus being >> generated from Cobbler now. One small problem -- timeouts don't >> appear to count down at all -- or else, they just don't work -- >> documentation is fairly lacking. The following configuration file >> (which has been generated from cobbler), shows that it will "Automatic >> boot in X seconds", though after this time expires, nothing >> happens.... what I really /need/ to happen is for it fall through to >> local boot rather than hanging around on the menu screen. Does >> anyone have any idea how to get this last bit working? This might >> be my test hardware specifically, though I kind of doubt it. >> >> Here's the file... >> >> DEFAULT menu.c32 >> PROMPT 0 >> MENU TITLE Cobbler | http://cobbler.et.redhat.com >> TIMEOUT 200 >> TOTALTIMEOUT 600 >> ONTIMEOUT local >> >> LABEL local >> MENU LABEL (local) >> MENU DEFAULT >> LOCALBOOT 0 >> >> LABEL fc62895 >> kernel /images/fc62895/vmlinuz >> MENU LABEL fc62895 >> append ksdevice=eth0 lang= text syslog=192.168.1.99:25150 >> devfs=nomount ramdisk_size=16438 append >> initrd=/images/fc62895/initrd.img >> ks=http://192.168.1.99/cobbler_track/kickstarts/fc62895/ks.cfg >> >> >> I've also tried "ONTIMEOUT LOCALBOOT 0" which also doesn't work, as >> well as leaving off the "ONTIMEOUT" bit and assuming the MENU DEFAULT >> would do the right thing. >> >> Provided I can't get this working, there's another decent way to make >> this happen, that is slightly less usable. That's to set the DEFAULT >> to "local" in the first few lines of the file, and we enable "PROMPT >> 1". Then the user has 20 or so seconds to type "menu" at the >> pxelinux "boot:" prompt, otherwise it would boot locally. If they >> type "menu", they would get the menu. It works, but it's not quite >> obvious. >> >> Anyone have any ideas on how to utilize timeouts and menu.c32 at the >> same time? >> >> Thanks, >> >> --Michael >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > I've had this same issue with the timeout, it never counts down. > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > How do you approach this? Having the entire data center reboot and having to visit menus to boot them would be bad :) It's not so bad if you don't have the BIOS'es set to PXE first every time though ... though it seems that some people do have (or want to have) configurations set up that way. It would be a pretty easy way (if not somewhat dangerous way) to reprovision boxes without needing to invoke koan. --MPD From mdehaan at redhat.com Thu Feb 15 19:21:59 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 15 Feb 2007 14:21:59 -0500 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D49A4B.7000509@redhat.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> <45D353A0.7080004@redhat.com> <45D49621.9090303@redhat.com> <45D49726.2030407@hostingsupport.com> <45D49A4B.7000509@redhat.com> Message-ID: <45D4B2D7.8080209@redhat.com> ... snip ... For those interested, I've pushed out the implementation that allows typing "menu" during the PXE boot process to get a menu or just pressing enter or waiting for timeout (currently 20 seconds) to get the local boot. I think it will work rather well, comments welcome, so check it out... every profile in Cobbler is shown in a sorted scrollable list... the list seems to scale fairly well -- I've tried it with 200 profiles thus far and it performed well. The way the config is set up now, it should be sane for situations where either (A) the system is set to PXE every time or (B) the system is set to PXE when their is no local OS... I wanted to ensure that we covered both as folks appear to want both. There aren't a lot of pressing features for Cobbler at this point, other than some internal stuff about the format of /var/lib/cobbler files (that shouldn't affect anyone -- it's mainly to make the API a bit more useful when feeding in huge chunks of template data), so I'll probably aim for a release of 0.4.0 sometime next week if there aren't any comments on what is already there -- I'd definitely welcome any suggestions for improvements on (or patches for) the PXE menu implementation. The PXE menu files should probably be based on templates to allow custom timeouts/captions and such, though that can be dealt with in a later release. $ yum install mercurial $ hg clone http://hg.et.redhat.com/hg/emd/applications/cobbler cobbler $ cd cobbler $ make --Michael From matts at hostingsupport.com Thu Feb 15 19:49:55 2007 From: matts at hostingsupport.com (Matt S Unix Administrator) Date: Thu, 15 Feb 2007 13:49:55 -0600 Subject: [et-mgmt-tools] Cobbler idea: PXE boot profile menu In-Reply-To: <45D49A4B.7000509@redhat.com> References: <20070214173458.GL6585@xdroop.com> <45D34CEF.9020705@redhat.com> <45D350E3.5080908@hostingsupport.com> <45D353A0.7080004@redhat.com> <45D49621.9090303@redhat.com> <45D49726.2030407@hostingsupport.com> <45D49A4B.7000509@redhat.com> Message-ID: <45D4B963.9010206@hostingsupport.com> Michael DeHaan wrote: > Matt S Unix Administrator wrote: >> Michael DeHaan wrote: >> >>> Michael DeHaan wrote: >>> >>>> Matt S Unix Administrator wrote: >>>> >>> .. snip ... >>> >>> Allright, so I've got a pretty good prototype of pxelinux menus being >>> generated from Cobbler now. One small problem -- timeouts don't >>> appear to count down at all -- or else, they just don't work -- >>> documentation is fairly lacking. The following configuration file >>> (which has been generated from cobbler), shows that it will "Automatic >>> boot in X seconds", though after this time expires, nothing >>> happens.... what I really /need/ to happen is for it fall through to >>> local boot rather than hanging around on the menu screen. Does >>> anyone have any idea how to get this last bit working? This might >>> be my test hardware specifically, though I kind of doubt it. >>> >>> Here's the file... >>> >>> DEFAULT menu.c32 >>> PROMPT 0 >>> MENU TITLE Cobbler | http://cobbler.et.redhat.com >>> TIMEOUT 200 >>> TOTALTIMEOUT 600 >>> ONTIMEOUT local >>> >>> LABEL local >>> MENU LABEL (local) >>> MENU DEFAULT >>> LOCALBOOT 0 >>> >>> LABEL fc62895 >>> kernel /images/fc62895/vmlinuz >>> MENU LABEL fc62895 >>> append ksdevice=eth0 lang= text syslog=192.168.1.99:25150 >>> devfs=nomount ramdisk_size=16438 append >>> initrd=/images/fc62895/initrd.img >>> ks=http://192.168.1.99/cobbler_track/kickstarts/fc62895/ks.cfg >>> >>> >>> I've also tried "ONTIMEOUT LOCALBOOT 0" which also doesn't work, as >>> well as leaving off the "ONTIMEOUT" bit and assuming the MENU DEFAULT >>> would do the right thing. >>> >>> Provided I can't get this working, there's another decent way to make >>> this happen, that is slightly less usable. That's to set the DEFAULT >>> to "local" in the first few lines of the file, and we enable "PROMPT >>> 1". Then the user has 20 or so seconds to type "menu" at the >>> pxelinux "boot:" prompt, otherwise it would boot locally. If they >>> type "menu", they would get the menu. It works, but it's not quite >>> obvious. >>> >>> Anyone have any ideas on how to utilize timeouts and menu.c32 at the >>> same time? >>> >>> Thanks, >>> >>> --Michael >>> >>> _______________________________________________ >>> et-mgmt-tools mailing list >>> et-mgmt-tools at redhat.com >>> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >>> >> I've had this same issue with the timeout, it never counts down. >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > > How do you approach this? Having the entire data center reboot and > having to visit menus to boot them would be bad :) > > It's not so bad if you don't have the BIOS'es set to PXE first every > time though ... though it seems that some people do have (or want to > have) configurations set up that way. It would be a pretty easy way > (if not somewhat dangerous way) to reprovision boxes without needing > to invoke koan. > > --MPD > > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools Our implementation usually involves a person selecting the option they required and then upon reboot selecting a chainload menu entry which boots to ide0. It's unfortunate that the timeout doesn't seem to work with the menu, but the solution you provided is workable for us. From mdehaan at redhat.com Thu Feb 15 23:02:17 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 15 Feb 2007 18:02:17 -0500 Subject: [et-mgmt-tools] Cobbler and kickstart templating... In-Reply-To: <45D24C8C.507@redhat.com> References: <45D24C8C.507@redhat.com> Message-ID: <45D4E679.4030600@redhat.com> Michael DeHaan wrote: > Friends, Romans, Cobbler users, > > Seeing I personally have a few uses for a better templating system > than what I have now, I'm moving Cobbler back to using Cheetah for > kickstart templating in 0.4.0. Originally the move away from Cheetah > was done to accomdate RHEL3 based cobbler servers, but I can't > continue to justify not having a good templating engine, so I'm > cutting those lose. > > Kickstart files you have now will be unaffected (and will still work > as is), though now "TEMPLATE::foo" just becomes an alias for the much > simpler $foo. Variables that aren't found in --ksmeta arguments (say > $thisisnotavariable) just don't get expanded. > > I'll probably bundle Cheetah in the cobbler RPM (as I did before) to > make RHEL4 happy (good for Centos for having python-cheetah in the > repository). > This will be in the 0.4.0 release and is already pushed out to the > public hg repo if you want to get a jump start on testing it out. > You can learn to use the full power of Cheetah for your kickstart > templates here: http://www.cheetahtemplate.org/learn.html > > --Michael > > > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools I lied somewhat :) Cobbler 0.4.0 will apparently require escaping of certain dollar signs in kickstart files. Particularly, the list hard drives macro in /etc/cobbler/kickstart-fc6.ks for instance, will be changed so it will work for new users. Those users already using those files will need to make those changes manually or to copy /etc/cobbler/kickstart-fc6.ks.rpmnew over the other file -- the RPM is set up not to overwrite files that users might have modified. Per the Cheetah manual: Placeholders and directives can be escaped by putting a backslash before them. |\$var| and |\#if| will be output as literal text. From ccoffing at novell.com Fri Feb 16 01:09:56 2007 From: ccoffing at novell.com (Charles Coffing) Date: Thu, 15 Feb 2007 18:09:56 -0700 Subject: [et-mgmt-tools] virt-manager: global name 'grabbedKeyboard' is not defined Message-ID: <45D4A15F.D169.003C.0@novell.com> There's a trivial bug in the latest virt-manager hg: --- src/vncViewer/vnc.py~ 2007-02-15 17:54:59.000000000 -0700 +++ src/vncViewer/vnc.py 2007-02-15 17:55:06.000000000 -0700 @@ -424,7 +424,7 @@ def set_autograb_keyboard(self, grab): self.autograbkey = grab - if grab == False and grabbedKeyboard: + if grab == False and self.grabbedKeyboard: self.ungrab_keyboard() From berrange at redhat.com Fri Feb 16 02:07:54 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Fri, 16 Feb 2007 02:07:54 +0000 Subject: [et-mgmt-tools] virt-manager: global name 'grabbedKeyboard' is not defined In-Reply-To: <45D4A15F.D169.003C.0@novell.com> References: <45D4A15F.D169.003C.0@novell.com> Message-ID: <20070216020753.GA13030@redhat.com> On Thu, Feb 15, 2007 at 06:09:56PM -0700, Charles Coffing wrote: > There's a trivial bug in the latest virt-manager hg: > > --- src/vncViewer/vnc.py~ 2007-02-15 17:54:59.000000000 -0700 > +++ src/vncViewer/vnc.py 2007-02-15 17:55:06.000000000 -0700 > @@ -424,7 +424,7 @@ > > def set_autograb_keyboard(self, grab): > self.autograbkey = grab > - if grab == False and grabbedKeyboard: > + if grab == False and self.grabbedKeyboard: > self.ungrab_keyboard() Thanks, have applied the patch to the mercurial repo. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From james at cloud9.co.uk Fri Feb 16 13:29:23 2007 From: james at cloud9.co.uk (James Fidell) Date: Fri, 16 Feb 2007 13:29:23 +0000 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D38643.10309@redhat.com> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> Message-ID: <45D5B1B3.5060504@cloud9.co.uk> Michael DeHaan wrote: > Apologies on being lazy here, though per the manpage... > > mirror > > The addresss of the mirror. This needs to be either an rsync:// url > or an ssh location usable with rsync. > The mirror address should specify an exact repository to mirror -- > just one architecture > and just one distribution. If you have a seperate repo to mirror for > a different arch, add that > repo seperately. > > Here's an example of what looks like a good URL: > > rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync > protocol) > user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) > > What I thought was a related problem is just something I'm seeing during > the unit tests, so that was a bit of wild goose chase. > > Your problem is that you didn't rsync enough data, and don't have the > comps file. Do an import using the URL format specified in the > manpage and you should be good to go. I would suggest cleaning up > /var/www/cobbler/ks_mirror manually before doing that, just to free up > the extra space on your drives. > > Hope this helps, if not, let me know... This time I did: cobbler import --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 --name=fedoracore6 which i think is as close to the example given as I can get. I still get the same error (though obviously it's looking for the comps.xml file in a slightly different location this time): Error: groupfile /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot be found. James From mdehaan at redhat.com Fri Feb 16 15:27:29 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 16 Feb 2007 10:27:29 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D5B1B3.5060504@cloud9.co.uk> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> Message-ID: <45D5CD61.9090904@redhat.com> James Fidell wrote: > Michael DeHaan wrote: > > >> Apologies on being lazy here, though per the manpage... >> >> mirror >> >> The addresss of the mirror. This needs to be either an rsync:// url >> or an ssh location usable with rsync. >> The mirror address should specify an exact repository to mirror -- >> just one architecture >> and just one distribution. If you have a seperate repo to mirror for >> a different arch, add that >> repo seperately. >> >> Here's an example of what looks like a good URL: >> >> rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync >> protocol) >> user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) >> >> What I thought was a related problem is just something I'm seeing during >> the unit tests, so that was a bit of wild goose chase. >> >> Your problem is that you didn't rsync enough data, and don't have the >> comps file. Do an import using the URL format specified in the >> manpage and you should be good to go. I would suggest cleaning up >> /var/www/cobbler/ks_mirror manually before doing that, just to free up >> the extra space on your drives. >> >> Hope this helps, if not, let me know... >> > > This time I did: > > cobbler import > --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 > --name=fedoracore6 > > which i think is as close to the example given as I can get. I still > get the same error (though obviously it's looking for the comps.xml file > in a slightly different location this time): > > Error: groupfile > /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot > be found. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Ok, thanks, I'll check into this further then... --MPD From james at cloud9.co.uk Fri Feb 16 16:26:24 2007 From: james at cloud9.co.uk (James Fidell) Date: Fri, 16 Feb 2007 16:26:24 +0000 Subject: [et-mgmt-tools] cobbler repo add --local-file parameter Message-ID: <45D5DB30.30903@cloud9.co.uk> My kickstart files are getting created with entries lilke this: %post wget http://aaa.bbb.ccc.ddd/cobbler_track/repo_mirror/fedoracore6-updates/config.repo --output-document=/etc/yum.repos.d/.repo (note the name of the output document). My python is fairly weak, but it looks like what comes before ".repo" should be given with the --local-file parameter to "cobbler repo add". So I tried it: # cobbler repo add --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/updates/6/i386/ --name=fedoracore6-updates --local-file=local-updates this command doesn't take a parameter named '--local-file' # cobbler --help | grep 'repo add' cobbler repo add --name= --mirror=rsync://
[--local-file=name] cobbler repo add --name= --mirror=user at address:/path [--local-file=name] What am I missing here? The manpage also suggests that if --local-file is missing then the repo won't be added to /etc/yum.repos.d, but from the above excerpt of the kickstart file, that doesn't appear to be the case. And whilst on the subject, the manpage doesn't mention anything about creating the config.repo file. I'm assuming that's something I have to do manually after syncing the repository. Correct? (Centos 4.4 + cobbler-0.3.9-1 built from the src.rpm, if that's relevant.) James From mdehaan at redhat.com Fri Feb 16 16:47:08 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 16 Feb 2007 11:47:08 -0500 Subject: [et-mgmt-tools] cobbler repo add --local-file parameter In-Reply-To: <45D5DB30.30903@cloud9.co.uk> References: <45D5DB30.30903@cloud9.co.uk> Message-ID: <45D5E00C.6050900@redhat.com> James Fidell wrote: > My kickstart files are getting created with entries lilke this: > > %post > wget > http://aaa.bbb.ccc.ddd/cobbler_track/repo_mirror/fedoracore6-updates/config.repo > --output-document=/etc/yum.repos.d/.repo > > (note the name of the output document). > Yeah, this is broken :) What I should do is just not fill in a wget line to create the repo file if you don't specify --local-filename to the "repo add" command. Good catch! I'll fix this for 0.4.0 (out next week). > My python is fairly weak, but it looks like what comes before ".repo" > should be given with the --local-file parameter to "cobbler repo add". > So I tried it: > > # cobbler repo add > --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/updates/6/i386/ > --name=fedoracore6-updates --local-file=local-updates > this command doesn't take a parameter named '--local-file' > # cobbler --help | grep 'repo add' > cobbler repo add --name= --mirror=rsync://
> [--local-file=name] > cobbler repo add --name= --mirror=user at address:/path > [--local-file=name] > > What am I missing here? > > The manpage also suggests that if --local-file is missing then the repo > won't be added to /etc/yum.repos.d, but from the above excerpt of the > kickstart file, that doesn't appear to be the case. > It's a bug in the manpage. It's "--local-filename" apparently. I'll fix the docs here too. > And whilst on the subject, the manpage doesn't mention anything about > creating the config.repo file. I'm assuming that's something I have to > do manually after syncing the repository. Correct? > Yep, this is what --local-filename is intended to do. It makes sure the kickstart creates the config.repo file (in this case, local-updates.repo) in /etc/yum/repos.d on the newly provisioned target system. You are actually the first user of this feature (repo mirroring) that I've heard of, other than myself, so I greatly appreciate your helping to work out the kinks in this. Thanks! > (Centos 4.4 + cobbler-0.3.9-1 built from the src.rpm, if that's > relevant.) > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From mdehaan at redhat.com Fri Feb 16 16:51:02 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 16 Feb 2007 11:51:02 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D5B1B3.5060504@cloud9.co.uk> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> Message-ID: <45D5E0F6.7080902@redhat.com> James Fidell wrote: > Michael DeHaan wrote: > > >> Apologies on being lazy here, though per the manpage... >> >> mirror >> >> The addresss of the mirror. This needs to be either an rsync:// url >> or an ssh location usable with rsync. >> The mirror address should specify an exact repository to mirror -- >> just one architecture >> and just one distribution. If you have a seperate repo to mirror for >> a different arch, add that >> repo seperately. >> >> Here's an example of what looks like a good URL: >> >> rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync >> protocol) >> user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) >> >> What I thought was a related problem is just something I'm seeing during >> the unit tests, so that was a bit of wild goose chase. >> >> Your problem is that you didn't rsync enough data, and don't have the >> comps file. Do an import using the URL format specified in the >> manpage and you should be good to go. I would suggest cleaning up >> /var/www/cobbler/ks_mirror manually before doing that, just to free up >> the extra space on your drives. >> >> Hope this helps, if not, let me know... >> > > This time I did: > > cobbler import > --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 > --name=fedoracore6 > > which i think is as close to the example given as I can get. I still > get the same error (though obviously it's looking for the comps.xml file > in a slightly different location this time): > > Error: groupfile > /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot > be found. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > You probably noticed this already, but the manpage section I copied was for repo management, not the import command. With the import command, you can pretty much point it across a rsync mirror that holds 40 distros and it's expected to import them all. I'm still retrying the exact command you gave for the import (using mirrorservice) to see what createrepo reports. So far I'm ok with internal mirrors and ibiblio, and the tree structure looks right. Should know shortly. --Michael From mdehaan at redhat.com Fri Feb 16 20:14:24 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 16 Feb 2007 15:14:24 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D5E0F6.7080902@redhat.com> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> <45D5E0F6.7080902@redhat.com> Message-ID: <45D610A0.5040509@redhat.com> Michael DeHaan wrote: > James Fidell wrote: >> Michael DeHaan wrote: >> >> >>> Apologies on being lazy here, though per the manpage... >>> >>> mirror >>> >>> The addresss of the mirror. This needs to be either an rsync:// url >>> or an ssh location usable with rsync. >>> The mirror address should specify an exact repository to mirror -- >>> just one architecture >>> and just one distribution. If you have a seperate repo to mirror >>> for >>> a different arch, add that >>> repo seperately. >>> >>> Here's an example of what looks like a good URL: >>> >>> rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync >>> protocol) >>> user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) >>> >>> What I thought was a related problem is just something I'm seeing >>> during >>> the unit tests, so that was a bit of wild goose chase. >>> >>> Your problem is that you didn't rsync enough data, and don't have the >>> comps file. Do an import using the URL format specified in the >>> manpage and you should be good to go. I would suggest cleaning up >>> /var/www/cobbler/ks_mirror manually before doing that, just to free up >>> the extra space on your drives. >>> >>> Hope this helps, if not, let me know... >>> >> >> This time I did: >> >> cobbler import >> --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 >> >> --name=fedoracore6 >> >> which i think is as close to the example given as I can get. I still >> get the same error (though obviously it's looking for the comps.xml file >> in a slightly different location this time): >> >> Error: groupfile >> /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot >> be found. >> >> James >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > You probably noticed this already, but the manpage section I copied > was for repo management, not the import command. With the import > command, you can pretty much point it across a rsync mirror that holds > 40 distros and it's expected to import them all. > > I'm still retrying the exact command you gave for the import (using > mirrorservice) to see what createrepo reports. So far I'm ok with > internal mirrors and ibiblio, and the tree structure looks right. > Should know shortly. > > --Michael > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools Ok, since I wanted to be through, I ran the exact same command you ran, against the same mirror URL... [root at mdehaan cobbler]# cobbler import --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 --name=test Here's the output I recieved on FC6 (note, I did tweak cobbler to print out the rsync command being run...other than that, the mirror import code is unchanged) sent 36680 bytes received 1835445089 bytes 190630.08 bytes/sec total size is 2668601165 speedup is 1.45 /var/www/cobbler/ks_mirror/test /var/www/cobbler/ks_mirror/test/i386 /var/www/cobbler/ks_mirror/test/i386/iso /var/www/cobbler/ks_mirror/test/i386/os /var/www/cobbler/ks_mirror/test/i386/os/images /var/www/cobbler/ks_mirror/test/i386/os/images/xen (distro added) (profile added) - looking for comps in /var/www/cobbler/ks_mirror/test/i386/os - createrepo --groupfile /var/www/cobbler/ks_mirror/test/i386/os/repodata/comps.xml /var/www/cobbler/ks_mirror/test/i386/os No problems. [mdehaan at mdehaan bar]$ createrepo -V 0.4.3 Is there a chance you have a different version of createrepo installed that when run with the arguments listed above, produces an error? I'm willing to make tweaks to the code to get the arguments correct for Centos 4.X builds, if you could point me to what those correct arguments need to be. Doesn't seem like it could be too different. If you can run the createrepo command manually with the above URL, I'd be interested to know the output (that, and to get the version number)... Thanks, --Michael From mdehaan at redhat.com Fri Feb 16 20:21:41 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 16 Feb 2007 15:21:41 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D610A0.5040509@redhat.com> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> <45D5E0F6.7080902@redhat.com> <45D610A0.5040509@redhat.com> Message-ID: <45D61255.5050404@redhat.com> Michael DeHaan wrote: > Michael DeHaan wrote: >> James Fidell wrote: >>> Michael DeHaan wrote: >>> >>> >>>> Apologies on being lazy here, though per the manpage... >>>> >>>> mirror >>>> >>>> The addresss of the mirror. This needs to be either an rsync:// >>>> url >>>> or an ssh location usable with rsync. >>>> The mirror address should specify an exact repository to mirror -- >>>> just one architecture >>>> and just one distribution. If you have a seperate repo to >>>> mirror for >>>> a different arch, add that >>>> repo seperately. >>>> >>>> Here's an example of what looks like a good URL: >>>> >>>> rsync://yourmirror.example.com/fedora-linux-core/6/i386 (for rsync >>>> protocol) >>>> user at yourmirror.example.com/fedora-linux-core/6/i386 (for SSH) >>>> >>>> What I thought was a related problem is just something I'm seeing >>>> during >>>> the unit tests, so that was a bit of wild goose chase. >>>> >>>> Your problem is that you didn't rsync enough data, and don't have the >>>> comps file. Do an import using the URL format specified in the >>>> manpage and you should be good to go. I would suggest cleaning up >>>> /var/www/cobbler/ks_mirror manually before doing that, just to free up >>>> the extra space on your drives. >>>> >>>> Hope this helps, if not, let me know... >>>> >>> >>> This time I did: >>> >>> cobbler import >>> --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 >>> >>> --name=fedoracore6 >>> >>> which i think is as close to the example given as I can get. I still >>> get the same error (though obviously it's looking for the comps.xml >>> file >>> in a slightly different location this time): >>> >>> Error: groupfile >>> /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml >>> cannot >>> be found. >>> >>> James >>> >>> _______________________________________________ >>> et-mgmt-tools mailing list >>> et-mgmt-tools at redhat.com >>> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >>> >> You probably noticed this already, but the manpage section I copied >> was for repo management, not the import command. With the import >> command, you can pretty much point it across a rsync mirror that >> holds 40 distros and it's expected to import them all. >> >> I'm still retrying the exact command you gave for the import (using >> mirrorservice) to see what createrepo reports. So far I'm ok with >> internal mirrors and ibiblio, and the tree structure looks right. >> Should know shortly. >> >> --Michael >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools > > Ok, since I wanted to be through, I ran the exact same command you > ran, against the same mirror URL... > > [root at mdehaan cobbler]# cobbler import > --mirror=rsync://rsync.mirrorservice.org/sites/download.fedora.redhat.com/pub/fedora/linux/core/6/i386 > --name=test > > Here's the output I recieved on FC6 (note, I did tweak cobbler to > print out the rsync command being run...other than that, the mirror > import code is unchanged) > > sent 36680 bytes received 1835445089 bytes 190630.08 bytes/sec > total size is 2668601165 speedup is 1.45 > /var/www/cobbler/ks_mirror/test > /var/www/cobbler/ks_mirror/test/i386 > /var/www/cobbler/ks_mirror/test/i386/iso > /var/www/cobbler/ks_mirror/test/i386/os > /var/www/cobbler/ks_mirror/test/i386/os/images > /var/www/cobbler/ks_mirror/test/i386/os/images/xen > (distro added) > (profile added) > - looking for comps in /var/www/cobbler/ks_mirror/test/i386/os > - createrepo --groupfile > /var/www/cobbler/ks_mirror/test/i386/os/repodata/comps.xml > /var/www/cobbler/ks_mirror/test/i386/os > > No problems. > > [mdehaan at mdehaan bar]$ createrepo -V > 0.4.3 > > Is there a chance you have a different version of createrepo installed > that when run with the arguments listed above, produces an error? I'm > willing to make tweaks to the code to get the arguments correct for > Centos 4.X builds, if you could point me to what those correct > arguments need to be. Doesn't seem like it could be too different. > > If you can run the createrepo command manually with the above URL, I'd > be interested to know the output (that, and to get the version number)... > > Thanks, > > --Michael > > > > BTW, if interested, here's the patch that prints out what the createrepo command is -- if you have already rsync'd the content, you can just run the command again, rsync will be quick, and it will do the createrepo relatively quickly. I expect you'll see the same command being run with possibly different output... diff -r 5d481259fc319c30f69a4a1d569ccd3b7202a050 cobbler/action_import.py --- a/cobbler/action_import.py Thu Feb 15 18:06:30 2007 -0500 +++ b/cobbler/action_import.py Fri Feb 16 10:32:50 2007 -0500 @@ -222,6 +222,7 @@ class Importer: # they'll share same repo files. if not processed_repos.has_key(comps_path): cmd = "createrepo --groupfile %s %s" % (comps_file, comps_path) + print "- %s" % cmd sub_process.call(cmd,shell=True) print "- repository updated" processed_repos[comps_path] = 1 diff -r 5d481259fc319c30f69a4a1d569ccd3b7202a050 cobbler/action_sync.py From james at cloud9.co.uk Mon Feb 19 11:20:01 2007 From: james at cloud9.co.uk (James Fidell) Date: Mon, 19 Feb 2007 11:20:01 +0000 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D610A0.5040509@redhat.com> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> <45D5E0F6.7080902@redhat.com> <45D610A0.5040509@redhat.com> Message-ID: <45D987E1.20302@cloud9.co.uk> Michael DeHaan wrote: > Is there a chance you have a different version of createrepo installed > that when run with the arguments listed above, produces an error? I'm > willing to make tweaks to the code to get the arguments correct for > Centos 4.X builds, if you could point me to what those correct arguments > need to be. Doesn't seem like it could be too different. Looks like I'm running the same version as you: # createrepo -V 0.4.3 > If you can run the createrepo command manually with the above URL, I'd > be interested to know the output (that, and to get the version number)... # ls -l /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml -rw-r--r-- 1 root root 853899 Oct 18 01:27 /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml # createrepo --groupfile /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml /var/www/cobbler/ks_mirror/fedoracore6/i386/os Error: groupfile /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot be found. etc. I've just noticed this in the createrepo manpage though: EXAMPLES Here is an example of a repository with a groups file. Note that the groups file should be in the same directory as the rpm packages (i.e. /path/to/rpms/comps.xml). createrepo -g comps.xml /path/to/rpms # createrepo --groupfile comps.xml /var/www/cobbler/ks_mirror/fedoracore6/i386/os still failed, but # createrepo --groupfile repodata/comps.xml /var/www/cobbler/ks_mirror/fedoracore6/i386/os Seems to work fine for me. I don't understand why what is apparently the same version of createrepo should work differently for us unless something is actually different between the centos release and whichever one you're using. James From mdehaan at redhat.com Mon Feb 19 15:30:22 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 10:30:22 -0500 Subject: [et-mgmt-tools] cobbler import --mirror error In-Reply-To: <45D987E1.20302@cloud9.co.uk> References: <45D33E5C.3050401@cloud9.co.uk> <45D34ADB.40601@redhat.com> <45D38643.10309@redhat.com> <45D5B1B3.5060504@cloud9.co.uk> <45D5E0F6.7080902@redhat.com> <45D610A0.5040509@redhat.com> <45D987E1.20302@cloud9.co.uk> Message-ID: <45D9C28E.2010403@redhat.com> James Fidell wrote: > Michael DeHaan wrote: > > >> Is there a chance you have a different version of createrepo installed >> that when run with the arguments listed above, produces an error? I'm >> willing to make tweaks to the code to get the arguments correct for >> Centos 4.X builds, if you could point me to what those correct arguments >> need to be. Doesn't seem like it could be too different. >> > > Looks like I'm running the same version as you: > > # createrepo -V > 0.4.3 > > >> If you can run the createrepo command manually with the above URL, I'd >> be interested to know the output (that, and to get the version number)... >> > > # ls -l /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml > -rw-r--r-- 1 root root 853899 Oct 18 01:27 > /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml > > # createrepo --groupfile > /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml > /var/www/cobbler/ks_mirror/fedoracore6/i386/os > Error: groupfile > /var/www/cobbler/ks_mirror/fedoracore6/i386/os/repodata/comps.xml cannot > be found. > > etc. > > I've just noticed this in the createrepo manpage though: > > EXAMPLES > Here is an example of a repository with a groups file. Note > that the > groups file should be in the same directory as the rpm packages > (i.e. > /path/to/rpms/comps.xml). > > createrepo -g comps.xml /path/to/rpms > > # createrepo --groupfile comps.xml > /var/www/cobbler/ks_mirror/fedoracore6/i386/os > > still failed, but > > # createrepo --groupfile repodata/comps.xml > /var/www/cobbler/ks_mirror/fedoracore6/i386/os > > Seems to work fine for me. I don't understand why what is apparently > the same version of createrepo should work differently for us unless > something is actually different between the centos release and whichever > one you're using. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Strange :) Looking over the createrepo source, it appears that some of the arguments join those parameter's paths with basedir, rather than accepting relative paths. The default value for basedir is apparently the current working directory. I'm going to try adding a "--basedir /" to the createrepo command to see if that makes it happier. Since you've worked around this manually, your kickstart config should be good to go for now. Basically you just have to do the createrepo in case any files in /etc/cobbler/rsync.exclude left out some RPMs to save disk/bandwidth. --Michael From hbrock at redhat.com Mon Feb 19 16:58:16 2007 From: hbrock at redhat.com (Hugh Brock) Date: Mon, 19 Feb 2007 11:58:16 -0500 Subject: [et-mgmt-tools] virt-manager os-choice feature Message-ID: <45D9D728.5070903@redhat.com> Hi everyone. I've just committed a patch that adds UI in virt-manager to control the "choose your fully virtualized OS" flags in python-virtinst. Details are at http://hg.et.redhat.com/virt/applications/virt-manager--devel?cs=80e0238ab362 At the moment, choosing an OS type affects only the acpi= and apic= flags, according to a dictionary of those settings in the header of virtinst/FullVirtGuest. However I imagine we will add more os-dependent settings later as the occasion arises. Enjoy, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mdehaan at redhat.com Mon Feb 19 19:04:01 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 14:04:01 -0500 Subject: [et-mgmt-tools] Cobbler 0.4.0 released Message-ID: <45D9F4A1.2000309@redhat.com> Cobbler 0.4.0 is out. RPMs are already up on et.redhat.com and will be pushed out to the mirrors shortly. The main feature for this release is support for PXE menus. For those wanting to use PXE menus, the best thing to do is set all machines in your datacenter/lab to "PXE first". To access the PXE menu, type "menu" at the prompt, otherwise cobbler will fall through to local boot in 20 seconds. If you have a cobbler definition made for a system, that system will bypass the PXE menu. This is true whether you have a certain MAC set to a specific profile (ex: cobbler system add --name=AA:BB:CC:DD:EE:FF ...) or if you have a "default" system entry (cobbler system add --name=default --profile=blah). If you have a default system entry and want to use the new menus, just run "cobbler system delete --name=default" and then "cobbler sync". Presto, you've got PXE menus. This is also documented in the manpage. As mentioned previously on the list, I'm also reinstating the usage of Cheetah for vastly improved kickstart templating. See http://www.cheetahtemplate.org/learn for more information. Basically the existing kickstart templating is now improved with the ability to do if-statements and the like. Cobbler will allow either TEMPLATE::foo or $foo for kickstart templating substitution. The one thing I've found is that a macro such as $(list-hard-drives) does need to be escaped to \$(list-hard-drives) in kickstart files. Things like passwords (crypted passwords often contain "$") don't need to be escaped and are otherwise fine. If you do escape them though, it won't hurt. The new /etc/cobbler/kickstart* files have them escaped. If you have your own kickstart files, it would be worth checking for any dollarsigns that were unescaped, or you might see an error next time you run "cobbler sync". This release also no longer requires running "cobbler sync" after "profile add" or "system add" commands as much as cobbler used to. Editing configuration files behind the scenes, or wanting to regenerate PXE menus will, however, still require running "cobbler sync". The features made to minimize the need for sync were mainly done for applications using cobbler as a provisioning library and those users that occasionally forget to run sync and post questions to the list :) I've also incorporated some random minor bugfixes, including passing "--basedir" to createrepo and manpage fixes related to the "--local-filename" option as mentioned last week. Questions? Comments? Fire away. --Michael From mdehaan at redhat.com Mon Feb 19 19:52:56 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 14:52:56 -0500 Subject: [et-mgmt-tools] Re: Cobbler 0.4.0 released In-Reply-To: <45D9F4A1.2000309@redhat.com> References: <45D9F4A1.2000309@redhat.com> Message-ID: <45DA0018.7010806@redhat.com> Michael DeHaan wrote: > Cobbler 0.4.0 is out. RPMs are already up on et.redhat.com and will > be pushed out to the mirrors shortly. > > The main feature for this release is support for PXE menus. For > those wanting to use PXE menus, the best thing to do is set all > machines in your datacenter/lab to "PXE first". To access the PXE > menu, type "menu" at the prompt, otherwise cobbler will fall through > to local boot in 20 seconds. If you have a cobbler definition made > for a system, that system will bypass the PXE menu. This is true > whether you have a certain MAC set to a specific profile (ex: cobbler > system add --name=AA:BB:CC:DD:EE:FF ...) or if you have a "default" > system entry (cobbler system add --name=default --profile=blah). If > you have a default system entry and want to use the new menus, just > run "cobbler system delete --name=default" and then "cobbler sync". > Presto, you've got PXE menus. This is also documented in the manpage. > > As mentioned previously on the list, I'm also reinstating the usage of > Cheetah for vastly improved kickstart templating. See > http://www.cheetahtemplate.org/learn for more information. Basically > the existing kickstart templating is now improved with the ability to > do if-statements and the like. Cobbler will allow either > TEMPLATE::foo or $foo for kickstart templating substitution. The one > thing I've found is that a macro such as $(list-hard-drives) does need > to be escaped to \$(list-hard-drives) in kickstart files. Things > like passwords (crypted passwords often contain "$") don't need to be > escaped and are otherwise fine. If you do escape them though, it > won't hurt. The new /etc/cobbler/kickstart* files have them > escaped. If you have your own kickstart files, it would be worth > checking for any dollarsigns that were unescaped, or you might see an > error next time you run "cobbler sync". > > This release also no longer requires running "cobbler sync" after > "profile add" or "system add" commands as much as cobbler used to. > Editing configuration files behind the scenes, or wanting to > regenerate PXE menus will, however, still require running "cobbler > sync". The features made to minimize the need for sync were mainly > done for applications using cobbler as a provisioning library and > those users that occasionally forget to run sync and post questions to > the list :) > I've also incorporated some random minor bugfixes, including passing > "--basedir" to createrepo and manpage fixes related to the > "--local-filename" option as mentioned last week. > > Questions? Comments? Fire away. > > --Michael > RHEL4/Centos users: Apparently menu.c32 (required for PXE menus) isn't provided by the syslinux package on RHEL4... I'll probably end up bundling it and storing it in /var/lib/cobbler. For now, you can copy it yourself to /usr/lib/syslinux/menu.c32 and things will work fine. I'll release a 0.4.1 fix that will bundle this file shortly. From mdehaan at redhat.com Mon Feb 19 20:38:46 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 15:38:46 -0500 Subject: [et-mgmt-tools] Re: Cobbler 0.4.0 released In-Reply-To: <45DA0018.7010806@redhat.com> References: <45D9F4A1.2000309@redhat.com> <45DA0018.7010806@redhat.com> Message-ID: <45DA0AD6.7060200@redhat.com> Michael DeHaan wrote: > Michael DeHaan wrote: >> Cobbler 0.4.0 is out. RPMs are already up on et.redhat.com and will >> be pushed out to the mirrors shortly. >> >> The main feature for this release is support for PXE menus. For >> those wanting to use PXE menus, the best thing to do is set all >> machines in your datacenter/lab to "PXE first". To access the PXE >> menu, type "menu" at the prompt, otherwise cobbler will fall through >> to local boot in 20 seconds. If you have a cobbler definition made >> for a system, that system will bypass the PXE menu. This is true >> whether you have a certain MAC set to a specific profile (ex: >> cobbler system add --name=AA:BB:CC:DD:EE:FF ...) or if you have a >> "default" system entry (cobbler system add --name=default >> --profile=blah). If you have a default system entry and want to >> use the new menus, just run "cobbler system delete --name=default" >> and then "cobbler sync". Presto, you've got PXE menus. This is >> also documented in the manpage. >> >> As mentioned previously on the list, I'm also reinstating the usage >> of Cheetah for vastly improved kickstart templating. See >> http://www.cheetahtemplate.org/learn for more information. >> Basically the existing kickstart templating is now improved with the >> ability to do if-statements and the like. Cobbler will allow either >> TEMPLATE::foo or $foo for kickstart templating substitution. The >> one thing I've found is that a macro such as $(list-hard-drives) does >> need to be escaped to \$(list-hard-drives) in kickstart files. >> Things like passwords (crypted passwords often contain "$") don't >> need to be escaped and are otherwise fine. If you do escape them >> though, it won't hurt. The new /etc/cobbler/kickstart* files have >> them escaped. If you have your own kickstart files, it would be >> worth checking for any dollarsigns that were unescaped, or you might >> see an error next time you run "cobbler sync". >> >> This release also no longer requires running "cobbler sync" after >> "profile add" or "system add" commands as much as cobbler used to. >> Editing configuration files behind the scenes, or wanting to >> regenerate PXE menus will, however, still require running "cobbler >> sync". The features made to minimize the need for sync were mainly >> done for applications using cobbler as a provisioning library and >> those users that occasionally forget to run sync and post questions >> to the list :) I've also incorporated some random minor bugfixes, >> including passing "--basedir" to createrepo and manpage fixes related >> to the "--local-filename" option as mentioned last week. >> >> Questions? Comments? Fire away. >> >> --Michael >> > RHEL4/Centos users: > > Apparently menu.c32 (required for PXE menus) isn't provided by the > syslinux package on RHEL4... I'll probably end up bundling it and > storing it in /var/lib/cobbler. > For now, you can copy it yourself to /usr/lib/syslinux/menu.c32 and > things will work fine. I'll release a 0.4.1 fix that will bundle this > file shortly. > > Apparently bundling Cheetah is causing some evil problems on older systems, I'm going to back that bit out and add it as a Requires: in the RPM. This will be released shortly as 0.4.1. RHEL4/Centos users ca get Cheetah here: http://www.python.org/pyvault/centos-4-i386/repodata/repoview/python-cheetah-0-0.9.18-1.el4.pyv.html .. actually I think Centos already has it in the default repo(s) ... correct me if I'm wrong. --Michael From mdehaan at redhat.com Mon Feb 19 21:45:08 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 16:45:08 -0500 Subject: [et-mgmt-tools] Re: Cobbler 0.4.0 released In-Reply-To: <45DA0AD6.7060200@redhat.com> References: <45D9F4A1.2000309@redhat.com> <45DA0018.7010806@redhat.com> <45DA0AD6.7060200@redhat.com> Message-ID: <45DA1A64.30206@redhat.com> Michael DeHaan wrote: > Michael DeHaan wrote: >> Michael DeHaan wrote: >>> Cobbler 0.4.0 is out. RPMs are already up on et.redhat.com and >>> will be pushed out to the mirrors shortly. >>> >>> The main feature for this release is support for PXE menus. For >>> those wanting to use PXE menus, the best thing to do is set all >>> machines in your datacenter/lab to "PXE first". To access the PXE >>> menu, type "menu" at the prompt, otherwise cobbler will fall through >>> to local boot in 20 seconds. If you have a cobbler definition made >>> for a system, that system will bypass the PXE menu. This is true >>> whether you have a certain MAC set to a specific profile (ex: >>> cobbler system add --name=AA:BB:CC:DD:EE:FF ...) or if you have a >>> "default" system entry (cobbler system add --name=default >>> --profile=blah). If you have a default system entry and want to >>> use the new menus, just run "cobbler system delete --name=default" >>> and then "cobbler sync". Presto, you've got PXE menus. This is >>> also documented in the manpage. >>> >>> As mentioned previously on the list, I'm also reinstating the usage >>> of Cheetah for vastly improved kickstart templating. See >>> http://www.cheetahtemplate.org/learn for more information. >>> Basically the existing kickstart templating is now improved with the >>> ability to do if-statements and the like. Cobbler will allow either >>> TEMPLATE::foo or $foo for kickstart templating substitution. The >>> one thing I've found is that a macro such as $(list-hard-drives) >>> does need to be escaped to \$(list-hard-drives) in kickstart >>> files. Things like passwords (crypted passwords often contain "$") >>> don't need to be escaped and are otherwise fine. If you do escape >>> them though, it won't hurt. The new /etc/cobbler/kickstart* files >>> have them escaped. If you have your own kickstart files, it would >>> be worth checking for any dollarsigns that were unescaped, or you >>> might see an error next time you run "cobbler sync". >>> >>> This release also no longer requires running "cobbler sync" after >>> "profile add" or "system add" commands as much as cobbler used >>> to. Editing configuration files behind the scenes, or wanting to >>> regenerate PXE menus will, however, still require running "cobbler >>> sync". The features made to minimize the need for sync were mainly >>> done for applications using cobbler as a provisioning library and >>> those users that occasionally forget to run sync and post questions >>> to the list :) I've also incorporated some random minor bugfixes, >>> including passing "--basedir" to createrepo and manpage fixes >>> related to the "--local-filename" option as mentioned last week. >>> >>> Questions? Comments? Fire away. >>> >>> --Michael >>> >> RHEL4/Centos users: >> >> Apparently menu.c32 (required for PXE menus) isn't provided by the >> syslinux package on RHEL4... I'll probably end up bundling it and >> storing it in /var/lib/cobbler. For now, you can copy it yourself to >> /usr/lib/syslinux/menu.c32 and things will work fine. I'll release a >> 0.4.1 fix that will bundle this file shortly. >> >> > > Apparently bundling Cheetah is causing some evil problems on older > systems, I'm going to back that bit out and add it as a Requires: in > the RPM. This will be released shortly as 0.4.1. > > RHEL4/Centos users ca get Cheetah here: > http://www.python.org/pyvault/centos-4-i386/repodata/repoview/python-cheetah-0-0.9.18-1.el4.pyv.html > .. actually I think Centos already has it in the default repo(s) ... > correct me if I'm wrong. > > --Michael > The Cheetah changes are fixed and pushed, along with the inclusion of the menu.c32 file. RHEL4 users can satisfy the Cheetah requirements by doing the following: wget http://www.python.org/pyvault/centos-4-i386/python23-cheetah-0.9.18-1.el4.pyv.i386.rpm wget http://www.python.org/pyvault/centos-4-i386/python-cheetah-0-0.9.18-1.el4.pyv.rpm rpm -i python*cheetah*.rpm --nodeps Fedora/Centos users can just let yum suck in the dependencies. I'll update this info and put it on the website for future reference. Remember to change any "$(foo)" macros code in your kickstart files to "\$(foo)" or you _will_ get errors during cobbler sync. --Michael From mdehaan at redhat.com Mon Feb 19 23:04:34 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Mon, 19 Feb 2007 18:04:34 -0500 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux Message-ID: <45DA2D02.1000305@redhat.com> I've noticed that the pxelinux that ships with RHEL4 (test platform: RHEL4U4) does not work well at all with new PXE menus, though the version shipping with FC6 works great. Basically it doesn't understand the menu label options. If you have RHEL4's version of syslinux installed, the fix here is to install a newer syslinux. I'll file the bug for good measure (if this is even a bug?) and may bundle a known-good pxelinux.0 in the next release. The workaround is pretty simple though. If anyone still has questions, feel free to ask. --Michael From berrange at redhat.com Tue Feb 20 04:14:27 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 20 Feb 2007 04:14:27 +0000 Subject: [et-mgmt-tools] virt-manager os-choice feature In-Reply-To: <45D9D728.5070903@redhat.com> References: <45D9D728.5070903@redhat.com> Message-ID: <20070220041427.GA353@redhat.com> On Mon, Feb 19, 2007 at 11:58:16AM -0500, Hugh Brock wrote: > Hi everyone. > > I've just committed a patch that adds UI in virt-manager to control the > "choose your fully virtualized OS" flags in python-virtinst. Details are at > http://hg.et.redhat.com/virt/applications/virt-manager--devel?cs=80e0238ab362 > > At the moment, choosing an OS type affects only the acpi= and apic= > flags, according to a dictionary of those settings in the header of > virtinst/FullVirtGuest. However I imagine we will add more os-dependent > settings later as the occasion arises. Looks good, but can you change the two drop down lists to be sorted alphabetically :-) Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Tue Feb 20 04:21:21 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 20 Feb 2007 04:21:21 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager Message-ID: <20070220042121.GB353@redhat.com> I've now completed the first iteration of support for QEMU / KVM in the virt-install and virt-manager tools. For virt-install there are a couple of new command line args. First you need to request the QEMU driver, with '--connect [URI]'. Then you can optionally specify a CPU architecture and whether to enable acceleration (it picks KVM if available, otherwise falls back to KQEMU). So as an example virt-install --connect qemu:///session \ --arch x86_65 --accelerate --name "Windows XP" --cdrom /home/demo/xp.iso --file /home/demo/xp.img --ram 500 For virt-manager, the initial connect dialog box now allows you to choose between a QEMU and Xen connection. From there everything else in the virt-manager UI is basically the same as for Xen. The only difference is that when creating a new VM, you can't choose paravirt - on the otherhand you do get a choice of CPU architecture x86/x86_64/ppc/sparc/mips :-) NB, it requires either the latest libvirt CVS code, or the libvirt 0.2.0-2 RPM from Fedora rawhide (the plain 0.2.0 release had a couple of bugs). As I said this is the first iteration just to get the UI up and running. Once people have played around with it, I fully expect some UI tweaks to improve the user experiance. The code is in mercurial, otherwise I plan todo a new release of both apps in the very near future. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From markmc at redhat.com Tue Feb 20 10:39:46 2007 From: markmc at redhat.com (Mark McLoughlin) Date: Tue, 20 Feb 2007 10:39:46 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager In-Reply-To: <20070220042121.GB353@redhat.com> References: <20070220042121.GB353@redhat.com> Message-ID: <1171967986.3259.46.camel@blaa> Hi Dan, Cool stuff ... On Tue, 2007-02-20 at 04:21 +0000, Daniel P. Berrange wrote: > As I said this is the first iteration just to get the UI up and running. > Once people have played around with it, I fully expect some UI tweaks to > improve the user experiance. Some suggestions: - It'd be really nice not to have the connect dialog when you start up virt-manager - e.g. think of a first time user who selects "Virtual Machine Manager" from the menus and is then asked about "Open a connection" ... that's fairly intimidating - Perhaps by default connect to Xen if run as root, and connect to QEMU when run as non-root - I'd be tempted[1] to not expose the notion of different hypervisors in the UI beyond that. Maybe have a --qemu command line option or something. If we do need to expose this in the UI, I think the most natural place is in the "create VM" dialog[2]. - Failing that, maybe expose the hypervisor type as radio items in the File menu, something like File Edit View Help -> New Machine Restore saved machine --- [X] Manage Xen VMs [ ] Manage QEMU VMs [ ] Manage Remote VMs ... --- Close Quit - Also, suggest "Connect to" rather than "Open a connection", the latter concept would probably only make sense to programmers - I think I'd only use the "Connect to" concept with remote management - connecting to a remote machine is a well understood concept, but connecting to a hypervisor is a little weird. Perhaps "Manage Xen guests" vs. Manage QEMU guests" as above Cheers, Mark. [1] - As discussed here : http://www.redhat.com/archives/libvir-list/2007-January/msg00093.html From rjones at redhat.com Tue Feb 20 11:04:12 2007 From: rjones at redhat.com (Richard Jones) Date: Tue, 20 Feb 2007 11:04:12 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager In-Reply-To: <1171967986.3259.46.camel@blaa> References: <20070220042121.GB353@redhat.com> <1171967986.3259.46.camel@blaa> Message-ID: <2E01A34E-3C4D-4EC5-94E9-C340F4CB022E@redhat.com> On 20 Feb 2007, at 10:39, Mark McLoughlin wrote: > Hi Dan, > Cool stuff ... > > On Tue, 2007-02-20 at 04:21 +0000, Daniel P. Berrange wrote: > >> As I said this is the first iteration just to get the UI up and >> running. >> Once people have played around with it, I fully expect some UI >> tweaks to >> improve the user experiance. > > Some suggestions: > > - It'd be really nice not to have the connect dialog when you start > up virt-manager - e.g. think of a first time user who selects > "Virtual Machine Manager" from the menus and is then asked about > "Open a connection" ... that's fairly intimidating Agreed. > - Perhaps by default connect to Xen if run as root, and connect to > QEMU when run as non-root Or perhaps even connect with name = NULL following our earlier suggestion to make this "do something sensible" :-? > > - I think I'd only use the "Connect to" concept with remote > management - connecting to a remote machine is a well understood > concept, but connecting to a hypervisor is a little weird. Perhaps > "Manage Xen guests" vs. Manage QEMU guests" as above What's the general plan for being able to manage both through the same virt-manager interface? [1] Rich. [1] I'm obviously still not able to run any of this stuff on any machine I have, so apologies if this is a dumb question. From markmc at redhat.com Tue Feb 20 11:11:22 2007 From: markmc at redhat.com (Mark McLoughlin) Date: Tue, 20 Feb 2007 11:11:22 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager In-Reply-To: <2E01A34E-3C4D-4EC5-94E9-C340F4CB022E@redhat.com> References: <20070220042121.GB353@redhat.com> <1171967986.3259.46.camel@blaa> <2E01A34E-3C4D-4EC5-94E9-C340F4CB022E@redhat.com> Message-ID: <1171969882.3259.63.camel@blaa> On Tue, 2007-02-20 at 11:04 +0000, Richard Jones wrote: > On 20 Feb 2007, at 10:39, Mark McLoughlin wrote: > > - I think I'd only use the "Connect to" concept with remote > > management - connecting to a remote machine is a well understood > > concept, but connecting to a hypervisor is a little weird. Perhaps > > "Manage Xen guests" vs. Manage QEMU guests" as above > > What's the general plan for being able to manage both through the > same virt-manager interface? If we're going to expose the different hypervisors through the UI, my instinct would be that there should be a "run on Xen" vs. "run on QEMU" toggle in the "Create VM" dialog and that all guests should appear in a flat list with a "Hypervisor Type" column. But there is the argument that people shouldn't ever really need to choose and so it shouldn't be in the UI at all. Cheers, Mark. From berrange at redhat.com Tue Feb 20 12:28:22 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 20 Feb 2007 12:28:22 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager In-Reply-To: <1171967986.3259.46.camel@blaa> References: <20070220042121.GB353@redhat.com> <1171967986.3259.46.camel@blaa> Message-ID: <20070220122822.GB28263@redhat.com> On Tue, Feb 20, 2007 at 10:39:46AM +0000, Mark McLoughlin wrote: > Hi Dan, > Cool stuff ... > > On Tue, 2007-02-20 at 04:21 +0000, Daniel P. Berrange wrote: > > > As I said this is the first iteration just to get the UI up and running. > > Once people have played around with it, I fully expect some UI tweaks to > > improve the user experiance. > > Some suggestions: > > - It'd be really nice not to have the connect dialog when you start > up virt-manager - e.g. think of a first time user who selects > "Virtual Machine Manager" from the menus and is then asked about > "Open a connection" ... that's fairly intimidating > > - Perhaps by default connect to Xen if run as root, and connect to > QEMU when run as non-root Yes, that pretty much aligns with what i was thinking. Anticipating the remote management, the flat list of VMs would change to a tree list with two levels, the first being the host, the second being the VMs on that host. Saving the list of URIs for managed hosts in GConf so we don't need to ask the user each tim ethe app runs - and neatly avoid having to put up the initial connect dialog at startup. We'd also automatically connect to either Xen or QEMU on the local host according to which was available. > - I'd be tempted[1] to not expose the notion of different hypervisors > in the UI beyond that. Maybe have a --qemu command line option or > something. If we do need to expose this in the UI, I think the most > natural place is in the "create VM" dialog[2]. We can already specify at arbitrary libvirt connect URI on the command line via the arg '--connect URI', so we have a fallback option for experts to use. > - Also, suggest "Connect to" rather than "Open a connection", the > latter concept would probably only make sense to programmers Or 'Connect to host'. > - I think I'd only use the "Connect to" concept with remote > management - connecting to a remote machine is a well understood > concept, but connecting to a hypervisor is a little weird. Perhaps > "Manage Xen guests" vs. Manage QEMU guests" as abovea I like the idea of just picking Xen or QEMU as appropriate (root vs non-root) and allowing people to use the --connect command line arg if desired. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Tue Feb 20 12:30:52 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 20 Feb 2007 12:30:52 +0000 Subject: [et-mgmt-tools] QEMU & KVM support in virt-install/virt-manager In-Reply-To: <2E01A34E-3C4D-4EC5-94E9-C340F4CB022E@redhat.com> References: <20070220042121.GB353@redhat.com> <1171967986.3259.46.camel@blaa> <2E01A34E-3C4D-4EC5-94E9-C340F4CB022E@redhat.com> Message-ID: <20070220123052.GC28263@redhat.com> On Tue, Feb 20, 2007 at 11:04:12AM +0000, Richard Jones wrote: > On 20 Feb 2007, at 10:39, Mark McLoughlin wrote: > > - I think I'd only use the "Connect to" concept with remote > > management - connecting to a remote machine is a well understood > > concept, but connecting to a hypervisor is a little weird. Perhaps > > "Manage Xen guests" vs. Manage QEMU guests" as above > > What's the general plan for being able to manage both through the > same virt-manager interface? [1] I'd like to de-emphasise the idea of connecting to hypervisors in faavour of connecting to hosts - and automatically pick either Xen or QEMU as a default connection for localhost.As mentioned in the previous mail, the flat tree view would instead be a 2 level tree, showing hosts, and the VMs running on them. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From hbrock at redhat.com Tue Feb 20 15:36:42 2007 From: hbrock at redhat.com (Hugh Brock) Date: Tue, 20 Feb 2007 10:36:42 -0500 Subject: [et-mgmt-tools] virt-manager os-choice feature In-Reply-To: <20070220041427.GA353@redhat.com> References: <45D9D728.5070903@redhat.com> <20070220041427.GA353@redhat.com> Message-ID: <45DB158A.4000700@redhat.com> Daniel P. Berrange wrote: > On Mon, Feb 19, 2007 at 11:58:16AM -0500, Hugh Brock wrote: >> Hi everyone. >> >> I've just committed a patch that adds UI in virt-manager to control the >> "choose your fully virtualized OS" flags in python-virtinst. Details are at >> http://hg.et.redhat.com/virt/applications/virt-manager--devel?cs=80e0238ab362 >> >> At the moment, choosing an OS type affects only the acpi= and apic= >> flags, according to a dictionary of those settings in the header of >> virtinst/FullVirtGuest. However I imagine we will add more os-dependent >> settings later as the occasion arises. > > Looks good, but can you change the two drop down lists to be sorted > alphabetically :-) > Yeah, good idea... Python does something strange with the list ordering, the dictionary is created in alphabetical order, but then it gets either reversed or scrambled when it pulls the keys out to put them in the model for the pull-down widgets. Anyway, yeah, I'll sort them... --H -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From hbrock at redhat.com Tue Feb 20 18:35:21 2007 From: hbrock at redhat.com (Hugh Brock) Date: Tue, 20 Feb 2007 13:35:21 -0500 Subject: [et-mgmt-tools] os-list now sorted Message-ID: <45DB3F69.9070409@redhat.com> http://hg.et.redhat.com/virt/applications/virt-manager--devel?cs=b81a95033b72 --H -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From fukuta.saori at jp.fujitsu.com Wed Feb 21 03:07:42 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Wed, 21 Feb 2007 12:07:42 +0900 Subject: [et-mgmt-tools] [PATCH] waiting VNC port number Message-ID: <20070221120701.669D.FUKUTA.SAORI@jp.fujitsu.com> Hi, I try to install with virt-install and sometimes vncviewer get an error message that says "unable to connect to host:Connection refused(111)". So, I cannot continue installation. The attached patch to resolve this issues in the following way: 1) Check whether the domain has VNCport number. 2) If the number is exist, use the VNCport number. 3) If not, check again every 0.25 second up to 40 seconds. Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virt-install (python-virtinst-0.100.0) =================================================================== --- virt-install 2007-02-19 16:09:12.000000000 +0900 +++ virt-install.new 2007-02-19 16:12:05.000000000 +0900 @@ -306,11 +306,22 @@ def get_xml_string(dom, path): return None def vnc_console(dom): - import time; time.sleep(2) # FIXME: ugh. - vncport = get_xml_string(dom, - "/domain/devices/graphics[@type='vnc']/@port") + import time + import commands + num = 0 + vncport = None + cmd = "/usr/bin/xenstore-read /local/domain/%s/console/vnc-port" %(dom.ID()) + while num < ( 40 / 0.25 ): # 40 seconds, .25 second sleeps + ( ret, port ) = commands.getstatusoutput( cmd ) + if ret == 0: + vncport = port + break + else: + num += 1 + time.sleep(0.25) if vncport == None: - vncport = 5900 + dom.ID() + print >> sys.stderr, "Unable to connect to graphical console; vncport number is not found." + return None vncport = int(vncport) vnchost = "localhost" if not os.path.exists("/usr/bin/vncviewer"): From fukuta.saori at jp.fujitsu.com Wed Feb 21 03:09:02 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Wed, 21 Feb 2007 12:09:02 +0900 Subject: [et-mgmt-tools] [PATCH] to break out of the infinite loop Message-ID: <20070221120806.66AC.FUKUTA.SAORI@jp.fujitsu.com> Hi, There's a trivial bug in the latest virtinst/Guest.py. This patch adds a increment to fix this problem. Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virtinst/Guest.py (python-virtinst-0.100.0) =================================================================== --- Guest.py 2007-02-19 16:09:28.000000000 +0900 +++ Guest.py.new 2007-02-19 16:11:11.000000000 +0900 @@ -392,6 +392,7 @@ class Guest(object): break except libvirt.libvirtError: pass + num += 1 time.sleep(0.25) if d is None: From fukuta.saori at jp.fujitsu.com Wed Feb 21 03:09:02 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Wed, 21 Feb 2007 12:09:02 +0900 Subject: [et-mgmt-tools] [PATCH] domain name must not be numeric only Message-ID: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> Hi, The attached patch adds to check the domain name that must not be numeric only. Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virtinst/Guest.py (python-virtinst-0.100.0) =================================================================== --- Guest.py 2007-02-21 11:04:43.000000000 +0900 +++ Guest.py.num 2007-02-21 11:10:28.000000000 +0900 @@ -204,6 +204,8 @@ class Guest(object): return self._name def set_name(self, val): # FIXME: need some validation here + if re.match("^[0-9]*$", val): + raise ValueError, "Domain name must not be numeric only" if re.match("^[a-zA-Z0-9_]*$", val) == None: raise ValueError, "Domain name must be alphanumeric or _" if len(val) > 50: From berrange at redhat.com Wed Feb 21 14:05:19 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 21 Feb 2007 14:05:19 +0000 Subject: [et-mgmt-tools] [PATCH] domain name must not be numeric only In-Reply-To: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> References: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070221140519.GC7306@redhat.com> On Wed, Feb 21, 2007 at 12:09:02PM +0900, Saori Fukuta wrote: > Hi, > > The attached patch adds to check the domain name that must not be > numeric only. > > Signed-off-by: Saori Fukuta Looks good, will apply. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Wed Feb 21 14:07:13 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 21 Feb 2007 14:07:13 +0000 Subject: [et-mgmt-tools] [PATCH] to break out of the infinite loop In-Reply-To: <20070221120806.66AC.FUKUTA.SAORI@jp.fujitsu.com> References: <20070221120806.66AC.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070221140713.GD7306@redhat.com> On Wed, Feb 21, 2007 at 12:09:02PM +0900, Saori Fukuta wrote: > Hi, > > There's a trivial bug in the latest virtinst/Guest.py. > This patch adds a increment to fix this problem. > > Signed-off-by: Saori Fukuta Will apply. Thanks, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Wed Feb 21 14:11:12 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 21 Feb 2007 14:11:12 +0000 Subject: [et-mgmt-tools] [PATCH] waiting VNC port number In-Reply-To: <20070221120701.669D.FUKUTA.SAORI@jp.fujitsu.com> References: <20070221120701.669D.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070221141112.GE7306@redhat.com> On Wed, Feb 21, 2007 at 12:07:42PM +0900, Saori Fukuta wrote: > Hi, > > I try to install with virt-install and sometimes vncviewer get an error > message that says "unable to connect to host:Connection refused(111)". > So, I cannot continue installation. > > The attached patch to resolve this issues in the following way: > > 1) Check whether the domain has VNCport number. > 2) If the number is exist, use the VNCport number. > 3) If not, check again every 0.25 second up to 40 seconds. This is no good because it is Xen specific - the code breaks when used with the QEMU / KVM backend to libvirt. I think this is actually a combination of two bugs: - libvirt looks in xenstore to find the port number, but if it is missing hardcodes 5900 + domid. So there is a race condition where if virt-install asks for the XML dump, before the framebuffer daemon has started up, it will get the wrong port number. This is a hardcoded fallback port number is a hangover from old Xen 3.0.2 where the port number was fixed & not kept in xenstore at all. We need to fix libvirt to only use this code on Xen <= 3.0.3 - Once the above fix in libvirt is done, we can make virt-install do a loop like the one you have below, but instead of looking in xenstore it will look at the port in the XML - waiting for it to change from '-1' to an actual port number Regards Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From fj1826dm at aa.jp.fujitsu.com Thu Feb 22 06:44:12 2007 From: fj1826dm at aa.jp.fujitsu.com (Masayuki Sunou) Date: Thu, 22 Feb 2007 15:44:12 +0900 Subject: [et-mgmt-tools] About CPU Usage of virt-manager Message-ID: <200702221544.JCG34350.N3G7E29K@aa.jp.fujitsu.com> Hi, I have a question about CPU Usage of virt-manager. Virt-manager calculates CPU Usage based on physical CPU. Because of that, when the host has many CPU, guest's CPU Usage is difficult to know. (SMP etc.) example: When the host has 32 CPU, and guest's CPU Usage is 100%, virt-manager displays CPU Usage as 3%. --> 100%/32CPU=3% Therefore, I think that virt-manager should calculate CPU Usage based on virtual CPU.(like "xentop") Because "Xentop" calculates CPU Usage based on virtual CPU, I think that virt-manager should also calculate CPU Usage based on virtual CPU. Thanks, Masayuki Sunou From cobbler at msquared.id.au Thu Feb 22 10:41:51 2007 From: cobbler at msquared.id.au (Msquared) Date: Thu, 22 Feb 2007 19:41:51 +0900 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? Message-ID: <20070222104151.GA23723@sliderule.msquared.com.au> I'm having some trouble understanding where everything goes under cobbler. If I use 'cobbler import' or 'cobbler distro add', do I put the files somewhere in /var/www/cobbler first? It seems that 'cobbler import' doesn't actually copy files for me, so I'm not really sure what it does. The walkthrough at http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler doesn't seem to help me, as it doesn't tell me what I need to put where. I've tried to copy the contents of my Fedora 6 DVD to /var/www/cobbler/kickstarts/Fedora6 and run 'cobbler import --path=/var/www/cobbler/kickstarts/Fedora6', and then run 'cobbler list', but the output looks nothing like the output presented on http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler, instead it looks like this: Distros: var_www_cobbler_kickstarts_Fedora6_images_pxeboot var_www_cobbler_kickstarts_Fedora6_images_xen Repos: Profiles: var_www_cobbler_kickstarts_Fedora6_images_pxeboot var_www_cobbler_kickstarts_Fedora6_images_xen Systems: I can't seem to get 'cobbler distro add' to work either, not that I'm entirely sure what it does. Where does it get the repository files from? Do I need to do some preparation of the files from the DVD, or download a mirror first? I tried to run cobbler distro add but ended up with a very strange error: cobbler distro add --name=myxendistro --kernel=/var/www/cobbler/kickstarts/Fedora6/images/xen/vmlinuz --initrd=/var/www/cobbler/kickstarts/Fedora6/images/xen/initrd.img The result is: Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 505, in main BootCLI(sys.argv).run() File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 102, in run self.curry_args(self.args[1:], self.commands['toplevel']) File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 421, in curry_args commands[args[0]](args[1:]) File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 469, in distro return self.curry_args(args, self.commands['distro']) File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 421, in curry_args commands[args[0]](args[1:]) File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 386, in distro_edit return self.apply_args(args,commands,on_ok) File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 409, in apply_args on_ok() File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 385, in on_ok = lambda: self.api.distros().add(distro, with_copy=self.api.sync_flag) File "/usr/lib/python2.4/site-packages/cobbler/collection.py", line 97, in add lite_sync.add_single_distro(ref.name) File "/usr/lib/python2.4/site-packages/cobbler/action_litesync.py", line 60, in add_single_distro self.sync.write_distro_file(distro) File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 733, in write_distro_file clone.kernel_options = self.hash_to_string(clone.kernel_options) File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 885, in hash_to_string buffer = buffer + key + "=" + value + " " TypeError: cannot concatenate 'str' and 'int' objects Any ideas? Regards, Msquared... From berrange at redhat.com Thu Feb 22 12:26:58 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Thu, 22 Feb 2007 12:26:58 +0000 Subject: [et-mgmt-tools] About CPU Usage of virt-manager In-Reply-To: <200702221544.JCG34350.N3G7E29K@aa.jp.fujitsu.com> References: <200702221544.JCG34350.N3G7E29K@aa.jp.fujitsu.com> Message-ID: <20070222122658.GA5893@redhat.com> On Thu, Feb 22, 2007 at 03:44:12PM +0900, Masayuki Sunou wrote: > Hi, > > I have a question about CPU Usage of virt-manager. > > Virt-manager calculates CPU Usage based on physical CPU. > Because of that, when the host has many CPU, guest's CPU Usage is difficult > to know. (SMP etc.) > > example: > When the host has 32 CPU, and guest's CPU Usage is 100%, virt-manager > displays CPU Usage as 3%. > --> 100%/32CPU=3% > > Therefore, I think that virt-manager should calculate CPU Usage based on > virtual CPU.(like "xentop") > > Because "Xentop" calculates CPU Usage based on virtual CPU, I think that > virt-manager should also calculate CPU Usage based on virtual CPU. The reason for calculating usage based on physical CPU is to illustrate what % of the host machine's resources are being used by a particular machine. eg it makes is very easy to see that 'Guest X' is taking up 50% of the host's compute cycles. Simiarly memory is calculated against physical RAM. While calculating CPU usage based on virtual CPUs would make it easier to see how much of its own allocation a guest is using, it would be harder to visualize overall host utilization. Perhaps we need to support both.. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From mdehaan at redhat.com Thu Feb 22 15:57:03 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 22 Feb 2007 10:57:03 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <20070222104151.GA23723@sliderule.msquared.com.au> References: <20070222104151.GA23723@sliderule.msquared.com.au> Message-ID: <45DDBD4F.4070709@redhat.com> Msquared wrote: > I'm having some trouble understanding where everything goes under cobbler. > > If I use 'cobbler import' or 'cobbler distro add', do I put the files > somewhere in /var/www/cobbler first? > Cobbler will do this for you. The purpose of cobbler distro add is to register the locations of the files you are interested provisioning. Cobbler will automagically copy them to the right locations needed to enable provisioning (which reside in /tftpboot and /var/www/cobbler, respectively). If you want to force cobbler to "re-do" this, just run "cobbler sync". If you are passing in an rsync mirror to "cobbler import", cobbler _does_ copy files for the rest of the tree (not just the kernel/initrd), and this is actually a pretty powerful way to start off. What happens then is you create a local mirror of whatever distribution, including the entire kickstart tree. This makes local provisioning very very fast. However, if you already have the operating system tree locally, that's not as important. You really don't want to be copying files by hand into the /var/www/cobbler directory, as cobbler will want to manage the contents of this directory itself. If you have a kickstart tree you just want to make available, (and you don't want to do the rsync:// import, which is really a great way to do this), it's easy enough to just copy that tree into your apache root /var/www/html or even cobbler's reserved-for-random-uses-of-people-who-know-what-they-are-doing directory /var/www/cobbler/local_mirror. In general though, I don't expect people to have to do that -- they either are using a http://, ftp://, or nfs:// kickstart tree someone else has already set up, or they'll be doing an import from a rync mirror to establish a new kickstart tree themselves that they can use for provisioning. > It seems that 'cobbler import' doesn't actually copy files for me, so I'm > not really sure what it does. > > > The walkthrough at http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler > doesn't seem to help me, as it doesn't tell me what I need to put where. > > I've tried to copy the contents of my Fedora 6 DVD to > /var/www/cobbler/kickstarts/Fedora6 and run 'cobbler import > --path=/var/www/cobbler/kickstarts/Fedora6', and then run 'cobbler list', > but the output looks nothing like the output presented on > http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler, instead it looks > like this: > First off, I do have to thank Dave a lot for making Cobbler posts on his site. Cobbler has changed just a little bit those original articles are written, and you should probably be looking at the latest manpage -- http://cobbler.et.redhat.com/cobbler.html The long-output command that generates what's on Dave's site is now "cobbler report", and "cobbler list" has been abbreviated to make it more usable. That's why the output looks different. > Distros: > var_www_cobbler_kickstarts_Fedora6_images_pxeboot > var_www_cobbler_kickstarts_Fedora6_images_xen > Repos: > Profiles: > var_www_cobbler_kickstarts_Fedora6_images_pxeboot > var_www_cobbler_kickstarts_Fedora6_images_xen > Systems: > This actually looks fine, as it shows cobbler has imported 2 distros and made a profile for each of them. That's good. > > I can't seem to get 'cobbler distro add' to work either, not that I'm > entirely sure what it does. Where does it get the repository files from? > Do I need to do some preparation of the files from the DVD, or download a > mirror first? > Cobbler has a tiered concepts of Distributions, Profiles, and Systems. Distributions contain kernel and initrd information. Profiles take Distributions and add kickstarts to them. Systems contain Profile information and potentially some system specific information. Whenever you run these commands, it updates the cobbler "database" in /var/lib/cobbler/* -- and will copy files needed around to locations in /tftpboot and /var/www/cobbler. So these commands are doing something :) If you want to download from a mirror, the best way to do that is to skip "distro add" altogether and just do: cobbler import --mirror=rsync://blah --mirror-name=blah as mentioned in the manpage. That sets you up with a local mirror of everything you need, with the kickstart files configured appropriately to point into that tree over http:// -- fully automatic installs that you can customize later as needed. More advanced use cases will require running distro add commands manually, but if you just want a small lab setup for PXE booting Fedora and a few other distros, doing an rsync:// import isn't a bad way to go and will pretty much fill out "reasonably sane" kickstarts for you automatically. > I tried to run cobbler distro add but ended up with a very strange error: > > cobbler distro add --name=myxendistro > --kernel=/var/www/cobbler/kickstarts/Fedora6/images/xen/vmlinuz > --initrd=/var/www/cobbler/kickstarts/Fedora6/images/xen/initrd.img > > The result is: > > Traceback (most recent call last): > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 505, in main > BootCLI(sys.argv).run() > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 102, in run > self.curry_args(self.args[1:], self.commands['toplevel']) > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 421, in curry_args > commands[args[0]](args[1:]) > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 469, in distro > return self.curry_args(args, self.commands['distro']) > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 421, in curry_args > commands[args[0]](args[1:]) > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 386, in distro_edit > return self.apply_args(args,commands,on_ok) > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 409, in apply_args > on_ok() > File "/usr/lib/python2.4/site-packages/cobbler/cobbler.py", line 385, in > on_ok = lambda: self.api.distros().add(distro, with_copy=self.api.sync_flag) > File "/usr/lib/python2.4/site-packages/cobbler/collection.py", line 97, in add > lite_sync.add_single_distro(ref.name) > File "/usr/lib/python2.4/site-packages/cobbler/action_litesync.py", line 60, in add_single_distro > self.sync.write_distro_file(distro) > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 733, in write_distro_file > clone.kernel_options = self.hash_to_string(clone.kernel_options) > File "/usr/lib/python2.4/site-packages/cobbler/action_sync.py", line 885, in hash_to_string > buffer = buffer + key + "=" + value + " " > TypeError: cannot concatenate 'str' and 'int' objects > Hmm, this worked fine for me but what I imagine is happening is that somewhere in the configuration files, some of the parameters are being stored as integers instead of strings. Not being able to reproduce this, but just looking at the code, this looks like a fix: diff -r fbdb0203b3ee41519942c705ec800333b70c3365 cobbler/action_sync.py --- a/cobbler/action_sync.py Wed Feb 21 10:30:33 2007 -0500 +++ b/cobbler/action_sync.py Thu Feb 22 10:44:21 2007 -0500 @@ -874,8 +874,8 @@ class BootSync: for key in hash: value = hash[key] if value is None: - buffer = buffer + key + " " + buffer = buffer + str(key) + " " else: - buffer = buffer + key + "=" + value + " " + buffer = buffer + str(key) + "=" + str(value) + " " return buffer The question is how did they get this way? I don't know, as unless the files in /var/lib/cobbler were edited directly, I would expect that file to contain strings as opposed to integers -- if you want to email me your /var/lib/cobbler/distros file, that would be useful in tracking down the problem. > > Any ideas? > > Regards, Msquared... > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From mdehaan at redhat.com Thu Feb 22 17:39:42 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 22 Feb 2007 12:39:42 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DDBD4F.4070709@redhat.com> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> Message-ID: <45DDD55E.6080404@redhat.com> > First off, I do have to thank Dave a lot for making Cobbler posts on > his site. Cobbler has changed just a little bit those original > articles are written, and you should probably be looking at the latest > manpage -- http://cobbler.et.redhat.com/cobbler.html > > The long-output command that generates what's on Dave's site is now > "cobbler report", and "cobbler list" has been abbreviated to make it > more usable. That's why the output looks different. >> Distros: >> var_www_cobbler_kickstarts_Fedora6_images_pxeboot >> var_www_cobbler_kickstarts_Fedora6_images_xen >> Repos: >> Profiles: >> var_www_cobbler_kickstarts_Fedora6_images_pxeboot >> var_www_cobbler_kickstarts_Fedora6_images_xen >> Systems: > This actually looks fine, as it shows cobbler has imported 2 distros > and made a profile for each of them. That's good. Let me retract that. That means import worked, though the directory you copied files to will likely be destroyed by cobbler, since that's a directory cobbler uses for it's own purposes. If you had importanted into another directory, like, say, /opt/kickstarts, you'd be a lot better off. Again, the rsync:// import works very well and is a good way to go. Just run "cobbler profile remove --name=x" for each of the profiles above, and then do the same for the distros, and you can start over. The import worked, that's just not a good place to put those files (it would be about as safe as putting them in /tmp). --Michael From berrange at redhat.com Thu Feb 22 22:12:46 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Thu, 22 Feb 2007 22:12:46 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.1 / virt-install 0.101.0 releases Message-ID: <20070222221246.GQ22473@redhat.com> Following on from the recent release of libvirt 0.2.0, I'm pleased to announce the releases of virt-manager 0.3.1, and virt-install 0.101.0 are now available for download from: http://virt-manager.org/download.html * Virtual Machine Manager 0.3.1 * This release introduces support for managing QEMU / KVM virtual machines using the new libvirt QEMU driver backend. This requires a new libvirt (at least 0.2.0) to enable the QEMU driver. It also requires an install of the virtinst package of at least version 0.101.0 to support QEMU. The dual cursor problem is worked around by grabbing the mouse pointer upon first button press (release with Ctrl+Alt). The progress bar display when creating new VMs has had its appearance tweaked. The new VM creation wizard also allows the user to specify the type of guest OS being installed. This will allow the setup of virtual hardware to be optimized for the needs of specific guest OS. http://virt-manager.org/download/sources/virt-manager/virt-manager-0.3.1.tar.gz * Virt Install 0.101.0 * This release introduces the first support for provisioning of QEMU / KVM virtual machines. This requires an updated libvirt of at least 0.2.0 to enable the qemu driver. The code for fetching bootable kernel/initrd images has been refactored to facilitate porting to deal with other non Fedora / RHEL operating systems. Prototype support for dealing with SuSE paravirtualized installs is provided. The bootable ISO image for fully virtualized installs can also be fetched at runtime from an NFS/HTTP/FTP site, or equivalent initrd and kernels for baremetal. A list of guest operating systems is now maintained to enable virtual hardware configuration to be optimized to suit a particular guest OS. http://virt-manager.org/download/sources/virtinst/virtinst-0.101.0.tar.gz NB, the QEMU / KVM support is at a very early stage of development - we have released this snapshot just to get it is exposed to the real world. We fully expect people to find a wide variety of bugs - the next release of libvirt & virt-manager will apply far more polish to the QEMU support to get it into a generally usable state. NB, to use QEMU support in virt-install, you must supply an explicit connect URI for the hypervisor. eg virt-install --connect qemu:///session Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From fukuta.saori at jp.fujitsu.com Fri Feb 23 01:56:04 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Fri, 23 Feb 2007 10:56:04 +0900 Subject: [et-mgmt-tools] [PATCH] waiting VNC port number In-Reply-To: <20070221141112.GE7306@redhat.com> References: <20070221120701.669D.FUKUTA.SAORI@jp.fujitsu.com> <20070221141112.GE7306@redhat.com> Message-ID: <20070223105543.DAE7.FUKUTA.SAORI@jp.fujitsu.com> Hi, Dan Thank you for your comments. On Wed, 21 Feb 2007 14:11:12 +0000 "Daniel P. Berrange"wrote: > This is no good because it is Xen specific - the code breaks when used > with the QEMU / KVM backend to libvirt. I think this is actually a > combination of two bugs: > > - libvirt looks in xenstore to find the port number, but if it is > missing hardcodes 5900 + domid. So there is a race condition where > if virt-install asks for the XML dump, before the framebuffer daemon > has started up, it will get the wrong port number. > > This is a hardcoded fallback port number is a hangover from old Xen > 3.0.2 where the port number was fixed & not kept in xenstore at all. > We need to fix libvirt to only use this code on Xen <= 3.0.3 Yes, I think so, too. I try to think about how to fix this bug. > - Once the above fix in libvirt is done, we can make virt-install do > a loop like the one you have below, but instead of looking in xenstore > it will look at the port in the XML - waiting for it to change from '-1' > to an actual port number You are right, I will consider again as you pointed out. Thanks, Saori Fukuta From fukuta.saori at jp.fujitsu.com Fri Feb 23 02:36:38 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Fri, 23 Feb 2007 11:36:38 +0900 Subject: [et-mgmt-tools] [PATCH] domain name must not be numeric only In-Reply-To: <20070221140519.GC7306@redhat.com> References: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> <20070221140519.GC7306@redhat.com> Message-ID: <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> Hi, Dan On Wed, 21 Feb 2007 14:05:19 +0000 "Daniel P. Berrange" wrote: > On Wed, Feb 21, 2007 at 12:09:02PM +0900, Saori Fukuta wrote: > > Hi, > > > > The attached patch adds to check the domain name that must not be > > numeric only. > > > > Signed-off-by: Saori Fukuta > > Looks good, will apply. By the way, I find a slight mistake here. How about this for a message? Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virtinst/Guest.py (python-virtinst-0.100.0) =================================================================== --- Guest.py 2007-02-21 11:04:43.000000000 +0900 +++ Guest.py.msg 2007-02-23 11:13:07.000000000 +0900 @@ -207,7 +207,7 @@ class Guest(object): if re.match("^[a-zA-Z0-9_]*$", val) == None: raise ValueError, "Domain name must be alphanumeric or _" if len(val) > 50: - raise ValueError, "Domain name must be less than 50 characters" + raise ValueError, "Domain name must be less or equal 50 characters" if type(val) != type("string"): raise ValueError, "Domain name must be a string" self._name = val From fj0873gn at aa.jp.fujitsu.com Fri Feb 23 06:03:34 2007 From: fj0873gn at aa.jp.fujitsu.com (Nobuhiro Itou) Date: Fri, 23 Feb 2007 15:03:34 +0900 Subject: [et-mgmt-tools] [PATCH] check the VNC port number Message-ID: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> Hi, When I execute "virt-install --vnc --vncport=5900", domain starts with 5900 + domain ID. I think that it is necessary to check, because we cannot appoint the VNC port number equal to or less than 5900. The attached patch adds to check the VNC port number. Signed-off-by: Nobuhiro Itou Thanks, Nobuhiro Itou. Index: virtinst/Guest.py (python-virtinst-0.100.0) =================================================================== --- Guest.py 2007-02-23 02:28:02.000000000 +0900 +++ Guest.py.vncport_check 2007-02-23 02:24:07.000000000 +0900 @@ -157,7 +157,9 @@ class VirtualGraphics: class VNCVirtualGraphics(VirtualGraphics): def __init__(self, *args): self.name = "vnc" - if len(args) >= 1 and args[0]: + if len(args) >= 1 and not args[0] is None: + if args[0] < 5901: + raise ValueError, "Invalid value for vncport, vncport must be no less than 5901" self.port = args[0] else: self.port = -1 From cobbler at msquared.id.au Fri Feb 23 07:05:24 2007 From: cobbler at msquared.id.au (Msquared) Date: Fri, 23 Feb 2007 16:05:24 +0900 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DDD55E.6080404@redhat.com> <45DDBD4F.4070709@redhat.com> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DDD55E.6080404@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> Message-ID: <20070223070524.GA3332@sliderule.msquared.com.au> Thanks for the patch, that fixed cobbler for me, which allowed me to retry some of the things that had failed for me before. On Thu, Feb 22, 2007 at 10:57:03AM -0500, Michael DeHaan wrote: > >If I use 'cobbler import' or 'cobbler distro add', do I put the files > >somewhere in /var/www/cobbler first? > > Cobbler will do this for you. I think I'm confused about what belongs where. I thought the import command would import the contents of the repository (in this case, my CD) into somewhere that cobbler managed. It didn't seem to do that, so I put the files somewhere that I thought should be accessible to the machine being provisioned (ie: inside Cobbler's web space). The first thing I noticed was that cobbler import would not work when I gave it the mounted DVD: cobbler wanted the area to be writeable. So if I put the contents of my Fedora 6 DVD into /opt/dist/fc6 and then run cobbler import --path=/opt/dist/fc6 cobbler sync Do I have to leave my Fedora 6 files in /opt/dist/fc6? > If you are passing in an rsync mirror to "cobbler import", cobbler > _does_ copy files for the rest of the tree (not just the kernel/initrd), > and this is actually a pretty powerful way to start off. But I already have a copy of the files on a DVD, and so I just want to have cobbler import that instead. Firstly, I don't even know what rsync URL I should provide, and secondly I'm sure it would be faster from the DVD than across my internet connection. :-) > However, if you already have the operating system tree locally, that's > not as important. I still want cobbler to copy the contents of the DVD into somewhere that it manages, though. I managed to do it like so: cobbler import --mirror=root at localhost:/media/cdrom --mirror-name=testimport However, will this cause me grief in future? What happens when I 'cobbler sync'? Will it try to re-mirror based on whatever is in /media/cdrom? Can I stop it from doing that? Also, should I make the mirror-name something that is related to where I'm getting the files from (eg: fc6dvd or fc6public), related to what the files are for (eg: fc6), or related to my provisioning server (eg: localprovserver; ie: the machine I am running cobbler on)? Of course, my next question would be how do I update my local copy of the DVD from updates provided to the distribution(s). Can I change the URL it uses from root at localhost:/media/cdrom to one of the official Fedora 6 mirrors? Ah, distro and profile names. :) When I ran my cobbler import with the ssh mirror, I ended up with a distro and a profile named thus: var_www_cobbler_ks_mirror_testimport_cdrom_images_xen Is there a way of making this shorter, or more meaningful (without editing config files)? Also, if I use 'cobbler distro add', how does it know where to get the rest of the files from? (ie: how does it know where the repository is?) I've tried to import from the DVD with these, but none of them work: cobbler import --mirror=root at localhost > >The walkthrough at > >http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler doesn't seem to > >help me, as it doesn't tell me what I need to put where. > > First off, I do have to thank Dave a lot for making Cobbler posts on his > site. It's definitely a step in the right direction. It encouraged me to take the plunge and experiment anyway. However, I would like to see a lot more higher-level documentation, plus documentation of the level Dave provided, but with more explanations about what values you should use, or at least sensible suggestions. I'm willing to help write that documentation, if that helps. I think tools like cobbler/koan could make a huge difference in the uptake of Xen, and while Xen works wonderfully when running, I've found that setting it up is the hardest part. > >I can't seem to get 'cobbler distro add' to work either, not that I'm > >entirely sure what it does. Where does it get the repository files > >from? Do I need to do some preparation of the files from the DVD, or > >download a mirror first? > > Cobbler has a tiered concepts of Distributions, Profiles, and Systems. > Distributions contain kernel and initrd information. Profiles take > Distributions and add kickstarts to them. Systems contain Profile > information and potentially some system specific information. Whenever > you run these commands, it updates the cobbler "database" in > /var/lib/cobbler/* -- and will copy files needed around to locations in > /tftpboot and /var/www/cobbler. So these commands are doing something :) Ah, I guess I assumed that the cobbler 'database' included the actual repositories of files, too. Although most of the documentation discusses Distributions, Profiles, and Systems (as you mention above), the output of 'cobbler list' includes 'Repos'. I guess this is one part that I'm not sure how it all works together. I'll go and review the updated documentation you listed. > If you want to download from a mirror, the best way to do that is to > skip "distro add" altogether and just do: > > cobbler import --mirror=rsync://blah --mirror-name=blah > > as mentioned in the manpage. I couldn't just mirror 3-4G without explaining it to someone. :-) Since I had the files on CD, I've been trying to work out how to use them instead. I should have explicitly asked "how to I get cobbler to mirror from my CD", so that you could have answered with an SSH URL example, rather than trying to convince me to use an RSYNC URL, which I didn't want to have to set up. :-) On Thu, Feb 22, 2007 at 12:39:42PM -0500, Michael DeHaan wrote: > Let me retract that. That means import worked, though the directory you > copied files to will likely be destroyed by cobbler, since that's a > directory cobbler uses for it's own purposes. If you had importanted > into another directory, like, say, /opt/kickstarts, you'd be a lot > better off. Again, the rsync:// import works very well and is a good way > to go. Thanks for the tip. I've stopped poking around in /var/www/cobbler. :-) Regards, Msquared... From mark.cave-ayland at ilande.co.uk Fri Feb 23 09:10:59 2007 From: mark.cave-ayland at ilande.co.uk (Mark Cave-Ayland) Date: Fri, 23 Feb 2007 09:10:59 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <20070215115806.GA16485@redhat.com> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> <1171530362.7896.8.camel@mca-desktop> <20070215115806.GA16485@redhat.com> Message-ID: <1172221859.21229.21.camel@mca-desktop> On Thu, 2007-02-15 at 11:58 +0000, Daniel P. Berrange wrote: > > Another quick couple of questions: firstly is there a way to bypass > > "Open Connection" dialog on startup so that it assumes that the default > > "Local Xen Host" is selected? I can see the -C option for a URI but > > can't see a way to specify the equivalent of "Local Xen Host" on the > > command line? > > virt-manager --connect Xen > > Will do the trick IIRC. Actually, I've just found out this doesn't work. I've finally got around to creating a menu entry in Ubuntu which reads: "gksu virt-manager --connect Xen" and when selected, it does bypass the selection dialog and go straight to the list of virtual machines. However, when I try and start my Windows XP virtual machine, nothing happens! I eventually tried running "virt-manager --connect Xen" directly and got the following message on the console: "libvir: Domain error winxp: operation virDomainCreate forbidden for read only access". However, if I I just execute "virt-manager" and select "Local host" and hit "Connect", then I can start the VMs fine. The only conclusion I can come to is that if you specify "--connect Xen" on the command line for virt-manager that it behaves the same as if you tick the "Read only" checkbox in the "Open Connection" dialog. Am I right in thinking this is a bug? FWIW, I'm using the latest versions of everything: libvirt 0.2.0, virtinst-0.101.0 and virt-manager 0.3.1. Many thanks, Mark. From fj1826dm at aa.jp.fujitsu.com Fri Feb 23 09:55:10 2007 From: fj1826dm at aa.jp.fujitsu.com (Masayuki Sunou) Date: Fri, 23 Feb 2007 18:55:10 +0900 Subject: [et-mgmt-tools] About CPU Usage of virt-manager In-Reply-To: <20070222122658.GA5893@redhat.com> References: <200702221544.JCG34350.N3G7E29K@aa.jp.fujitsu.com> <20070222122658.GA5893@redhat.com> Message-ID: <200702231855.FBF78620.2N9EKG37@aa.jp.fujitsu.com> Hi, Dan > Perhaps we need to support both.. > Well, I think so. I consider this correction again because I think that priority is low. Thanks, Masayuki Sunou. -------------------------------------------------------------------- > On Thu, Feb 22, 2007 at 03:44:12PM +0900, Masayuki Sunou wrote: > > Hi, > > > > I have a question about CPU Usage of virt-manager. > > > > Virt-manager calculates CPU Usage based on physical CPU. > > Because of that, when the host has many CPU, guest's CPU Usage is difficult > > to know. (SMP etc.) > > > > example: > > When the host has 32 CPU, and guest's CPU Usage is 100%, virt-manager > > displays CPU Usage as 3%. > > --> 100%/32CPU=3% > > > > Therefore, I think that virt-manager should calculate CPU Usage based on > > virtual CPU.(like "xentop") > > > > Because "Xentop" calculates CPU Usage based on virtual CPU, I think that > > virt-manager should also calculate CPU Usage based on virtual CPU. > > The reason for calculating usage based on physical CPU is to illustrate > what % of the host machine's resources are being used by a particular > machine. eg it makes is very easy to see that 'Guest X' is taking up 50% > of the host's compute cycles. Simiarly memory is calculated against > physical RAM. While calculating CPU usage based on virtual CPUs would > make it easier to see how much of its own allocation a guest is using, it > would be harder to visualize overall host utilization. > > Perhaps we need to support both.. > > Dan. > -- > |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| > |=- Perl modules: http://search.cpan.org/~danberr/ -=| > |=- Projects: http://freshmeat.net/~danielpb/ -=| > |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From berrange at redhat.com Fri Feb 23 12:00:05 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Fri, 23 Feb 2007 12:00:05 +0000 Subject: [et-mgmt-tools] [PATCH] waiting VNC port number In-Reply-To: <20070223105543.DAE7.FUKUTA.SAORI@jp.fujitsu.com> References: <20070221120701.669D.FUKUTA.SAORI@jp.fujitsu.com> <20070221141112.GE7306@redhat.com> <20070223105543.DAE7.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070223120004.GA14993@redhat.com> On Fri, Feb 23, 2007 at 10:56:04AM +0900, Saori Fukuta wrote: > On Wed, 21 Feb 2007 14:11:12 +0000 "Daniel P. Berrange"wrote: > > This is no good because it is Xen specific - the code breaks when used > > with the QEMU / KVM backend to libvirt. I think this is actually a > > combination of two bugs: > > > > - libvirt looks in xenstore to find the port number, but if it is > > missing hardcodes 5900 + domid. So there is a race condition where > > if virt-install asks for the XML dump, before the framebuffer daemon > > has started up, it will get the wrong port number. > > > > This is a hardcoded fallback port number is a hangover from old Xen > > 3.0.2 where the port number was fixed & not kept in xenstore at all. > > We need to fix libvirt to only use this code on Xen <= 3.0.3 > > Yes, I think so, too. > I try to think about how to fix this bug. Check out the libvir-list archives - I posted a patch for this last night. > > - Once the above fix in libvirt is done, we can make virt-install do > > a loop like the one you have below, but instead of looking in xenstore > > it will look at the port in the XML - waiting for it to change from '-1' > > to an actual port number > > You are right, I will consider again as you pointed out. Excellant. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Fri Feb 23 12:01:54 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Fri, 23 Feb 2007 12:01:54 +0000 Subject: [et-mgmt-tools] [PATCH] domain name must not be numeric only In-Reply-To: <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> References: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> <20070221140519.GC7306@redhat.com> <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070223120154.GB14993@redhat.com> On Fri, Feb 23, 2007 at 11:36:38AM +0900, Saori Fukuta wrote: > Hi, Dan > > On Wed, 21 Feb 2007 14:05:19 +0000 "Daniel P. Berrange" wrote: > > On Wed, Feb 21, 2007 at 12:09:02PM +0900, Saori Fukuta wrote: > > > Hi, > > > > > > The attached patch adds to check the domain name that must not be > > > numeric only. > > > > > > Signed-off-by: Saori Fukuta > > > > Looks good, will apply. > > By the way, I find a slight mistake here. > How about this for a message? Looks good, have applied it. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Fri Feb 23 12:16:22 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Fri, 23 Feb 2007 12:16:22 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <1172221859.21229.21.camel@mca-desktop> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> <1171530362.7896.8.camel@mca-desktop> <20070215115806.GA16485@redhat.com> <1172221859.21229.21.camel@mca-desktop> Message-ID: <20070223121622.GD14993@redhat.com> On Fri, Feb 23, 2007 at 09:10:59AM +0000, Mark Cave-Ayland wrote: > On Thu, 2007-02-15 at 11:58 +0000, Daniel P. Berrange wrote: > > > > Another quick couple of questions: firstly is there a way to bypass > > > "Open Connection" dialog on startup so that it assumes that the default > > > "Local Xen Host" is selected? I can see the -C option for a URI but > > > can't see a way to specify the equivalent of "Local Xen Host" on the > > > command line? > > > > virt-manager --connect Xen > > > > Will do the trick IIRC. > > > Actually, I've just found out this doesn't work. I've finally got around > to creating a menu entry in Ubuntu which reads: "gksu virt-manager > --connect Xen" and when selected, it does bypass the selection dialog > and go straight to the list of virtual machines. However, when I try and > start my Windows XP virtual machine, nothing happens! > > I eventually tried running "virt-manager --connect Xen" directly and got > the following message on the console: "libvir: Domain error winxp: > operation virDomainCreate forbidden for read only access". However, if I > I just execute "virt-manager" and select "Local host" and hit "Connect", > then I can start the VMs fine. > > The only conclusion I can come to is that if you specify "--connect Xen" > on the command line for virt-manager that it behaves the same as if you > tick the "Read only" checkbox in the "Open Connection" dialog. Am I > right in thinking this is a bug? Looking at the code, that is indeed true. I'll see about fixing that. In fact I think I'll probably remove the 'read only' checkbox from the dialog too, and have virt-manager automatically try a full read write connection, falling back to read only when doing Xen as non-root. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From dhawal at netmagicsolutions.com Fri Feb 23 14:05:36 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Fri, 23 Feb 2007 19:35:36 +0530 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux In-Reply-To: <45DA2D02.1000305@redhat.com> References: <45DA2D02.1000305@redhat.com> Message-ID: <45DEF4B0.7040402@netmagicsolutions.com> Michael DeHaan wrote: > I've noticed that the pxelinux that ships with RHEL4 (test platform: > RHEL4U4) does not work well at all with new PXE menus, though the > version shipping with FC6 works great. Basically it doesn't understand > the menu label options. > > If you have RHEL4's version of syslinux installed, the fix here is to > install a newer syslinux. I'll file the bug for good measure (if this > is even a bug?) and may bundle a known-good pxelinux.0 in the next > release. The workaround is pretty simple though. If anyone still > has questions, feel free to ask. Finally got the time to try out the newer 0.4.2 version.. the default menu gets created nicely and i am trying to use a local kickstart file rather than a URL (to explore templating options).. However, while trying to pxeboot a test server, it just stays at the menu (tab works fine though).. looks like i need to upgrade syslinux OR pxelinux.0 as you've mentioned. Now since syslinux needs an upgrade (the latest being 3.36) do you have any recommendations for the version? - dhawal From jim at meyering.net Fri Feb 23 13:47:37 2007 From: jim at meyering.net (Jim Meyering) Date: Fri, 23 Feb 2007 14:47:37 +0100 Subject: [et-mgmt-tools] [PATCH]: give a sensible diagnostic for an empty name In-Reply-To: <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> (Saori Fukuta's message of "Fri, 23 Feb 2007 11:36:38 +0900") References: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> <20070221140519.GC7306@redhat.com> <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <877iu956km.fsf_-_@rho.meyering.net> Hi Dan, I noticed that get_name doesn't diagnose an empty name properly. Instead, it says it must not be numeric only. Give a sensible diagnostic for an empty name. Even though the code now detects an empty name early, don't let subsequent tests for digit-only and non-alphanumeric names match on an empty name. diff -r 982e9920511c -r c97207424b57 virtinst/Guest.py --- a/virtinst/Guest.py Wed Feb 21 11:42:53 2007 -0500 +++ b/virtinst/Guest.py Fri Feb 23 14:42:50 2007 +0100 @@ -243,9 +243,11 @@ class Guest(object): def get_name(self): return self._name def set_name(self, val): - if re.match("^[0-9]*$", val): + if len(val) == 0: + raise ValueError, "Domain name must be nonempty" + if re.match("^[0-9]+$", val): raise ValueError, "Domain name must not be numeric only" - if re.match("^[a-zA-Z0-9_]*$", val) == None: + if re.match("^[a-zA-Z0-9_]+$", val) == None: raise ValueError, "Domain name must be alphanumeric or _" if len(val) > 50: raise ValueError, "Domain name must be less than 50 characters" From berrange at redhat.com Fri Feb 23 14:43:12 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Fri, 23 Feb 2007 14:43:12 +0000 Subject: [et-mgmt-tools] Re: [PATCH]: give a sensible diagnostic for an empty name In-Reply-To: <877iu956km.fsf_-_@rho.meyering.net> References: <20070221120858.66AF.FUKUTA.SAORI@jp.fujitsu.com> <20070221140519.GC7306@redhat.com> <20070223113548.DB05.FUKUTA.SAORI@jp.fujitsu.com> <877iu956km.fsf_-_@rho.meyering.net> Message-ID: <20070223144312.GA23711@redhat.com> On Fri, Feb 23, 2007 at 02:47:37PM +0100, Jim Meyering wrote: > Hi Dan, > > I noticed that get_name doesn't diagnose an empty name properly. > Instead, it says it must not be numeric only. > > Give a sensible diagnostic for an empty name. > Even though the code now detects an empty name early, don't let > subsequent tests for digit-only and non-alphanumeric names match > on an empty name. Applied - thanks, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From mdehaan at redhat.com Fri Feb 23 14:49:25 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 09:49:25 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <20070223070524.GA3332@sliderule.msquared.com.au> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DDD55E.6080404@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <20070223070524.GA3332@sliderule.msquared.com.au> Message-ID: <45DEFEF5.2020802@redhat.com> Msquared wrote: > Thanks for the patch, that fixed cobbler for me, which allowed me to retry > some of the things that had failed for me before. > Good deal, I will include it in the next release after I look over the files you sent me (thanks!). > > On Thu, Feb 22, 2007 at 10:57:03AM -0500, Michael DeHaan wrote: > > >>> If I use 'cobbler import' or 'cobbler distro add', do I put the files >>> somewhere in /var/www/cobbler first? >>> >> Cobbler will do this for you. >> > > I think I'm confused about what belongs where. > > I thought the import command would import the contents of the repository > (in this case, my CD) into somewhere that cobbler managed. It didn't seem > to do that, so I put the files somewhere that I thought should be > accessible to the machine being provisioned (ie: inside Cobbler's web > space). > > The first thing I noticed was that cobbler import would not work when I > gave it the mounted DVD: cobbler wanted the area to be writeable. > > So if I put the contents of my Fedora 6 DVD into /opt/dist/fc6 and then run > > cobbler import --path=/opt/dist/fc6 > cobbler sync > > Do I have to leave my Fedora 6 files in /opt/dist/fc6? > > Ah, you're trying to do DVD imports. Well, it depends on what you want to do. Probably you're going to want to use the entire kickstart tree from that DVD, so that's as good as a place to leave them as any. What I normally do in this situation is add a symlink, like /var/www/cobbler/local_mirror/fc6tree -> /opt/dist/fc6 And then, you can use the following parameter in your kickstart files for the "--url" url --url http://servername/cobbler_track/local_mirror/fc6tree/restofpath Note that you want to include the full path to the "os" or "tree" directory, and don't want to stop at the root. That's what I mean by "restofpath". >> If you are passing in an rsync mirror to "cobbler import", cobbler >> _does_ copy files for the rest of the tree (not just the kernel/initrd), >> and this is actually a pretty powerful way to start off. >> > > But I already have a copy of the files on a DVD, and so I just want to > have cobbler import that instead. Firstly, I don't even know what rsync > URL I should provide, and secondly I'm sure it would be faster from the > DVD than across my internet connection. :-) > You're right, it's not faster :) It just puts files in the right place, and allows them to update if they ever change remotely. The DVD case can still work just fine. I think you've given an excellent reason to add "how to import a kickstart tree from a DVD" to the manpage. > >> However, if you already have the operating system tree locally, that's >> not as important. >> > > I still want cobbler to copy the contents of the DVD into somewhere that > it manages, though. I managed to do it like so: > > cobbler import --mirror=root at localhost:/media/cdrom --mirror-name=testimport > > Actually that will work. Creative workaround! :) This will file the results of the import in /var/www/cobbler/ks_mirror and is a really smart way to go about it. > However, will this cause me grief in future? What happens when I 'cobbler > sync'? Will it try to re-mirror based on whatever is in /media/cdrom? > Can I stop it from doing that? > Cobbler sync won't try to re-mirror that, because cobbler knows kickstart trees really don't change. Sync will build out a lot of other interestingness in /var/lib/cobbler and /tftpboot, but it will leave the contents of /var/lib/cobbler/ks_mirror untouched. > Also, should I make the mirror-name something that is related to where I'm > getting the files from (eg: fc6dvd or fc6public), related to what the > files are for (eg: fc6), or related to my provisioning server (eg: > localprovserver; ie: the machine I am running cobbler on)? > The mirror name is basically just to pick something you can remember later. > Of course, my next question would be how do I update my local copy of the > DVD from updates provided to the distribution(s). Can I change the URL it > uses from root at localhost:/media/cdrom to one of the official Fedora 6 > mirrors? > > Nah, not quite. Look at the manpage for "cobbler repo add" for adding a repository. You'll add the updates repository this way and then reference them in the profile. Since you've actually imported the profile, you're going to need to go into /var/lib/cobbler/profiles to list the repos you use. This isn't really as scary as it sounds, but if you're concerned, you can back up the profiles file first. > Ah, distro and profile names. :) > > When I ran my cobbler import with the ssh mirror, I ended up with a distro > and a profile named thus: > > var_www_cobbler_ks_mirror_testimport_cdrom_images_xen > > Is there a way of making this shorter, or more meaningful (without editing > config files)? > > Import uses the paths because they are safely namespaced. I probably should consider making them auto-shorten themselves in a future release. In the meantime, it's pretty safe to edit /var/lib/cobbler/profiles and change the "name:" fields. You can do the same for the distros if you like, just make sure the "distro" fields in the profile match the distro names. Really though, you could just edit the profiles and leave it at that. > Also, if I use 'cobbler distro add', how does it know where to get the > rest of the files from? (ie: how does it know where the repository is?) > > I've tried to import from the DVD with these, but none of them work: > > cobbler import --mirror=root at localhost > > I thought you said you just got that working earlier? Anyhow, import also requires a "--mirror-name". Distro add is a way to manually add distros without using the "import" shortcut, and requires full paths to the kernel and initrd files. >>> The walkthrough at >>> http://wiki.xdroop.com/space/RedHat/kickstart/Cobbler doesn't seem to >>> help me, as it doesn't tell me what I need to put where. >>> >> First off, I do have to thank Dave a lot for making Cobbler posts on his >> site. >> > > It's definitely a step in the right direction. It encouraged me to take > the plunge and experiment anyway. However, I would like to see a lot more > higher-level documentation, plus documentation of the level Dave provided, > but with more explanations about what values you should use, or at least > sensible suggestions. > +1. The easiest way for this to happen is for you to write up something, put it up on the web, and I can link it from the Cobbler webpage. All of that material can be potentially merged to generate a bit of a user manual later, and that's an excellent idea. As with the DVD import, folks using cobbler often think of things I don't think of, and it's great to see what the things they want to do with provisioning are. > I'm willing to help write that documentation, if that helps. I think > tools like cobbler/koan could make a huge difference in the uptake of Xen, > and while Xen works wonderfully when running, I've found that setting it > up is the hardest part. > Great! > >>> I can't seem to get 'cobbler distro add' to work either, not that I'm >>> entirely sure what it does. Where does it get the repository files >>> from? Do I need to do some preparation of the files from the DVD, or >>> download a mirror first? >>> >> >> Cobbler has a tiered concepts of Distributions, Profiles, and Systems. >> Distributions contain kernel and initrd information. Profiles take >> Distributions and add kickstarts to them. Systems contain Profile >> information and potentially some system specific information. Whenever >> you run these commands, it updates the cobbler "database" in >> /var/lib/cobbler/* -- and will copy files needed around to locations in >> /tftpboot and /var/www/cobbler. So these commands are doing something :) >> > > Ah, I guess I assumed that the cobbler 'database' included the actual > repositories of files, too. Although most of the documentation discusses > Distributions, Profiles, and Systems (as you mention above), the output of > 'cobbler list' includes 'Repos'. I guess this is one part that I'm not > sure how it all works together. I'll go and review the updated > documentation you listed. > Yeah, repos are a relatively new concept, and are used for attaching things like "fc6updatesi386" to a profile, so that they can automatically install and configure a yum mirror when they provision it, as well as using a yum mirror at install time to install packages. For instance, an ISV software repository or a college with special software in a yum repository that was not included in Fedora Extras. > >> If you want to download from a mirror, the best way to do that is to >> skip "distro add" altogether and just do: >> >> cobbler import --mirror=rsync://blah --mirror-name=blah >> >> as mentioned in the manpage. >> > > I couldn't just mirror 3-4G without explaining it to someone. :-) Since > I had the files on CD, I've been trying to work out how to use them > instead. > Yeah, usage of the ssh parameters from the DVD will make a mirrored copy of what's on the CD, but that's better than leaving it mounted. Good work there. > I should have explicitly asked "how to I get cobbler to mirror from my > CD", so that you could have answered with an SSH URL example, rather than > trying to convince me to use an RSYNC URL, which I didn't want to have to > set up. :-) > Yeah. Given that the way to make rsync work locally seems to work, I'll include a section on that in the documentation. I like that a lot. > > On Thu, Feb 22, 2007 at 12:39:42PM -0500, Michael DeHaan wrote: > > >> Let me retract that. That means import worked, though the directory you >> copied files to will likely be destroyed by cobbler, since that's a >> directory cobbler uses for it's own purposes. If you had importanted >> into another directory, like, say, /opt/kickstarts, you'd be a lot >> better off. Again, the rsync:// import works very well and is a good way >> to go. >> > > Thanks for the tip. I've stopped poking around in /var/www/cobbler. :-) > > Regards, Msquared... > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From mdehaan at redhat.com Fri Feb 23 14:52:13 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 09:52:13 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DEFEF5.2020802@redhat.com> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DDD55E.6080404@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <20070223070524.GA3332@sliderule.msquared.com.au> <45DEFEF5.2020802@redhat.com> Message-ID: <45DEFF9D.6060400@redhat.com> >> > Ah, you're trying to do DVD imports. Well, it depends on what you > want to do. Probably you're going to want to use the entire kickstart > tree from that DVD, so that's as good as a place to leave them as any. > > What I normally do in this situation is add a symlink, like > > /var/www/cobbler/local_mirror/fc6tree -> /opt/dist/fc6 Correction: the way you invoked the import (using the root:localhost@ syntax), you don't need to do this at all. Check to make sure content did get copied to /var/www/cobbler/ks_mirror though ... From dhawal at netmagicsolutions.com Fri Feb 23 15:02:14 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Fri, 23 Feb 2007 20:32:14 +0530 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DDBD4F.4070709@redhat.com> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> Message-ID: <45DF01F6.1050507@netmagicsolutions.com> Michael DeHaan wrote: > Hmm, this worked fine for me but what I imagine is happening is that > somewhere in the configuration files, some of the parameters are being > stored as integers instead of strings. Not being able to reproduce this, > but just looking at the code, this looks like a fix: > > diff -r fbdb0203b3ee41519942c705ec800333b70c3365 cobbler/action_sync.py > --- a/cobbler/action_sync.py Wed Feb 21 10:30:33 2007 -0500 > +++ b/cobbler/action_sync.py Thu Feb 22 10:44:21 2007 -0500 > @@ -874,8 +874,8 @@ class BootSync: > for key in hash: > value = hash[key] > if value is None: > - buffer = buffer + key + " " > + buffer = buffer + str(key) + " " > else: > - buffer = buffer + key + "=" + value + " " > + buffer = buffer + str(key) + "=" + str(value) + " " > return buffer Just to add, i experienced a similar problem on 0.4.2 and this patch fixed it for me as well. - dhawal From dhawal at netmagicsolutions.com Fri Feb 23 15:08:14 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Fri, 23 Feb 2007 20:38:14 +0530 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux In-Reply-To: <45DEF4B0.7040402@netmagicsolutions.com> References: <45DA2D02.1000305@redhat.com> <45DEF4B0.7040402@netmagicsolutions.com> Message-ID: <45DF035E.2060600@netmagicsolutions.com> Dhawal Doshy wrote: > Michael DeHaan wrote: >> I've noticed that the pxelinux that ships with RHEL4 (test platform: >> RHEL4U4) does not work well at all with new PXE menus, though the >> version shipping with FC6 works great. Basically it doesn't >> understand the menu label options. >> >> If you have RHEL4's version of syslinux installed, the fix here is to >> install a newer syslinux. I'll file the bug for good measure (if >> this is even a bug?) and may bundle a known-good pxelinux.0 in the >> next release. The workaround is pretty simple though. If anyone >> still has questions, feel free to ask. > > Finally got the time to try out the newer 0.4.2 version.. the default > menu gets created nicely and i am trying to use a local kickstart file > rather than a URL (to explore templating options).. > > However, while trying to pxeboot a test server, it just stays at the > menu (tab works fine though).. looks like i need to upgrade syslinux OR > pxelinux.0 as you've mentioned. Now since syslinux needs an upgrade (the > latest being 3.36) do you have any recommendations for the version? Upgraded to syslinux 3.36 and all is well.. though as you mention a simple change to a newer pxelinux.0 would have worked equally well. wget http://www.kernel.org/pub/linux/utils/boot/syslinux/syslinux-3.36.tar.gz yum install netpbm-progs nasm rpmbuild -ta syslinux-3.36.tar.gz rpm -Fvh /usr/src/redhat/RPMS/i386/syslinux-3.36-1.i386.rpm # the other RPMs are not required. Change in /var/lib/cobbler/settings this line: standard: /usr/lib/syslinux/pxelinux.0 to this: standard: /usr/share/syslinux/pxelinux.0 A 'cobbler sync' makes the necessary changes.. Finally can anyone post some more examples of the ks templating?? cheers, - dhawal From mdehaan at redhat.com Fri Feb 23 15:48:29 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 10:48:29 -0500 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux In-Reply-To: <45DF035E.2060600@netmagicsolutions.com> References: <45DA2D02.1000305@redhat.com> <45DEF4B0.7040402@netmagicsolutions.com> <45DF035E.2060600@netmagicsolutions.com> Message-ID: <45DF0CCD.9060404@redhat.com> > > Finally can anyone post some more examples of the ks templating?? > I assume you've seen the very basic examples in /etc/cobbler, these are the default files that cobbler uses when one runs "cobbler import". The simplest thing there is how it fills in $tree to point to the location of your kickstart tree as accessible from cobbler over http://. If you do a cobbler import (say, from an rsync mirror, or from a DVD as msquared has shown us how to do) and then run a "cobbler report", you'll see each profile uses a custom value for "tree", but in the kickstart, it's just shown as "$tree". One thing that would be useful (I may have mentioned this before) is if one was provisioning a set of machines that were to use static IP's.... the profile would be common to all of them, but the different ip's would be added with --ksmeta="ip=192.168.1.50" for the various system entries. This would allow their IP's to be set up appropriately in Anaconda. It would also be possible for two profiles to share a common kickstart source file, and template that differently. > cheers, > - dhawal > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools From mdehaan at redhat.com Fri Feb 23 15:54:48 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 10:54:48 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DF01F6.1050507@netmagicsolutions.com> References: <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DF01F6.1050507@netmagicsolutions.com> Message-ID: <45DF0E48.3060107@redhat.com> Dhawal Doshy wrote: > Michael DeHaan wrote: >> Hmm, this worked fine for me but what I imagine is happening is that >> somewhere in the configuration files, some of the parameters are >> being stored as integers instead of strings. Not being able to >> reproduce this, but just looking at the code, this looks like a fix: >> >> diff -r fbdb0203b3ee41519942c705ec800333b70c3365 cobbler/action_sync.py >> --- a/cobbler/action_sync.py Wed Feb 21 10:30:33 2007 -0500 >> +++ b/cobbler/action_sync.py Thu Feb 22 10:44:21 2007 -0500 >> @@ -874,8 +874,8 @@ class BootSync: >> for key in hash: >> value = hash[key] >> if value is None: >> - buffer = buffer + key + " " >> + buffer = buffer + str(key) + " " >> else: >> - buffer = buffer + key + "=" + value + " " >> + buffer = buffer + str(key) + "=" + str(value) + " " >> return buffer > > Just to add, i experienced a similar problem on 0.4.2 and this patch > fixed it for me as well. > > - dhawal > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools Thanks for the heads up. I'll see about getting this fix pushed out. From mdehaan at redhat.com Fri Feb 23 15:57:36 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 10:57:36 -0500 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux In-Reply-To: <45DEF4B0.7040402@netmagicsolutions.com> References: <45DA2D02.1000305@redhat.com> <45DEF4B0.7040402@netmagicsolutions.com> Message-ID: <45DF0EF0.5080207@redhat.com> Dhawal Doshy wrote: > Michael DeHaan wrote: >> I've noticed that the pxelinux that ships with RHEL4 (test platform: >> RHEL4U4) does not work well at all with new PXE menus, though the >> version shipping with FC6 works great. Basically it doesn't >> understand the menu label options. >> >> If you have RHEL4's version of syslinux installed, the fix here is to >> install a newer syslinux. I'll file the bug for good measure (if >> this is even a bug?) and may bundle a known-good pxelinux.0 in the >> next release. The workaround is pretty simple though. If anyone >> still has questions, feel free to ask. > > Finally got the time to try out the newer 0.4.2 version.. the default > menu gets created nicely and i am trying to use a local kickstart file > rather than a URL (to explore templating options).. > > However, while trying to pxeboot a test server, it just stays at the > menu (tab works fine though).. looks like i need to upgrade syslinux > OR pxelinux.0 as you've mentioned. Now since syslinux needs an upgrade > (the latest being 3.36) do you have any recommendations for the version? > > - dhawal > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools It seems you've got it working (great!), though since you asked ... FC6 has 3.11-4, so anything equal to or newer than that should be fine. Looks like you're running newer and that shouldn't cause any problems. --MPD From mdehaan at redhat.com Fri Feb 23 16:46:23 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 11:46:23 -0500 Subject: [et-mgmt-tools] Using cobbler to mirror repos like updates/extras -- further information Message-ID: <45DF1A5F.10501@redhat.com> So the question has come up recently about how to use cobbler to manage updates for yum. Looking over the manpage, I think it could use some further explanation, and the mentioned rsync URLs aren't very helpful, so I'll update the manpage accordingly. In the meantime, here's some more information about repository management: --- A common use case is keeping a local mirror of FC-6 updates. cobbler repo add --mirror-url= rsync://distro.ibiblio.org/fedora-linux-core/updates/6/i386 --mirror=fc6i386updates --local-filename=updates When making new profiles, it would look something like this: cobbler profile add --name=profilefoo --kickstart=/path/to/kickstart --repos="fc6i386updates" If you want, you can manage more than one repository in cobbler. A good example to this would be managing a mirror of fc6 extras as well as updates. cobbler repo add --mirror-url=rsync://mirrors.usc.edu/fedora-core/extras/6/i386 --mirror=fc6i386updates --local-filename=updates Now, any profile that has a "--repos" setting that includes "fc6i386updates" it will (A) automatically be able to use the repo in any FC6+ kickstarts for the purpcoses of package installation (not useful for updates, but you get the idea) (B) automatically be configured to use this repo for package updates. This is done by the following template magic in the kickstart file. You are free to use your own kickstart files though /etc/cobbler/kickstart_fc6.ks is a good reference: $yum_repo_stanza (for adding the repo to anaconda) $yum_config_stanza (for setting up the repo for use by the provisioned system for future updates) If your kickstart file has both of these in it in the right place, like /etc/cobbler/kickstart_fc6.ks, your provisioned system will now be using your cobbler server as a updates mirror, as opposed to an external slower mirror. That's awesome, especially if you have a lot of systems and not a lot of external bandwidth. You'll probably want to put "cobbler reposync" on a crontab to keep all of your repos up to date. Running it nightly wouldn't be a bad idea. "cobbler reposync" will update any repos cobbler knows about. To list the configured repos, try "cobbler list repos" or "cobbler report repos". If something doesn't make sense, let me know, and hopefully I can clarify. Ideally this gives a better idea how to use your local provisioning mirror as an update mirror at the same time -- and more folks will start to set up local mirrors for provisioning purposes. This would be especially good for companies or labs that have a very large number of machines. Repository management for home users is really not required -- the default installs will be set up to use the mirror lists (which are external), but if you're really into speed or have bandwidth issues, these features could still prove interesting. Don't forget that repo management can also work with other kinds of yum repos -- including proprietary software or special tools that aren't part of a major distribution. Since cobbler uses rsync internally, it can accept user:password at host:path syntax as well as the rsync:// type paths. --Michael From dhawal at netmagicsolutions.com Fri Feb 23 17:35:32 2007 From: dhawal at netmagicsolutions.com (Dhawal Doshy) Date: Fri, 23 Feb 2007 23:05:32 +0530 Subject: [et-mgmt-tools] PXE boot menus & older versions of pxelinux/syslinux In-Reply-To: <45DF0CCD.9060404@redhat.com> References: <45DA2D02.1000305@redhat.com> <45DEF4B0.7040402@netmagicsolutions.com> <45DF035E.2060600@netmagicsolutions.com> <45DF0CCD.9060404@redhat.com> Message-ID: <45DF25E4.3030604@netmagicsolutions.com> Michael DeHaan wrote: > >> >> Finally can anyone post some more examples of the ks templating?? >> > > I assume you've seen the very basic examples in /etc/cobbler, these are > the default files that cobbler uses when one runs "cobbler import". > The simplest thing there is how it fills in $tree to point to the > location of your kickstart tree as accessible from cobbler over > http://. If you do a cobbler import (say, from an rsync mirror, or > from a DVD as msquared has shown us how to do) and then run a "cobbler > report", you'll see each profile uses a custom value for "tree", but in > the kickstart, it's just shown as "$tree". > > One thing that would be useful (I may have mentioned this before) is if > one was provisioning a set of machines that were to use static IP's.... > the profile would be common to all of them, but the different ip's would > be added with --ksmeta="ip=192.168.1.50" for the various system > entries. This would allow their IP's to be set up appropriately in > Anaconda. > > It would also be possible for two profiles to share a common kickstart > source file, and template that differently. Thanks Michael, that explains a lot more that what i could initially understand.. will let you know how it works out. - dhawal From cobbler at msquared.id.au Fri Feb 23 18:33:17 2007 From: cobbler at msquared.id.au (Msquared) Date: Sat, 24 Feb 2007 03:33:17 +0900 Subject: [et-mgmt-tools] Using cobbler to mirror repos like updates/extras -- further information In-Reply-To: <45DF1A5F.10501@redhat.com> References: <45DF1A5F.10501@redhat.com> Message-ID: <20070223183317.GB3332@sliderule.msquared.com.au> On Fri, Feb 23, 2007 at 11:46:23AM -0500, Michael DeHaan wrote: > A common use case is keeping a local mirror of FC-6 updates. What happens if I have a local mirror of the Fedora 6 DVD plus a local mirror of the corresponding FC6 updates? Can I add both to a profile, and kickstart will do the Right Thing when building the new machine? Regards, Msquared... From mdehaan at redhat.com Fri Feb 23 19:14:26 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 14:14:26 -0500 Subject: [et-mgmt-tools] Using cobbler to mirror repos like updates/extras -- further information In-Reply-To: <20070223183317.GB3332@sliderule.msquared.com.au> References: <45DF1A5F.10501@redhat.com> <20070223183317.GB3332@sliderule.msquared.com.au> Message-ID: <45DF3D12.4040006@redhat.com> Msquared wrote: > On Fri, Feb 23, 2007 at 11:46:23AM -0500, Michael DeHaan wrote: > > >> A common use case is keeping a local mirror of FC-6 updates. >> > > What happens if I have a local mirror of the Fedora 6 DVD plus a local > mirror of the corresponding FC6 updates? Can I add both to a profile, and > kickstart will do the Right Thing when building the new machine? > > Regards, Msquared... > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > What we have here is a failure to communicate :) Two kinds of mirrors, basically. Different concepts, used for different things. "cobbler import" puts files into /var/www/cobbler/ks_mirror -- this is a kickstart tree mirror. These don't need to be updated, and are essential for doing full automated installations. However since Anaconda /does/ allow network installs, you don't /have/ to mirror them on your cobbler server if you already have a good kickstart tree available over http on your network. In this case, you'd just use "cobbler distro add", and save yourself the import steps. However, most home users won't have an fast kickstart tree available to them, and this is why cobbler import helps you make one. cobbler repo add ... puts files into /var/www/cobbler/repo_mirror -- these are yum repositories for things like extras & updates. These update quite frequently and are entirely optional for cobbler -- if you don't use them, it will use external repos as configured by default in yum. If you add the later (the yum repos) you'll either have to edit them into the "repos" field of an existing profile in /var/lib/cobbler/profiles, or you can add create new profiles with "cobbler profile add" using the "--repo" argument. Note that if you do this, and you're using the default cobbler kickstart files, you'll still need to pass in an appropriate value for the --ksmeta option "tree". This is why I'd say it's much easier to just edit /var/lib/cobbler/profiles with the profiles that cobbler imports for you. Add yes, it will do the right thing. To verify that it does the right thing, run a "cobbler sync", and check out the kickstart files as seen at /var/www/cobbler/kickstarts/profilename/ks.cfg -- you should see a good parameter passed in for "url" towards the top, plus references to your repo in the middle (search for "repo") and also in post, where it installs the repo file in /etc/yum.repos.d --Michael From cobbler at msquared.id.au Fri Feb 23 19:40:53 2007 From: cobbler at msquared.id.au (Msquared) Date: Sat, 24 Feb 2007 04:40:53 +0900 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <45DEFF9D.6060400@redhat.com> <45DEFEF5.2020802@redhat.com> References: <20070223070524.GA3332@sliderule.msquared.com.au> <45DEFEF5.2020802@redhat.com> <45DEFF9D.6060400@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DDD55E.6080404@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <20070223070524.GA3332@sliderule.msquared.com.au> <45DEFEF5.2020802@redhat.com> Message-ID: <20070223194053.GC3332@sliderule.msquared.com.au> On Fri, Feb 23, 2007 at 09:49:25AM -0500, Michael DeHaan wrote: > And then, you can use the following parameter in your kickstart files > for the "--url" > > url --url http://servername/cobbler_track/local_mirror/fc6tree/restofpath I was just about to ask about cobbler_track, but then I noticed that /etc/httpd/conf.d/cobbler.conf takes care of it. Nice! > Note that you want to include the full path to the "os" or "tree" > directory, and don't want to stop at the root. That's what I mean by > "restofpath". So in the case of a Fedora DVD, which of the following should that be: .../local_mirror/fc6/repodata .../local_mirror/fc6/Fedora .../local_mirror/fc6/Fedora/base .../local_mirror/fc6/Fedora/RPMS ? > I think you've given an excellent reason to add "how to import a > kickstart tree from a DVD" to the manpage. I'd like the manpage to clarify that 'cobbler import --path' doesn't actually copy any files about. Of course, in retrospect this makes sense to me, since you're importing from a location that is already locally accessible. :-) But at the time I was trying to understand all of this, I didn't know what to expect, and my brain was trying to fit everything into orthogonal little boxes, and assuming that import should unilaterally slurp from wherever it was given into its own local store made sense to me. > > cobbler import --mirror=root at localhost:/media/cdrom > > --mirror-name=testimport > > Actually that will work. Creative workaround! :) I realised that might work from this info here: http://cobbler.et.redhat.com/cobbler.html#repository_mirroring_workflow +1 for the manpage. :) There was something I noticed after doing that import, though: the import process appeared to use the last component of the path I gave as a component in the path inside ks_mirror. After pondering it momentarily, I realised it's an rsync feature related to the trailing slash (or lack thereof, in this case). Perhaps it's worth moentioning that cobbler import --mirror=root at localhost:/media/cdrom --mirror-name=fc6 will copy the files into /var/www/cobbler/ks_mirror/fc6/cdrom whereas cobbler import --mirror=root at localhost:/media/cdrom/ --mirror-name=fc6 will copy the files into /var/www/cobbler/ks_mirror/fc6 Most likely, the second one is the one that is desired. :) Also, should I import root at localhost:/media/cdrom/ or root at localhost:/media/cdrom/restofpath/ (to refer to your --url example above) ? > >However, will this cause me grief in future? What happens when I > >'cobbler sync'? Will it try to re-mirror based on whatever is in > >/media/cdrom? Can I stop it from doing that? > > Cobbler sync won't try to re-mirror that, because cobbler knows > kickstart trees really don't change. Oh yeah, that made me realise something: After running the import, I couldn't find anywhere in /var/lib/cobbler/* that referred to the files that were just imported, However, after my rsync-slash discovery, I retried the import but with the trailing slash, and noticed that there was more information in the profile, this time! Specifically: ks_meta: tree: 'http://provisioning//ks_mirror/fc6' Previously ks_meta was just {} (which means an empty list?) I think that answers my question about how the provisioning knows about the files that were imported, but shouldn't that be: tree: 'http://provisioning/cobbler/ks_mirror/fc6' ? > >Also, should I make the mirror-name something that is related to where > >I'm getting the files from (eg: fc6dvd or fc6public), related to what > >the files are for (eg: fc6), or related to my provisioning server (eg: > >localprovserver; ie: the machine I am running cobbler on)? > > The mirror name is basically just to pick something you can remember > later. I realise that, but if I have an idea of a convention that I could use, it may be easier to recall later. :) I'm sure those who have experience will have their own ideas, but it would be nice to have at least a suggestion for newbies. It's may be worth noting that the name you supply will be used by cobbler to build the name of a distro and a profile based on that import. Mind you, although it's easy to remember later that I called it 'fc6', it's a bit harder to remember that cobbler translated that into 'var_www_cobbler_ks_mirror_fc6_images_xen', so +1 for your auto-shorten idea. :) > Since you've actually imported the profile, you're going to need to go > into /var/lib/cobbler/profiles to list the repos you use. This isn't > really as scary as it sounds, but if you're concerned, you can back up > the profiles file first. Ah, so distros and repos are kind of independent? By this I mean that cobbler doesn't intrinsically know what repo is related to what distro. You make that association as part of tweaking the profile, right? > >I've tried to import from the DVD with these, but none of them work: > > > > cobbler import --mirror=root at localhost > > I thought you said you just got that working earlier? Anyhow, import > also requires a "--mirror-name". Oops, sorry about that. Email compose error. :) While I was composing the email, the very act of expressing the problem would sometimes suggest a possible solution, so I'd go ahead and play with it, and often it did help. I was part-way through composing that when I discovered the ssh magic that allowed me to import the DVD and have cobbler retrieve a copy for itself. I obviously didn't proof-read enough before I sent the email. > The easiest way for this to happen is for you to write up something, put > it up on the web, and I can link it from the Cobbler webpage. Groovy. I'll try to get some time this weekend to put up the sum total of what I have learnt thus far. > As with the DVD import, folks using cobbler often think of things I > don't think of, and it's great to see what the things they want to do > with provisioning are. In this case, I want to set up a server for experimentation with various web CMS and blog technologies. I decided that Xen would mean the ability to experiment endlessly and just rebuild when required, and cobbler/koan would even simplify the rebuild process. So far I have not been able to provision a Xen guest using the instructions at http://fedoraproject.org/wiki/FedoraXenQuickstartFC6 (although part of the problem may be that I have not set up X, and thus virt-install is perhaps more fiddly than virt-manager). I don't want X because I want a lean environment. The example given on that page using cobbler/koan looks like they copy/pasted from the cobbler manual, or something. They didn't even give suitable example paths and data. At the very least I want to provide enough docs to make that bit work, but I'm happy to go much further. > Yeah, repos are a relatively new concept, and are used for attaching > things like "fc6updatesi386" to a profile, so that they can > automatically install and configure a yum mirror when they provision it, > as well as using a yum mirror at install time to install packages. If I add an updates mirror to the profile, will updated packages be installed instead of the ones from the original DVD tree, where there are updates available? On Fri, Feb 23, 2007 at 09:52:13AM -0500, Michael DeHaan wrote: > >What I normally do in this situation is add a symlink, like > > > >/var/www/cobbler/local_mirror/fc6tree -> /opt/dist/fc6 > > Correction: the way you invoked the import (using the root:localhost@ > syntax), you don't need to do this at all. Yeah, I worked that out. > Check to make sure content did get copied to /var/www/cobbler/ks_mirror > though ... Yes, it did copy the files. Thanks! Regards, Msquared... From cobbler at msquared.id.au Fri Feb 23 19:54:18 2007 From: cobbler at msquared.id.au (Msquared) Date: Sat, 24 Feb 2007 04:54:18 +0900 Subject: [et-mgmt-tools] Using cobbler to mirror repos like updates/extras -- further information In-Reply-To: <45DF3D12.4040006@redhat.com> References: <45DF1A5F.10501@redhat.com> <20070223183317.GB3332@sliderule.msquared.com.au> <45DF3D12.4040006@redhat.com> Message-ID: <20070223195418.GD3332@sliderule.msquared.com.au> On Fri, Feb 23, 2007 at 02:14:26PM -0500, Michael DeHaan wrote: > Two kinds of mirrors, basically. Different concepts, used for different > things. Ah, I see. > To verify that it does the right thing, run a "cobbler sync", and check > out the kickstart files as seen at > /var/www/cobbler/kickstarts/profilename/ks.cfg -- you should see a good > parameter passed in for "url" towards the top, plus references to your > repo in the middle (search for "repo") and also in post, where it > installs the repo file in /etc/yum.repos.d Ah, I see. 'cobbler sync' builds the ks.cfg from the supplied kickstart file. I did notice that '$' must be escaped in the supplied kickstart file, since it's used to introduce ks meta data. Anaconda kindly generates anaconda-ks.cfg when it installs a system for you. Can that be used as a source for 'cobbler sync', as long as the '$' are escaped? If not, what other things ought to be done first? Regards, Msquared... From mdehaan at redhat.com Fri Feb 23 20:02:38 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 15:02:38 -0500 Subject: [et-mgmt-tools] Adding a distribution - where do I put the files? In-Reply-To: <20070223194053.GC3332@sliderule.msquared.com.au> References: <20070223070524.GA3332@sliderule.msquared.com.au> <45DEFEF5.2020802@redhat.com> <45DEFF9D.6060400@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <45DDD55E.6080404@redhat.com> <20070222104151.GA23723@sliderule.msquared.com.au> <45DDBD4F.4070709@redhat.com> <20070223070524.GA3332@sliderule.msquared.com.au> <45DEFEF5.2020802@redhat.com> <20070223194053.GC3332@sliderule.msquared.com.au> Message-ID: <45DF485E.7040702@redhat.com> Msquared wrote: > On Fri, Feb 23, 2007 at 09:49:25AM -0500, Michael DeHaan wrote: > > >> And then, you can use the following parameter in your kickstart files >> for the "--url" >> >> url --url http://servername/cobbler_track/local_mirror/fc6tree/restofpath >> > > I was just about to ask about cobbler_track, but then I noticed that > /etc/httpd/conf.d/cobbler.conf takes care of it. Nice! > > >> Note that you want to include the full path to the "os" or "tree" >> directory, and don't want to stop at the root. That's what I mean by >> "restofpath". >> > > So in the case of a Fedora DVD, which of the following should that be: > > .../local_mirror/fc6/repodata > .../local_mirror/fc6/Fedora > .../local_mirror/fc6/Fedora/base > .../local_mirror/fc6/Fedora/RPMS > > ? > > Yeah, since you did the DVD import, paths are shorter here. My fault. Just fc6, since that directory contains "repodata" and "images". >> I think you've given an excellent reason to add "how to import a >> kickstart tree from a DVD" to the manpage. >> > > I'd like the manpage to clarify that 'cobbler import --path' doesn't > actually copy any files about. Of course, in retrospect this makes sense > to me, since you're importing from a location that is already locally > accessible. :-) > That's been added. > But at the time I was trying to understand all of this, I didn't know what > to expect, and my brain was trying to fit everything into orthogonal > little boxes, and assuming that import should unilaterally slurp from > wherever it was given into its own local store made sense to me. > > >>> cobbler import --mirror=root at localhost:/media/cdrom >>> --mirror-name=testimport >>> >> >> Actually that will work. Creative workaround! :) >> > > I realised that might work from this info here: > > http://cobbler.et.redhat.com/cobbler.html#repository_mirroring_workflow > > +1 for the manpage. :) > > There was something I noticed after doing that import, though: the import > process appeared to use the last component of the path I gave as a > component in the path inside ks_mirror. > > After pondering it momentarily, I realised it's an rsync feature related > to the trailing slash (or lack thereof, in this case). Perhaps it's worth > moentioning that > > cobbler import --mirror=root at localhost:/media/cdrom --mirror-name=fc6 > > will copy the files into > > /var/www/cobbler/ks_mirror/fc6/cdrom > > whereas > > cobbler import --mirror=root at localhost:/media/cdrom/ --mirror-name=fc6 > > will copy the files into > > /var/www/cobbler/ks_mirror/fc6 > > Most likely, the second one is the one that is desired. :) > Yes, I can have the code ensure there's a trailing slash. If not, no major harm done, cobbler's smart enough to still make everything work. > Also, should I import root at localhost:/media/cdrom/ or > root at localhost:/media/cdrom/restofpath/ (to refer to your --url example > above) ? > > >>> However, will this cause me grief in future? What happens when I >>> 'cobbler sync'? Will it try to re-mirror based on whatever is in >>> /media/cdrom? Can I stop it from doing that? >>> >> >> Cobbler sync won't try to re-mirror that, because cobbler knows >> kickstart trees really don't change. >> > > Oh yeah, that made me realise something: After running the import, I > couldn't find anywhere in /var/lib/cobbler/* that referred to the files > that were just imported, > Other than the http:// paths in tree, that's true. > However, after my rsync-slash discovery, I retried the import but with the > trailing slash, and noticed that there was more information in the > profile, this time! Specifically: > > ks_meta: > tree: 'http://provisioning//ks_mirror/fc6' > Yes, you're one step ahead of me. > Previously ks_meta was just {} (which means an empty list?) > An empty dictionary, actually, but yes. The configuration files are YAML. > I think that answers my question about how the provisioning knows about > the files that were imported, but shouldn't that be: > > tree: 'http://provisioning/cobbler/ks_mirror/fc6' > > ? > > If the string "cobbler" is missing from your DVD import, that's pretty darn weird and needs to be fixed. >>> Also, should I make the mirror-name something that is related to where >>> I'm getting the files from (eg: fc6dvd or fc6public), related to what >>> the files are for (eg: fc6), or related to my provisioning server (eg: >>> localprovserver; ie: the machine I am running cobbler on)? >>> >> >> The mirror name is basically just to pick something you can remember >> later. >> > > I realise that, but if I have an idea of a convention that I could use, it > may be easier to recall later. :) I'm sure those who have experience > will have their own ideas, but it would be nice to have at least a > suggestion for newbies. > > It's may be worth noting that the name you supply will be used by cobbler > to build the name of a distro and a profile based on that import. > > Mind you, although it's easy to remember later that I called it 'fc6', > it's a bit harder to remember that cobbler translated that into > 'var_www_cobbler_ks_mirror_fc6_images_xen', so +1 for your auto-shorten > idea. :) > > >> Since you've actually imported the profile, you're going to need to go >> into /var/lib/cobbler/profiles to list the repos you use. This isn't >> really as scary as it sounds, but if you're concerned, you can back up >> the profiles file first. >> > > Ah, so distros and repos are kind of independent? By this I mean that > cobbler doesn't intrinsically know what repo is related to what distro. > You make that association as part of tweaking the profile, right? > Yes, repos are a profile concept. If there is a repo called "fancy_webserver_tools" we might only want to enable it for webserver profiles, for instance. > >>> I've tried to import from the DVD with these, but none of them work: >>> >>> cobbler import --mirror=root at localhost >>> >> I thought you said you just got that working earlier? Anyhow, import >> also requires a "--mirror-name". >> > > Oops, sorry about that. Email compose error. :) While I was composing > the email, the very act of expressing the problem would sometimes suggest > a possible solution, so I'd go ahead and play with it, and often it did > help. I was part-way through composing that when I discovered the ssh > magic that allowed me to import the DVD and have cobbler retrieve a copy > for itself. I obviously didn't proof-read enough before I sent the email. > > >> The easiest way for this to happen is for you to write up something, put >> it up on the web, and I can link it from the Cobbler webpage. >> > > Groovy. I'll try to get some time this weekend to put up the sum total of > what I have learnt thus far. > > >> As with the DVD import, folks using cobbler often think of things I >> don't think of, and it's great to see what the things they want to do >> with provisioning are. >> > > In this case, I want to set up a server for experimentation with various > web CMS and blog technologies. I decided that Xen would mean the ability > to experiment endlessly and just rebuild when required, and cobbler/koan > would even simplify the rebuild process. > > So far I have not been able to provision a Xen guest using the > instructions at http://fedoraproject.org/wiki/FedoraXenQuickstartFC6 > (although part of the problem may be that I have not set up X, and thus > virt-install is perhaps more fiddly than virt-manager). I don't want X > because I want a lean environment. > > The example given on that page using cobbler/koan looks like they > copy/pasted from the cobbler manual, or something. They didn't even give > suitable example paths and data. They happens to be the author, in this case. > At the very least I want to provide > enough docs to make that bit work, but I'm happy to go much further. > > >> Yeah, repos are a relatively new concept, and are used for attaching >> things like "fc6updatesi386" to a profile, so that they can >> automatically install and configure a yum mirror when they provision it, >> as well as using a yum mirror at install time to install packages. >> > > If I add an updates mirror to the profile, will updated packages be > installed instead of the ones from the original DVD tree, where there are > updates available? > Yes. From mdehaan at redhat.com Fri Feb 23 20:04:14 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 23 Feb 2007 15:04:14 -0500 Subject: [et-mgmt-tools] Using cobbler to mirror repos like updates/extras -- further information In-Reply-To: <20070223195418.GD3332@sliderule.msquared.com.au> References: <45DF1A5F.10501@redhat.com> <20070223183317.GB3332@sliderule.msquared.com.au> <45DF3D12.4040006@redhat.com> <20070223195418.GD3332@sliderule.msquared.com.au> Message-ID: <45DF48BE.9040606@redhat.com> Msquared wrote: > On Fri, Feb 23, 2007 at 02:14:26PM -0500, Michael DeHaan wrote: > > >> Two kinds of mirrors, basically. Different concepts, used for different >> things. >> > > Ah, I see. > > >> To verify that it does the right thing, run a "cobbler sync", and check >> out the kickstart files as seen at >> /var/www/cobbler/kickstarts/profilename/ks.cfg -- you should see a good >> parameter passed in for "url" towards the top, plus references to your >> repo in the middle (search for "repo") and also in post, where it >> installs the repo file in /etc/yum.repos.d >> > > Ah, I see. 'cobbler sync' builds the ks.cfg from the supplied kickstart > file. I did notice that '$' must be escaped in the supplied kickstart > file, since it's used to introduce ks meta data. > > Anaconda kindly generates anaconda-ks.cfg when it installs a system for > you. Can that be used as a source for 'cobbler sync', as long as the '$' > are escaped? If not, what other things ought to be done first? > Yes. In general, you don't have to escape all of the '$' characters, but only the ones that cause problems. \$(list-hard-drives) is the only such example that requires mandatory escaping in the cobbler files. Generally this means anaconda-cfg.ks files can be used without modification. > Regards, Msquared... > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From mark.cave-ayland at ilande.co.uk Sat Feb 24 11:01:28 2007 From: mark.cave-ayland at ilande.co.uk (Mark Cave-Ayland) Date: Sat, 24 Feb 2007 11:01:28 +0000 Subject: [et-mgmt-tools] virt-manager 0.3.0 can't see inactive vms? In-Reply-To: <20070223121622.GD14993@redhat.com> References: <1171474335.15655.17.camel@mca-desktop> <20070214174827.GD29628@redhat.com> <1171530362.7896.8.camel@mca-desktop> <20070215115806.GA16485@redhat.com> <1172221859.21229.21.camel@mca-desktop> <20070223121622.GD14993@redhat.com> Message-ID: <1172314888.5190.2.camel@mca-desktop> On Fri, 2007-02-23 at 12:16 +0000, Daniel P. Berrange wrote: > Looking at the code, that is indeed true. I'll see about fixing that. > In fact I think I'll probably remove the 'read only' checkbox from the > dialog too, and have virt-manager automatically try a full read write > connection, falling back to read only when doing Xen as non-root. > > Dan. Hi Dan, Yeah that would be great, especially if you could change the title of the main window so it has a "[Read only]" suffix if running as non-root/read only mode. It took a while for me to figure out what was happening without the console output... Kind regards, Mark. From fukuta.saori at jp.fujitsu.com Mon Feb 26 05:55:23 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Mon, 26 Feb 2007 14:55:23 +0900 Subject: [et-mgmt-tools] [PATCH] check a MAC address format Message-ID: <20070226145309.6B48.FUKUTA.SAORI@jp.fujitsu.com> Hi, When I install by virt-intall with invalid MAC address(e.g."--mac=AA:"), virt-install get an error message that says "unable to connect to host:Connection refused(111)". So, here's the patch adds to check a MAC address with following format: xx:xx:xx:xx:xx:xx ( xx should be in hexadecimal ) Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virtinst-0.101.0/virtinst/Guest.py =================================================================== --- Guest.py 2007-02-21 05:26:23.000000000 +0900 +++ Guest.py.mac 2007-02-26 11:54:51.000000000 +0900 @@ -141,6 +141,10 @@ class XenDisk(VirtualDisk): class VirtualNetworkInterface: def __init__(self, macaddr = None, bridge = None): + if macaddr is not None: + form = re.match("^([0-9a-fA-F]{2}:){5}[0-9a-fA-F]{2}$",macaddr) + if form is None: + raise ValueError, "Invalid value for MAC address" self.macaddr = macaddr self.bridge = bridge From satyaakam at gmail.com Mon Feb 26 08:00:30 2007 From: satyaakam at gmail.com (satyaakam goswami) Date: Mon, 26 Feb 2007 13:30:30 +0530 Subject: [et-mgmt-tools] RHEL 5.0 Beta 2 and Paravirtualization Message-ID: <6491e1350702260000w2ae3c6f0oe8ea412c55a22180@mail.gmail.com> Hi All, I did not know where to start thought this list will be good place to begin I am trying to install xen in paravirtualized mode , and RHEL 5 Beta comes with virt-manager 0.2.5 i am trying to install a guest system ie: RHEL 3.0 the preliminary questions i have are as follows 1)Does the virt-manager version 0.2.5 support RHEL 3.0 as guest system in paravirtualized mode. 2)Are there any other tools to start a kickstart installation for RHEL 3.0 as guest on RHEL 5.0 beta 2.0 Satya From rzarouali at gmail.com Mon Feb 26 09:49:41 2007 From: rzarouali at gmail.com (Rachid Zarouali) Date: Mon, 26 Feb 2007 10:49:41 +0100 Subject: [et-mgmt-tools] RHEL 5.0 Beta 2 and Paravirtualization In-Reply-To: <6491e1350702260000w2ae3c6f0oe8ea412c55a22180@mail.gmail.com> References: <6491e1350702260000w2ae3c6f0oe8ea412c55a22180@mail.gmail.com> Message-ID: hy satya, virt-manager with rhel5beta2 for me is not working (at this time i don't if it's rhel side or virt-manager side bug :( ). but if you plain to use paravirtualized guest on rhel5 dom0, remember to rebuild your xen initrd with the xenblk module (mkinitrd --with=xennet --preload=xenblk /boot/initrd- 2.6.18-1.2747.el5xenU.img 2.6.18-1.2747.el5xen) or your paravirtualized guest won't start (knocked my head of on this for quite a day :( ) have you tried virt-install ? which have worked perfectly for me . i've also played with xenman (wich needed a little bit of tweaking to work ;)). maybe tools like enomalism would work too. hope this helps rachid On 2/26/07, satyaakam goswami wrote: > > Hi All, > I did not know where to start thought this list will be good > place to begin I am trying to install xen in paravirtualized mode , > and RHEL 5 Beta comes with virt-manager 0.2.5 i am trying to install a > guest system ie: RHEL 3.0 the preliminary questions i have are as > follows > > 1)Does the virt-manager version 0.2.5 support RHEL 3.0 as guest system > in paravirtualized mode. > 2)Are there any other tools to start a kickstart installation for RHEL > 3.0 as guest on RHEL 5.0 beta 2.0 > > Satya > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > -------------- next part -------------- An HTML attachment was scrubbed... URL: From berrange at redhat.com Mon Feb 26 13:47:26 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Mon, 26 Feb 2007 13:47:26 +0000 Subject: [et-mgmt-tools] RHEL 5.0 Beta 2 and Paravirtualization In-Reply-To: <6491e1350702260000w2ae3c6f0oe8ea412c55a22180@mail.gmail.com> References: <6491e1350702260000w2ae3c6f0oe8ea412c55a22180@mail.gmail.com> Message-ID: <20070226134726.GB23075@redhat.com> On Mon, Feb 26, 2007 at 01:30:30PM +0530, satyaakam goswami wrote: > Hi All, > I did not know where to start thought this list will be good > place to begin I am trying to install xen in paravirtualized mode , > and RHEL 5 Beta comes with virt-manager 0.2.5 i am trying to install a > guest system ie: RHEL 3.0 the preliminary questions i have are as > follows > > 1)Does the virt-manager version 0.2.5 support RHEL 3.0 as guest system > in paravirtualized mode. No. This isn't a limitation of virt-manager - there is no paravirt kernel available for RHEL 3, so your only option is to do fullvirt for RHEL 3. > 2)Are there any other tools to start a kickstart installation for RHEL > 3.0 as guest on RHEL 5.0 beta 2.0 You can either use virt-manager or the command line tool virt-install. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From hbrock at redhat.com Mon Feb 26 23:05:01 2007 From: hbrock at redhat.com (Hugh Brock) Date: Mon, 26 Feb 2007 18:05:01 -0500 Subject: [et-mgmt-tools] virt-manager roadmap Message-ID: <45E3679D.4080503@redhat.com> All: I have just posted a new virt-manager roadmap at http://virt-manager.org/roadmap.html. It reflects our discussions with Fujitsu last week along with other features various developers and clients have requested. Please feel free to comment here on the list, or post patches to the roadmap as necessary. Thank you, --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From John.Sanabria at ece.uprm.edu Tue Feb 27 02:10:35 2007 From: John.Sanabria at ece.uprm.edu (John Sanabria) Date: Mon, 26 Feb 2007 22:10:35 -0400 (AST) Subject: [et-mgmt-tools] java support... Message-ID: <42408.136.145.116.114.1172542235.squirrel@ece.uprm.edu> Hi, there exist a "wrapper" to access from java to libvirt? or some ideas for to do that? regards. http://ece.uprm.edu/~s047267 http://del.icio.us/josanabr http://blog-grid.blogspot.com From florian.heigl at gmail.com Tue Feb 27 02:58:42 2007 From: florian.heigl at gmail.com (Florian Heigl) Date: Tue, 27 Feb 2007 03:58:42 +0100 Subject: [et-mgmt-tools] virt-manager roadmap In-Reply-To: <45E3679D.4080503@redhat.com> References: <45E3679D.4080503@redhat.com> Message-ID: <77abe410702261858p29931ff6vef6cedf5e8aa8be2@mail.gmail.com> 2007/2/27, Hugh Brock : > All: I have just posted a new virt-manager roadmap at > http://virt-manager.org/roadmap.html. It reflects our discussions with > Fujitsu last week along with other features various developers and > clients have requested. Please feel free to comment here on the list, or > post patches to the roadmap as necessary. I desparately miss simultanously connecting to multiple Xen hosts. From berrange at redhat.com Tue Feb 27 03:54:04 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 27 Feb 2007 03:54:04 +0000 Subject: [et-mgmt-tools] java support... In-Reply-To: <42408.136.145.116.114.1172542235.squirrel@ece.uprm.edu> References: <42408.136.145.116.114.1172542235.squirrel@ece.uprm.edu> Message-ID: <20070227035404.GB32072@redhat.com> On Mon, Feb 26, 2007 at 10:10:35PM -0400, John Sanabria wrote: > Hi, > > there exist a "wrapper" to access from java to libvirt? or some ideas for > to do that? None that I am aware of at this time, although it'd be desirable to have someone write bindings in the future. FYI, libvirt questions should be sent to the dedicated libvirt mailing list libvir-list at redhat.com Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Tue Feb 27 04:05:41 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Tue, 27 Feb 2007 04:05:41 +0000 Subject: [et-mgmt-tools] virt-manager roadmap In-Reply-To: <77abe410702261858p29931ff6vef6cedf5e8aa8be2@mail.gmail.com> References: <45E3679D.4080503@redhat.com> <77abe410702261858p29931ff6vef6cedf5e8aa8be2@mail.gmail.com> Message-ID: <20070227040541.GC32072@redhat.com> On Tue, Feb 27, 2007 at 03:58:42AM +0100, Florian Heigl wrote: > 2007/2/27, Hugh Brock : > >All: I have just posted a new virt-manager roadmap at > >http://virt-manager.org/roadmap.html. It reflects our discussions with > >Fujitsu last week along with other features various developers and > >clients have requested. Please feel free to comment here on the list, or > >post patches to the roadmap as necessary. > > I desparately miss simultanously connecting to multiple Xen hosts. This is probably the single biggest item on the roadmap - so big in fact that we've split it up into many tasks. To do remote managemnet 'right' is no easy task. First we have to make sure we're portable to any virt platform (Xen, KVM, QEMU, etc), so we don't want to simply use the Xen specific remote API - instead we're working on a libvirt daemon to provide remote management service for all libvirt backends. Security is incredibly important too - no existing remote maangement API for Xen is secure - they all assume you're tunnelling over SSH which isn't really a viable solution for large scale deployment. Thus we're making sure libvirt daemon uses TLS encryption & certificate authentication (same security model as used in web servers/browsers). Then there's the quesiton fo the remote console - VNC is not a secure protocol in its regular impl, so we're working to make QEMU, Xen paravirt framebugger, and the virt-manager VNC widget all support a TLS security extension for VNC. Storage & network management is also an important thing that needs addressing - we can't simply enumerate local devices / create files when managing remote hosts. We're working on all these problems & more in parallel, and if all goes to plan, come Fedora 8 we'll have a complete remote management solution for virt-manager with end-to-end TLS security for every component. Some bits may well be ready before Fedora 8, and if so we'll push updates out to Fedora 6 / 7 to let people leverage new capabilities as easily as possible. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From james at cloud9.co.uk Tue Feb 27 13:20:34 2007 From: james at cloud9.co.uk (James Fidell) Date: Tue, 27 Feb 2007 13:20:34 +0000 Subject: [et-mgmt-tools] Xen domU + koan Message-ID: <45E43022.1050208@cloud9.co.uk> Are there any examples and/or walkthroughs of how to set up a Xen domU instance using koan and cobbler? I *think* from reading the manpage that I should be running something like "koan --virt ..." from dom0, but it isn't clear to me how xen works out which configuration to use for the domU instance, which kernel to boot etc. James From hbrock at redhat.com Tue Feb 27 15:20:44 2007 From: hbrock at redhat.com (Hugh Brock) Date: Tue, 27 Feb 2007 10:20:44 -0500 Subject: [et-mgmt-tools] [PATCH] check a MAC address format In-Reply-To: <20070226145309.6B48.FUKUTA.SAORI@jp.fujitsu.com> References: <20070226145309.6B48.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <45E44C4C.2080302@redhat.com> Saori Fukuta wrote: > Hi, > > When I install by virt-intall with invalid MAC address(e.g."--mac=AA:"), > virt-install get an error message that says "unable to connect to > host:Connection refused(111)". > > So, here's the patch adds to check a MAC address with following format: > xx:xx:xx:xx:xx:xx ( xx should be in hexadecimal ) > > Signed-off-by: Saori Fukuta > > Thanks, > Saori Fukuta. > > Index: virtinst-0.101.0/virtinst/Guest.py > =================================================================== > --- Guest.py 2007-02-21 05:26:23.000000000 +0900 > +++ Guest.py.mac 2007-02-26 11:54:51.000000000 +0900 > @@ -141,6 +141,10 @@ class XenDisk(VirtualDisk): > > class VirtualNetworkInterface: > def __init__(self, macaddr = None, bridge = None): > + if macaddr is not None: > + form = re.match("^([0-9a-fA-F]{2}:){5}[0-9a-fA-F]{2}$",macaddr) > + if form is None: > + raise ValueError, "Invalid value for MAC address" > self.macaddr = macaddr > self.bridge = bridge > I have applied this, thanks! --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock at redhat.com | virtualization library http://libvirt.org From mdehaan at redhat.com Tue Feb 27 15:37:50 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 27 Feb 2007 10:37:50 -0500 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E43022.1050208@cloud9.co.uk> References: <45E43022.1050208@cloud9.co.uk> Message-ID: <45E4504E.4000806@redhat.com> James Fidell wrote: > Are there any examples and/or walkthroughs of how to set up a Xen domU > instance using koan and cobbler? > > I *think* from reading the manpage that I should be running something > like "koan --virt ..." from dom0, but it isn't clear to me how xen > works out which configuration to use for the domU instance, which kernel > to boot etc. > > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > Basically just set up cobbler and import a tree (or set it up manually) on the server side, just like you'd do with PXE. From the koan side, specify the name of a Xen profile, which is by definition one that has a distro that uses a Xen (domU) kernel. Koan will talk with the cobbler server to get the right parameters, just as they are configured in cobbler. --Michael From james at cloud9.co.uk Tue Feb 27 16:33:37 2007 From: james at cloud9.co.uk (James Fidell) Date: Tue, 27 Feb 2007 16:33:37 +0000 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E4504E.4000806@redhat.com> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com> Message-ID: <45E45D61.8080309@cloud9.co.uk> Michael DeHaan wrote: > Basically just set up cobbler and import a tree (or set it up manually) > on the server side, just like you'd do with PXE. Ok, that's done. > From the koan side, specify the name of a Xen profile, which is by > definition one that has a distro that uses a Xen (domU) kernel. In this case I want a fully virtual system, so I can use tell koan to use a normal cobbler profile, I think. > Koan will talk with the cobbler server to get the right parameters, just > as they are configured in cobbler. Then I start to lose the plot :) Presumably the domU must already be configured, otherwise there'd be no way to find out how its hard disk image is mapped from dom0, amongst other things? Can I literally do: 1. Set up cobbler 2. Create config file for domU 3. xm create ... 4. koan ... or are there other steps I've missed? James From mdehaan at redhat.com Tue Feb 27 16:49:32 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 27 Feb 2007 11:49:32 -0500 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E45D61.8080309@cloud9.co.uk> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com> <45E45D61.8080309@cloud9.co.uk> Message-ID: <45E4611C.6060406@redhat.com> James Fidell wrote: > Michael DeHaan wrote: > > >> Basically just set up cobbler and import a tree (or set it up manually) >> on the server side, just like you'd do with PXE. >> > > Ok, that's done. > > >> From the koan side, specify the name of a Xen profile, which is by >> definition one that has a distro that uses a Xen (domU) kernel. >> > > In this case I want a fully virtual system, so I can use tell koan to > use a normal cobbler profile, I think. > > >> Koan will talk with the cobbler server to get the right parameters, just >> as they are configured in cobbler. >> > > Then I start to lose the plot :) Presumably the domU must already be > configured, otherwise there'd be no way to find out how its hard disk > image is mapped from dom0, amongst other things? > > Can I literally do: > > 1. Set up cobbler > 2. Create config file for domU > 3. xm create ... > 4. koan ... > > or are there other steps I've missed? > Hmm, I think you may be looking for koan to do something else than what it's designed to do, but I'm not sure what. Koan replaces the xm create. It creates the domU image for you, same as "xm create" or (preferred) virtguest-install would do, except it does so using a reusable profile that can be controlled server side and shared with multiple dom0 machines for repeatable results. Suppose you have already used cobbler to install a datacenter/lab full of dom0 machines (maybe you installed them with cobbler, and if you did, you get bonus points). You want, on any of these machines, to be able to repeatably install virtual images, with specific kickstarts, and specific disk space and RAM requirements. Say you want to install a new virtual webserver on many of them. So you create a "fc6virtualwebserver" profile in cobbler. Now you can go around to these dom0 machines and repeatedly install the same virtual profile on each of them: koan --virt --profile=fc6virtualwebserver --server=bootserver.example.com For another example, say many people frequently wanted to install a special development environment on their workstations. You could create a cobbler profile for that environment, and then anyone who wanted to install it could just run: koan --virt --profile=specialDevelopmentEnvironment --server=bootserver.example.com This eliminates the need to know the arguments needed for virtguest-install or xm-create and allows the definitions of these profiles to be managed centrally, along with their associated kickstart trees and repository mirrors. That's what koan does. > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From John.Sanabria at ece.uprm.edu Tue Feb 27 18:32:38 2007 From: John.Sanabria at ece.uprm.edu (John Sanabria) Date: Tue, 27 Feb 2007 14:32:38 -0400 (AST) Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E4611C.6060406@redhat.com> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com><45E45D61.8080309@cloud9.co.uk> <45E4611C.6060406@redhat.com> Message-ID: <47041.136.145.116.114.1172601158.squirrel@ece.uprm.edu> Hi James, if you want you can review the COMMANDS used by me in order to create distro, then profile (http://pdcsrv.ece.uprm.edu/blog/?p=78) and then how to deploy a virtual machine with koan (http://pdcsrv.ece.uprm.edu/blog/?p=80). Time ago, Michael in private email give a theoretical details about the roles that plays cobbler and koan, but, since i'm a bad student, i can not understand all (sorry Michael, but many thanks for your explaination), however the COMMANDS in those documents, work for me! Hi Michael, i have a question. According to figure attached, there exist a machine where is declared a profile->distro->system. If an user is located to remote server is possible for him execute: koan --replace-self --server=15.20.10.10 --system=databaseserver and then, replace the old system for the new one (databaseserver)? What is the security restrictions to avoid that a malicious user drop a running system? Another one, is not possible execute: koan --virt --server=15.20.10.10 --system=databaseserver where, in databaseserver system is defined the physical host where the virtual machine will be deployed? why is that possible considered? thanks a lot! regards. http://ece.uprm.edu/~s047267 http://del.icio.us/josanabr http://blog-grid.blogspot.com -------------- next part -------------- A non-text attachment was scrubbed... Name: etmgttools.png Type: image/png Size: 19142 bytes Desc: not available URL: From mdehaan at redhat.com Tue Feb 27 19:15:29 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 27 Feb 2007 14:15:29 -0500 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <47041.136.145.116.114.1172601158.squirrel@ece.uprm.edu> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com><45E45D61.8080309@cloud9.co.uk> <45E4611C.6060406@redhat.com> <47041.136.145.116.114.1172601158.squirrel@ece.uprm.edu> Message-ID: <45E48351.3050104@redhat.com> John Sanabria wrote: > Hi James, > > if you want you can review the COMMANDS used by me in order to create > distro, then profile (http://pdcsrv.ece.uprm.edu/blog/?p=78) and then how > to deploy a virtual machine with koan > (http://pdcsrv.ece.uprm.edu/blog/?p=80). > I'm not sure I understand what "distro repository" is in your diagrams as the cobbler "database" is usually on the same box. I see this as: (cobbler server) <----- pulls data from ---- (target system with koan installed) And that's it. Boot configurations are defined on the cobbler server and accessed either via PXE or with koan. > Time ago, Michael in private email give a theoretical details about the > roles that plays cobbler and koan, but, since i'm a bad student, i can not > understand all (sorry Michael, but many thanks for your explaination), > however the COMMANDS in those documents, work for me! > > Hi Michael, i have a question. According to figure attached, there exist a > machine where is declared a profile->distro->system. If an user is located > to remote server is possible for him execute: > > koan --replace-self --server=15.20.10.10 --system=databaseserver > > and then, replace the old system for the new one (databaseserver)? What is > the security restrictions to avoid that a malicious user drop a running > system? > The user in question needs to be logged in to the remote system and have root on it. (Or, if using "cobbler enchant" needs to be in root's authorized_keys). > Another one, is not possible execute: > > koan --virt --server=15.20.10.10 --system=databaseserver > > where, in databaseserver system is defined the physical host where the > virtual machine will be deployed? why is that possible considered? > I'm sorry, I don't understand this question. BTW, best practices for koan suggest using mac addresses for cobbler system names, not hostnames. > thanks a lot! regards. > > http://ece.uprm.edu/~s047267 > http://del.icio.us/josanabr > http://blog-grid.blogspot.com > > > > > > > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------ > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools From John.Sanabria at ece.uprm.edu Tue Feb 27 21:24:07 2007 From: John.Sanabria at ece.uprm.edu (John Sanabria) Date: Tue, 27 Feb 2007 17:24:07 -0400 (AST) Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E48351.3050104@redhat.com> References: <45E43022.1050208@cloud9.co.uk><45E4504E.4000806@redhat.com><45E45D61.8080309@cloud9.co.uk><45E4611C.6060406@redhat.com><47041.136.145.116.114.1172601158.squirrel@ece.uprm.edu> <45E48351.3050104@redhat.com> Message-ID: <55917.136.145.116.114.1172611447.squirrel@ece.uprm.edu> > I'm not sure I understand what "distro repository" is in your diagrams > as the cobbler "database" is usually on the same > box. > yes, i assign a bad name for that element into the figure (http://pdcsrv.ece.uprm.edu/blog/?attachment_id=79), perhaps is better something like Linux Distro. > I see this as: > > (cobbler server) <----- pulls data from ---- (target system with koan > installed) > > And that's it. > > Boot configurations are defined on the cobbler server and accessed > either via PXE or with koan. hmmm!!! ok... i'll review again the koan man page. > The user in question needs to be logged in to the remote system and have > root on it. > (Or, if using "cobbler enchant" needs to be in root's authorized_keys). > ok, i'll try "cobbler enchant"! >> Another one, is not possible execute: >> >> koan --virt --server=15.20.10.10 --system=databaseserver >> >> where, in databaseserver system is defined the physical host where the >> virtual machine will be deployed? why is that possible considered? >> > I'm sorry, I don't understand this question. > ok, i re-wrote something and now i understand my misconception. thanks a lot. http://ece.uprm.edu/~s047267 http://del.icio.us/josanabr http://blog-grid.blogspot.com From james at cloud9.co.uk Tue Feb 27 22:54:03 2007 From: james at cloud9.co.uk (James Fidell) Date: Tue, 27 Feb 2007 22:54:03 +0000 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E4611C.6060406@redhat.com> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com> <45E45D61.8080309@cloud9.co.uk> <45E4611C.6060406@redhat.com> Message-ID: <45E4B68B.9010305@cloud9.co.uk> Michael DeHaan wrote: > Hmm, I think you may be looking for koan to do something else than what > it's designed to do, but I'm not sure what. In which case my lack of understanding has meant that I've expressed myself poorly. > Koan replaces the xm create. It creates the domU image for you, same > as "xm create" or (preferred) virtguest-install would do, except it does > so using a reusable profile that can be controlled server side and > shared with multiple dom0 machines for repeatable results. > Suppose you have already used cobbler to install a datacenter/lab full > of dom0 machines (maybe you installed them with cobbler, and if you did, > you get bonus points). This is exactly what I want to do. And I have done the dom0 install with cobbler, too :) Reading the source for koan, it looks like my confusion may have been because I couldn't see where certain things got configured (such as, say, the location of a file used for the domU virtual disk), when in fact they're (partially) hardcoded. One more question (for the moment :) From the code it appears that koan only creates paravirtual domU instances. Is that correct? If not, how do I get it to create a fully virtual domU instance? James From mdehaan at redhat.com Tue Feb 27 23:24:23 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 27 Feb 2007 18:24:23 -0500 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E4B68B.9010305@cloud9.co.uk> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com> <45E45D61.8080309@cloud9.co.uk> <45E4611C.6060406@redhat.com> <45E4B68B.9010305@cloud9.co.uk> Message-ID: <45E4BDA7.3030908@redhat.com> > > One more question (for the moment :) From the code it appears that koan > only creates paravirtual domU instances. Is that correct? If not, how > do I get it to create a fully virtual domU instance? > > This is correct. koan needs to be ported over to use the same virtinst-devel python library that virtguest-install uses, and when that happens, it will be capable of doing fullvirt installs (and anything else virtinst can install, which can include virtualization technology other than Xen -- I believe QEmu is the one virt-manager is supporting now). I've wanted to do this earlier -- though I doubt that will happen prior to mid March. Should someone be feeling adventurous, patches would definitely be accepted. If not, I'll get around to it when I can -- it's next up on the cobbler/koan feature list and should require modifications to koan only, not cobbler. I'm imagining koan will take a "--fullvirt" flag in addition to it's "--virt" one it takes now. The aformentioned koan change to the Xen images install location to play nice with SELinux should happen this week though, as I'm also due to release a few cobbler bugfixes and it makes sense to go ahead and upgrade koan at that time as well. The only changes to koan will be for the path location and for sorting the output from --list-profiles. Both of these are already in the upstream source on et.redhat.com. > James > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > From mdehaan at redhat.com Tue Feb 27 23:26:40 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 27 Feb 2007 18:26:40 -0500 Subject: [et-mgmt-tools] Xen domU + koan In-Reply-To: <45E4BDA7.3030908@redhat.com> References: <45E43022.1050208@cloud9.co.uk> <45E4504E.4000806@redhat.com> <45E45D61.8080309@cloud9.co.uk> <45E4611C.6060406@redhat.com> <45E4B68B.9010305@cloud9.co.uk> <45E4BDA7.3030908@redhat.com> Message-ID: <45E4BE30.2040503@redhat.com> Michael DeHaan wrote: > >> >> One more question (for the moment :) From the code it appears that koan >> only creates paravirtual domU instances. Is that correct? If not, how >> do I get it to create a fully virtual domU instance? >> >> > > This is correct. > koan needs to be ported over to use the same virtinst-devel python > library that virtguest-install uses, and when that happens, it will be > capable of doing fullvirt installs (and anything else virtinst can > install, which can include virtualization technology other than Xen -- > I believe QEmu is the one virt-manager is supporting now). > > I've wanted to do this earlier -- though I doubt that will happen > prior to mid March. Should someone be feeling adventurous, patches > would definitely be accepted. If not, I'll get around to it when I > can -- it's next up on the cobbler/koan feature list and should > require modifications to koan only, not cobbler. I'm imagining koan > will take a "--fullvirt" flag in addition to it's "--virt" one it > takes now. Actually, what needs to happen is there's a parameter on the profile already for paravirt -- and it's currently being ignored. koan needs to start paying attention to that parameter. So you would not get a new koan parameter, but rather would define the profile differently in cobbler. Anyhow, easy enough :) > > The aformentioned koan change to the Xen images install location to > play nice with SELinux should happen this week though, as I'm also due > to release a few cobbler bugfixes and it makes sense to go ahead and > upgrade koan at that time as well. The only changes to koan will be > for the path location and for sorting the output from > --list-profiles. Both of these are already in the upstream source on > et.redhat.com. > >> James >> >> _______________________________________________ >> et-mgmt-tools mailing list >> et-mgmt-tools at redhat.com >> https://www.redhat.com/mailman/listinfo/et-mgmt-tools >> > > From fj0873gn at aa.jp.fujitsu.com Wed Feb 28 05:47:36 2007 From: fj0873gn at aa.jp.fujitsu.com (Nobuhiro Itou) Date: Wed, 28 Feb 2007 14:47:36 +0900 Subject: [et-mgmt-tools] [PATCH][RESEND] check the VNC port number In-Reply-To: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> References: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> Message-ID: <200702281447.BFC51011.K9O80G4H@aa.jp.fujitsu.com> Hi, Dan Would you give me a comment on this patch? If not, please apply it. I think virt-install should check the VNC port number that the user cannot specify. > When I execute "virt-install --vnc --vncport=5900", domain starts > with 5900 + domain ID. > I think that it is necessary to check, because we cannot appoint > the VNC port number equal to or less than 5900. > > The attached patch adds to check the VNC port number. > > Signed-off-by: Nobuhiro Itou > Thanks, Nobuhiro Itou. Index: virtinst/Guest.py (python-virtinst-0.101.0) =================================================================== --- Guest.py 2007-02-28 23:18:37.000000000 +0900 +++ Guest.py.vncport_check 2007-02-28 14:15:49.000000000 +0900 @@ -179,7 +179,9 @@ class XenGraphics(VirtualGraphics): class VNCVirtualGraphics(XenGraphics): def __init__(self, *args): self.name = "vnc" - if len(args) >= 1 and args[0]: + if len(args) >= 1 and not args[0] is None: + if args[0] < 5901: + raise ValueError, "Invalid value for vncport, vncport must be no less than 5901" self.port = args[0] else: self.port = -1 From fukuta.saori at jp.fujitsu.com Wed Feb 28 08:47:29 2007 From: fukuta.saori at jp.fujitsu.com (Saori Fukuta) Date: Wed, 28 Feb 2007 17:47:29 +0900 Subject: [et-mgmt-tools] [PATCH] virt-install 0.101.0 can't install Message-ID: <20070228174639.7303.FUKUTA.SAORI@jp.fujitsu.com> Hi, I can't install with current virt-install ( version: 0.101.0 ). I think the cause is the difference in following: getting OS type from libvirt : Xen (capital letter) check string at virt-install : xen (small letter) So, I fix the string "xen" to "Xen" and I can install. But, I worry about that libvirt has some return code. e.g.) xen_internal.c ... return("Xen"); xend_internal.c ... return("XenDaemon"); xm_internal.c ... return ("XenXM"); Is that all right? Signed-off-by: Saori Fukuta Thanks, Saori Fukuta. Index: virt-install ( version: 0.101.0 ) =================================================================== --- virt-install 2007-02-28 22:49:27.000000000 +0900 +++ virt-install.new 2007-03-01 02:25:30.000000000 +0900 @@ -371,7 +371,7 @@ def main(): conn = libvirt.open(options.connect) type = None # check to ensure we're really on a xen kernel - if conn.getType() == "xen": + if conn.getType() == "Xen": type = "xen" check_xen() @@ -381,7 +381,7 @@ def main(): # first things first, are we trying to create a fully virt guest? hvm = False - if conn.getType() == "xen": + if conn.getType() == "Xen": if virtinst.util.is_hvm_capable(): hvm = options.fullvirt if hvm is None: From rjones at redhat.com Wed Feb 28 11:07:53 2007 From: rjones at redhat.com (Richard W.M. Jones) Date: Wed, 28 Feb 2007 11:07:53 +0000 Subject: [et-mgmt-tools] virt-manager roadmap In-Reply-To: <77abe410702261858p29931ff6vef6cedf5e8aa8be2@mail.gmail.com> References: <45E3679D.4080503@redhat.com> <77abe410702261858p29931ff6vef6cedf5e8aa8be2@mail.gmail.com> Message-ID: <45E56289.90705@redhat.com> Florian Heigl wrote: > 2007/2/27, Hugh Brock : >> All: I have just posted a new virt-manager roadmap at >> http://virt-manager.org/roadmap.html. It reflects our discussions with >> Fujitsu last week along with other features various developers and >> clients have requested. Please feel free to comment here on the list, or >> post patches to the roadmap as necessary. > > > I desparately miss simultanously connecting to multiple Xen hosts. As Dan says, this is a big, hairy issues which should be done "right". Please take a look at the patch I've been working on[1] and send your comments, ideas and improvements. Rich. [1] https://www.redhat.com/archives/libvir-list/2007-February/msg00209.html -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 "[Negative numbers] darken the very whole doctrines of the equations and make dark of the things which are in their nature excessively obvious and simple" (Francis Maseres FRS, mathematician, 1759) -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/x-pkcs7-signature Size: 3237 bytes Desc: S/MIME Cryptographic Signature URL: From berrange at redhat.com Wed Feb 28 12:11:56 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 28 Feb 2007 12:11:56 +0000 Subject: [et-mgmt-tools] [PATCH][RESEND] check the VNC port number In-Reply-To: <200702281447.BFC51011.K9O80G4H@aa.jp.fujitsu.com> References: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> <200702281447.BFC51011.K9O80G4H@aa.jp.fujitsu.com> Message-ID: <20070228121156.GB8287@redhat.com> On Wed, Feb 28, 2007 at 02:47:36PM +0900, Nobuhiro Itou wrote: > Hi, Dan > > Would you give me a comment on this patch? > If not, please apply it. > > I think virt-install should check the VNC port number that the user cannot specify. Sorry, I missed it first time around. It looks good - although it 5900 is a valid port number to use, so I'll just s/5901/5900/ when applying it. > > > When I execute "virt-install --vnc --vncport=5900", domain starts > > with 5900 + domain ID. > > I think that it is necessary to check, because we cannot appoint > > the VNC port number equal to or less than 5900. > > > > The attached patch adds to check the VNC port number. > > > > Signed-off-by: Nobuhiro Itou > > > > Index: virtinst/Guest.py (python-virtinst-0.101.0) > =================================================================== > --- Guest.py 2007-02-28 23:18:37.000000000 +0900 > +++ Guest.py.vncport_check 2007-02-28 14:15:49.000000000 +0900 > @@ -179,7 +179,9 @@ class XenGraphics(VirtualGraphics): > class VNCVirtualGraphics(XenGraphics): > def __init__(self, *args): > self.name = "vnc" > - if len(args) >= 1 and args[0]: > + if len(args) >= 1 and not args[0] is None: > + if args[0] < 5901: > + raise ValueError, "Invalid value for vncport, vncport must be no less than 5901" > self.port = args[0] > else: > self.port = -1 Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From berrange at redhat.com Wed Feb 28 12:14:53 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 28 Feb 2007 12:14:53 +0000 Subject: [et-mgmt-tools] [PATCH] virt-install 0.101.0 can't install In-Reply-To: <20070228174639.7303.FUKUTA.SAORI@jp.fujitsu.com> References: <20070228174639.7303.FUKUTA.SAORI@jp.fujitsu.com> Message-ID: <20070228121453.GC8287@redhat.com> On Wed, Feb 28, 2007 at 05:47:29PM +0900, Saori Fukuta wrote: > Hi, > > I can't install with current virt-install ( version: 0.101.0 ). > > I think the cause is the difference in following: > getting OS type from libvirt : Xen (capital letter) > check string at virt-install : xen (small letter) > So, I fix the string "xen" to "Xen" and I can install. > > But, I worry about that libvirt has some return code. > e.g.) > xen_internal.c ... return("Xen"); > xend_internal.c ... return("XenDaemon"); > xm_internal.c ... return ("XenXM"); > Is that all right? Not entirely right - although in this scenario at least we wil only ever see the first string returned, because the xen_internal.c takes priority over everything else. > Index: virt-install ( version: 0.101.0 ) > =================================================================== > --- virt-install 2007-02-28 22:49:27.000000000 +0900 > +++ virt-install.new 2007-03-01 02:25:30.000000000 +0900 > @@ -371,7 +371,7 @@ def main(): > conn = libvirt.open(options.connect) > type = None > # check to ensure we're really on a xen kernel > - if conn.getType() == "xen": > + if conn.getType() == "Xen": > type = "xen" > check_xen() > > @@ -381,7 +381,7 @@ def main(): > > # first things first, are we trying to create a fully virt guest? > hvm = False > - if conn.getType() == "xen": > + if conn.getType() == "Xen": > if virtinst.util.is_hvm_capable(): > hvm = options.fullvirt > if hvm is None: I'll have to investigate why I didn't see a failure during testing when I did the initial release, but in principle this is fine. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From rjones at redhat.com Wed Feb 28 13:10:10 2007 From: rjones at redhat.com (Richard W.M. Jones) Date: Wed, 28 Feb 2007 13:10:10 +0000 Subject: [et-mgmt-tools] [PATCH][RESEND] check the VNC port number In-Reply-To: <20070228121156.GB8287@redhat.com> References: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> <200702281447.BFC51011.K9O80G4H@aa.jp.fujitsu.com> <20070228121156.GB8287@redhat.com> Message-ID: <45E57F32.5080900@redhat.com> Daniel P. Berrange wrote: > On Wed, Feb 28, 2007 at 02:47:36PM +0900, Nobuhiro Itou wrote: >> Hi, Dan >> >> Would you give me a comment on this patch? >> If not, please apply it. >> >> I think virt-install should check the VNC port number that the user cannot specify. > > Sorry, I missed it first time around. It looks good - although it 5900 is a > valid port number to use, so I'll just s/5901/5900/ when applying it. Probably a stupid question, but why, other than convention, are ports < 5900 not valid for VNC? Rich. -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 "[Negative numbers] darken the very whole doctrines of the equations and make dark of the things which are in their nature excessively obvious and simple" (Francis Maseres FRS, mathematician, 1759) -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/x-pkcs7-signature Size: 3237 bytes Desc: S/MIME Cryptographic Signature URL: From berrange at redhat.com Wed Feb 28 14:36:03 2007 From: berrange at redhat.com (Daniel P. Berrange) Date: Wed, 28 Feb 2007 14:36:03 +0000 Subject: [et-mgmt-tools] [PATCH][RESEND] check the VNC port number In-Reply-To: <45E57F32.5080900@redhat.com> References: <200702231503.DHG43271.H49G0OK8@aa.jp.fujitsu.com> <200702281447.BFC51011.K9O80G4H@aa.jp.fujitsu.com> <20070228121156.GB8287@redhat.com> <45E57F32.5080900@redhat.com> Message-ID: <20070228143603.GB9555@redhat.com> On Wed, Feb 28, 2007 at 01:10:10PM +0000, Richard W.M. Jones wrote: > Daniel P. Berrange wrote: > >On Wed, Feb 28, 2007 at 02:47:36PM +0900, Nobuhiro Itou wrote: > >>Hi, Dan > >> > >>Would you give me a comment on this patch? > >>If not, please apply it. > >> > >>I think virt-install should check the VNC port number that the user > >>cannot specify. > > > >Sorry, I missed it first time around. It looks good - although it 5900 is a > >valid port number to use, so I'll just s/5901/5900/ when applying it. > > Probably a stupid question, but why, other than convention, are ports < > 5900 not valid for VNC? All the implementations we talk to in libvirt ultimately convert this port number into a display number by subtracting 5900. So if we allowed < 5900, they'd end up -ve and bad stuff would happen. So it seems like a sensible precaution to restrict it in this way, even though the raw RFB protocol has no such restriction. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=| From tforeman at ibsys.com Wed Feb 28 16:29:10 2007 From: tforeman at ibsys.com (Foreman, Tim) Date: Wed, 28 Feb 2007 10:29:10 -0600 Subject: [et-mgmt-tools] Small Cobbler nit to pick Message-ID: <514B41B4EB0E834EAFB17FA73E263B7009B2B4@exch001.ibsys.com> I just upgraded to 0.4.2-0 (from 0.2.8-1 - it's been a while) and noticed that the system name seems to be case sensitive: [root at kickstart ~]# cobbler system remove --name=00:17:A4:8B:BC:29 can't delete something that doesn't exist [root at kickstart ~]# cobbler system remove --name=00:17:a4:8b:bc:29 This seems like it should be case insensitive to make it fall into standards compliance. Why should a MAC address or a hostname care about case? -- Timothy W. Foreman ~ Security Administrator ~ tforeman at ibsys.com (651) 365-4181 ~ Internet Broadcasting ~ www.ibsys.com -- The Onion: Have you decided what you want to be when you grow up? Berkeley Breathed: Dad. The rest is frosting. From mdehaan at redhat.com Wed Feb 28 16:47:23 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 28 Feb 2007 11:47:23 -0500 Subject: [et-mgmt-tools] Small Cobbler nit to pick In-Reply-To: <514B41B4EB0E834EAFB17FA73E263B7009B2B4@exch001.ibsys.com> References: <514B41B4EB0E834EAFB17FA73E263B7009B2B4@exch001.ibsys.com> Message-ID: <45E5B21B.5090905@redhat.com> Foreman, Tim wrote: > I just upgraded to 0.4.2-0 (from 0.2.8-1 - it's been a while) and > noticed that the system name seems to be case sensitive: > > [root at kickstart ~]# cobbler system remove --name=00:17:A4:8B:BC:29 > can't delete something that doesn't exist > [root at kickstart ~]# cobbler system remove --name=00:17:a4:8b:bc:29 > > This seems like it should be case insensitive to make it fall into > standards compliance. > > Why should a MAC address or a hostname care about case? > -- > Timothy W. Foreman ~ Security Administrator ~ tforeman at ibsys.com > (651) 365-4181 ~ Internet Broadcasting ~ www.ibsys.com > -- > The Onion: Have you decided what you want to be when you grow up? > Berkeley Breathed: Dad. The rest is frosting. > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > You're right, it shouldn't. As a backwards compatibility note, I'm guessing no one is using profiles or distros that differ only by case. If you are, speak up! :) --Michael From David.Mackintosh at xdroop.com Wed Feb 28 20:32:37 2007 From: David.Mackintosh at xdroop.com (David Mackintosh) Date: Wed, 28 Feb 2007 15:32:37 -0500 Subject: [et-mgmt-tools] Typo on cobbler page Message-ID: <20070228203237.GE26058@xdroop.com> Second instruction, "Installation/Build For RHEL4" wget http://www.python.org/pyvault/centos-4-i386/python-cheetah-0-0.9.18-1.el4.pyv.rpm ...should be... wget http://www.python.org/pyvault/centos-4-i386/python-cheetah-0.9.18-1.el4.pyv.rpm ...? -- /\oo/\ / /()\ \ David Mackintosh | dave at xdroop.com | http://www.xdroop.com -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From mdehaan at redhat.com Wed Feb 28 20:38:00 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 28 Feb 2007 15:38:00 -0500 Subject: [et-mgmt-tools] Typo on cobbler page In-Reply-To: <20070228203237.GE26058@xdroop.com> References: <20070228203237.GE26058@xdroop.com> Message-ID: <45E5E828.7020702@redhat.com> David Mackintosh wrote: > Second instruction, "Installation/Build For RHEL4" > > wget http://www.python.org/pyvault/centos-4-i386/python-cheetah-0-0.9.18-1.el4.pyv.rpm > > ...should be... > > wget http://www.python.org/pyvault/centos-4-i386/python-cheetah-0.9.18-1.el4.pyv.rpm > > ...? > > > ------------------------------------------------------------------------ > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools Yes. Fixed. Thanks! From David.Mackintosh at xdroop.com Wed Feb 28 20:51:03 2007 From: David.Mackintosh at xdroop.com (David Mackintosh) Date: Wed, 28 Feb 2007 15:51:03 -0500 Subject: [et-mgmt-tools] Typo on cobbler page In-Reply-To: <45E5E828.7020702@redhat.com> References: <20070228203237.GE26058@xdroop.com> <45E5E828.7020702@redhat.com> Message-ID: <20070228205103.GF26058@xdroop.com> Possible bug: when installing 0.4.2 using the above instructions, I don't get a sample /var/lib/cobbler/settings file? # ls -l /var/lib/cobbler/settings ls: /var/lib/cobbler/settings: No such file or directory -- /\oo/\ / /()\ \ David Mackintosh | dave at xdroop.com | http://www.xdroop.com -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From tforeman at ibsys.com Wed Feb 28 20:58:05 2007 From: tforeman at ibsys.com (Foreman, Tim) Date: Wed, 28 Feb 2007 14:58:05 -0600 Subject: [et-mgmt-tools] Cobbler v0.4.2-0 templating not working? Message-ID: <514B41B4EB0E834EAFB17FA73E263B7009B2BA@exch001.ibsys.com> Did templating change or break between 0.2.8-1 and 0.4.2-0? I can't get the templating to work anymore. My kickstart file has entries like this in it: INSTALL_ENV=TEMPLATE::env When I add a system I use the --ksmeta option - like this: cobbler system add --name=00:17:A4:3D:B4:5D --profile=base_test \ --ksmeta="env=mtc hostname=ibsqa-xen ip0=10.9.90.10 mask0=255.255.0.0 gw=10.10.0.3 ip1=10.10.90.10 mask1=255.255.0.0" After I sync, the kickstart file the template line looks like this: INSTALL_ENV=$env And when I do a report --systems, the ksmeta data doesn't show up: system : 00:17:A4:3D:B4:5D profile : base_test kernel options : {} ks metadata : {} pxe address : Where is the ksmeta data going? -- Timothy W. Foreman ~ Security Administrator ~ tforeman at ibsys.com (651) 365-4181 ~ Internet Broadcasting ~ www.ibsys.com -- The Onion: Have you decided what you want to be when you grow up? Berkeley Breathed: Dad. The rest is frosting. From mdehaan at redhat.com Wed Feb 28 21:10:46 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 28 Feb 2007 16:10:46 -0500 Subject: [et-mgmt-tools] Typo on cobbler page In-Reply-To: <20070228205103.GF26058@xdroop.com> References: <20070228203237.GE26058@xdroop.com> <45E5E828.7020702@redhat.com> <20070228205103.GF26058@xdroop.com> Message-ID: <45E5EFD6.8030705@redhat.com> David Mackintosh wrote: > Possible bug: when installing 0.4.2 using the above instructions, I don't > get a sample /var/lib/cobbler/settings file? > > # ls -l /var/lib/cobbler/settings > ls: /var/lib/cobbler/settings: No such file or directory > Ok, cobbler settings, like the rest of the cobbler "database" files, are generated at first use when they don't exist. "cobbler check" should generate them. > > ------------------------------------------------------------------------ > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools From mdehaan at redhat.com Wed Feb 28 21:19:45 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 28 Feb 2007 16:19:45 -0500 Subject: [et-mgmt-tools] Cobbler v0.4.2-0 templating not working? In-Reply-To: <514B41B4EB0E834EAFB17FA73E263B7009B2BA@exch001.ibsys.com> References: <514B41B4EB0E834EAFB17FA73E263B7009B2BA@exch001.ibsys.com> Message-ID: <45E5F1F1.3070803@redhat.com> Foreman, Tim wrote: > Did templating change or break between 0.2.8-1 and 0.4.2-0? > > I can't get the templating to work anymore. > > My kickstart file has entries like this in it: > INSTALL_ENV=TEMPLATE::env > > When I add a system I use the --ksmeta option - like this: > > cobbler system add --name=00:17:A4:3D:B4:5D --profile=base_test \ > --ksmeta="env=mtc hostname=ibsqa-xen ip0=10.9.90.10 mask0=255.255.0.0 > gw=10.10.0.3 ip1=10.10.90.10 mask1=255.255.0.0" > > After I sync, the kickstart file the template line looks like this: > INSTALL_ENV=$env > > And when I do a report --systems, the ksmeta data doesn't show up: > system : 00:17:A4:3D:B4:5D > profile : base_test > kernel options : {} > ks metadata : {} > pxe address : > > Where is the ksmeta data going? > -- > Timothy W. Foreman ~ Security Administrator ~ tforeman at ibsys.com > (651) 365-4181 ~ Internet Broadcasting ~ www.ibsys.com > -- > The Onion: Have you decided what you want to be when you grow up? > Berkeley Breathed: Dad. The rest is frosting. > > _______________________________________________ > et-mgmt-tools mailing list > et-mgmt-tools at redhat.com > https://www.redhat.com/mailman/listinfo/et-mgmt-tools > This is definitely bug. Apparently the "item.py" source code now splits ksmeta parameters on "," not on whitespace as it did in previous versions. I am changing 0.4.3 back to split on whitespace in accordance with the manpage. If you use "," to seperate the items for now, you /should/ see them entered correctly, and on the upgrade, they should still be there. Cobbler is now storing internal datastructures as hashes and lists, rather than strings, and that's how this bug was introduced. The parsing function was fed in the wrong delimiter. The reason for changing the storage format is now much more data can be fed into the templating engine (especially if using the API and not the command line) where before it was hard to express strings that contained newlines and longer passages of text, like an SSH key. Basically it makes the YAML files in /var/lib/shadowmanager easier to edit and modify. diff -r 78521aa0ef20 cobbler/item.py --- a/cobbler/item.py Tue Feb 27 14:46:54 2007 -0500 +++ b/cobbler/item.py Wed Feb 28 16:16:14 2007 -0500 @@ -44,7 +44,7 @@ class Item(serializable.Serializable): The meta tags are used as input to the templating system to preprocess kickstart files """ - (success, value) = utils.input_string_or_hash(options,",") + (success, value) = utils.input_string_or_hash(options,None) if not success: return False else: --Michael From tforeman at ibsys.com Wed Feb 28 21:25:37 2007 From: tforeman at ibsys.com (Foreman, Tim) Date: Wed, 28 Feb 2007 15:25:37 -0600 Subject: [et-mgmt-tools] Cobbler v0.4.2-0 templating not working? Message-ID: <514B41B4EB0E834EAFB17FA73E263B7009B2BB@exch001.ibsys.com> > -----Original Message----- > From: et-mgmt-tools-bounces at redhat.com > [mailto:et-mgmt-tools-bounces at redhat.com]On Behalf Of Michael DeHaan > > This is definitely bug. Apparently the "item.py" source code now > splits ksmeta parameters on "," not on whitespace as it did > in previous > versions. > I am changing 0.4.3 back to split on whitespace in accordance > with the > manpage. > > If you use "," to seperate the items for now, you /should/ see them > entered correctly, and on the upgrade, they should still be there. > Cobbler is now storing internal datastructures as hashes and lists, > rather than strings, and that's how this bug was introduced. The > parsing function was fed in the wrong delimiter. Thank you. Using "," for a separator worked. I'll look forward to getting 0.4.3. :-) --Tim From mdehaan at redhat.com Wed Feb 28 22:46:25 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 28 Feb 2007 17:46:25 -0500 Subject: [et-mgmt-tools] Released -- Cobbler 0.4.3, Koan 0.2.6 Message-ID: <45E60641.5060209@redhat.com> Primarily bugfixes. Cobbler 0.4.3 - Added netboot_enabled option for systems to control install loops in programmatic context. (you can ignore this, it's not surfaced in the CLI) - Disabling anchors in YAML serialization (which make files harder to edit) - Fix bug in ksmeta argument processing, takes whitespace again, not commas - Fix bug in old-style deserialization with str and int concatenation Koan 0.2.6 - Store images in a SELinux friendly path - the output of --list-profiles is now sorted Out on et.redhat.com now and waiting on the Fedora mirrors... --Michael