From dan at half-asleep.com Mon Aug 7 15:33:06 2006 From: dan at half-asleep.com (Daniel Segall) Date: 7 Aug 2006 11:33:06 -0400 Subject: Integration with another deployment system Message-ID: <7rJ0qEBf.1154964786.5783280.worlock@half-asleep.com> Hello all, I've just switched jobs, and inherited a mess of an atliris express deployment solution. While it works well for windoze installs, I find it to be more trouble than it's worth for linux installs. I'd like to setup another kickstart server here (which I have no problems doing), but I'm not sure how to pass a windoze job to the altiris server from a pxelinux.cfg. Does anyone have any pointers on how to pass a windoze install from a linux kickstart server to a windoze deployment server? Thanks -Dan From skvidal at linux.duke.edu Mon Aug 7 15:35:32 2006 From: skvidal at linux.duke.edu (seth vidal) Date: Mon, 07 Aug 2006 11:35:32 -0400 Subject: Integration with another deployment system In-Reply-To: <7rJ0qEBf.1154964786.5783280.worlock@half-asleep.com> References: <7rJ0qEBf.1154964786.5783280.worlock@half-asleep.com> Message-ID: <1154964932.9077.7.camel@cutter> On Mon, 2006-08-07 at 11:33 -0400, Daniel Segall wrote: > Hello all, > > I've just switched jobs, and inherited a mess of an atliris express > deployment solution. While it works well for windoze installs, I find it > to be more trouble than it's worth for linux installs. > > I'd like to setup another kickstart server here (which I have no > problems doing), but I'm not sure how to pass a windoze job to the > altiris server from a pxelinux.cfg. Does anyone have any pointers on how > to pass a windoze install from a linux kickstart server to a windoze > deployment server? > the altiris server here just uses: label altiris kernel altiris/linux append initrd=altiris/rootfs.gz root=/dev/ram0 rw ramdisk_size=32768 vga=extended and it seems to work fine for it from our linux-based pxe/tftp server -sv From dan at half-asleep.com Mon Aug 7 15:42:04 2006 From: dan at half-asleep.com (Daniel Segall) Date: 7 Aug 2006 11:42:04 -0400 Subject: Integration with another deployment system In-Reply-To: <1154964932.9077.7.camel@cutter> Message-ID: On 8/7/2006, "seth vidal" wrote: >On Mon, 2006-08-07 at 11:33 -0400, Daniel Segall wrote: >> Hello all, >> >> I've just switched jobs, and inherited a mess of an atliris express >> deployment solution. While it works well for windoze installs, I find it >> to be more trouble than it's worth for linux installs. >> >> I'd like to setup another kickstart server here (which I have no >> problems doing), but I'm not sure how to pass a windoze job to the >> altiris server from a pxelinux.cfg. Does anyone have any pointers on how >> to pass a windoze install from a linux kickstart server to a windoze >> deployment server? >> > >the altiris server here just uses: > > >label altiris > kernel altiris/linux > append initrd=altiris/rootfs.gz root=/dev/ram0 rw ramdisk_size=32768 >vga=extended > > >and it seems to work fine for it from our linux-based pxe/tftp server > >-sv > Thanks for the quick response Seth. What is actually contained in the kernel & initrd image? Meaning,... how does the server know where to go? Thanks, -Dan From skvidal at linux.duke.edu Mon Aug 7 15:43:47 2006 From: skvidal at linux.duke.edu (seth vidal) Date: Mon, 07 Aug 2006 11:43:47 -0400 Subject: Integration with another deployment system In-Reply-To: References: Message-ID: <1154965427.9077.12.camel@cutter> On Mon, 2006-08-07 at 11:42 -0400, Daniel Segall wrote: > On 8/7/2006, "seth vidal" wrote: > > >On Mon, 2006-08-07 at 11:33 -0400, Daniel Segall wrote: > >> Hello all, > >> > >> I've just switched jobs, and inherited a mess of an atliris express > >> deployment solution. While it works well for windoze installs, I find it > >> to be more trouble than it's worth for linux installs. > >> > >> I'd like to setup another kickstart server here (which I have no > >> problems doing), but I'm not sure how to pass a windoze job to the > >> altiris server from a pxelinux.cfg. Does anyone have any pointers on how > >> to pass a windoze install from a linux kickstart server to a windoze > >> deployment server? > >> > > > >the altiris server here just uses: > > > > > >label altiris > > kernel altiris/linux > > append initrd=altiris/rootfs.gz root=/dev/ram0 rw ramdisk_size=32768 > >vga=extended > > > > > >and it seems to work fine for it from our linux-based pxe/tftp server > > > >-sv > > > > Thanks for the quick response Seth. What is actually contained in the > kernel & initrd image? Meaning,... how does the server know where to go? > it's the kernel and initrd that we pulled off of the altiris server running the intel pxe server. We just copied them over to our other pxe server. -sv From Elizabeth.Brosch at thomson.com Thu Aug 10 20:02:40 2006 From: Elizabeth.Brosch at thomson.com (Elizabeth.Brosch at thomson.com) Date: Thu, 10 Aug 2006 16:02:40 -0400 Subject: Kickstart fails mounting nfs source Message-ID: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D1@tshuspaphimbx01.ERF.THOMSON.COM> All: I have some old HP DL140s that I need to install. The NICs on these servers do not support PXE therefore I have to boot from CDROM. Once booted I key in "ks" and the server obtains its network information from DHCP but then immediately fails. I see an error on the console" mount: bad file number". When I hit F3 I see that the real error is "failed to mount nfs source". Has anyone successfully kickstarted a server by using NFS instead of PXE? Thanks in advance. Liz Brosch Thomson Scientific & Healthcare TSH Systems Services Philadelphia, PA 19104 (215) 823-6144 office (267) 784-9166 cell -------------- next part -------------- An HTML attachment was scrubbed... URL: From Chip.Shabazian at bankofamerica.com Thu Aug 10 20:28:59 2006 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Thu, 10 Aug 2006 13:28:59 -0700 Subject: Kickstart fails mounting nfs source In-Reply-To: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D1@tshuspaphimbx01.ERF.THOMSON.COM> Message-ID: What version/update are you building from? Can you access the nfs mount if installing manually (not doing ks on the boot line)? ________________________________ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Elizabeth.Brosch at thomson.com Sent: Thursday, August 10, 2006 1:03 PM To: kickstart-list at redhat.com Subject: Kickstart fails mounting nfs source All: I have some old HP DL140s that I need to install. The NICs on these servers do not support PXE therefore I have to boot from CDROM. Once booted I key in "ks" and the server obtains its network information from DHCP but then immediately fails. I see an error on the console" mount: bad file number". When I hit F3 I see that the real error is "failed to mount nfs source". Has anyone successfully kickstarted a server by using NFS instead of PXE? Thanks in advance. Liz Brosch Thomson Scientific & Healthcare TSH Systems Services Philadelphia, PA 19104 (215) 823-6144 office (267) 784-9166 cell -------------- next part -------------- An HTML attachment was scrubbed... URL: From Elizabeth.Brosch at Thomson.com Thu Aug 10 20:33:41 2006 From: Elizabeth.Brosch at Thomson.com (Elizabeth.Brosch at Thomson.com) Date: Thu, 10 Aug 2006 16:33:41 -0400 Subject: Kickstart fails mounting nfs source Message-ID: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D6@tshuspaphimbx01.ERF.THOMSON.COM> Hi Chip. I am building from RHEL v4 Update 3. Also, unfortunately I do not have another server in this VLAN that I can test the NFS from. I have 12 servers in this VLAN. The first one I manually installed then configured for kickstart/dhcp/tftp. I am booting from CD #1 for the second server and entering "ks". Liz Brosch ________________________________ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Shabazian, Chip Sent: Thursday, August 10, 2006 4:29 PM To: Discussion list about Kickstart Subject: RE: Kickstart fails mounting nfs source What version/update are you building from? Can you access the nfs mount if installing manually (not doing ks on the boot line)? -------------- next part -------------- An HTML attachment was scrubbed... URL: From imarks at comcast.net Thu Aug 10 22:51:34 2006 From: imarks at comcast.net (Ian) Date: Thu, 10 Aug 2006 18:51:34 -0400 Subject: Kickstart fails mounting nfs source In-Reply-To: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D6@tshuspaphimbx01.ERF.THOMSON.COM> References: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D6@tshuspaphimbx01.ERF.THOMSON.COM> Message-ID: <44DBB876.3090207@comcast.net> Are you able to run tcpdump/snoop on the nfs server to see if the kickstart process attempts to connect. You could test the NFS mounts via booting Disk 1 into rescue mode and manually attempting to mount, or by starting the install with the "linux askmethod" mode and then verifying the mount points by selecting an install via NFS. Ian Elizabeth.Brosch at Thomson.com wrote: > Hi Chip. > > I am building from RHEL v4 Update 3. Also, unfortunately I do not > have another server in this VLAN that I can test the NFS from. I have > 12 servers in this VLAN. The first one I manually installed then > configured for kickstart/dhcp/tftp. > > I am booting from CD #1 for the second server and entering "ks". > > Liz Brosch > > ------------------------------------------------------------------------ > *From:* kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com] *On Behalf Of *Shabazian, Chip > *Sent:* Thursday, August 10, 2006 4:29 PM > *To:* Discussion list about Kickstart > *Subject:* RE: Kickstart fails mounting nfs source > > What version/update are you building from? > > Can you access the nfs mount if installing manually (not doing ks on > the boot line)? > > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From dag at wieers.com Fri Aug 11 05:47:42 2006 From: dag at wieers.com (Dag Wieers) Date: Fri, 11 Aug 2006 07:47:42 +0200 (CEST) Subject: Kickstart fails mounting nfs source In-Reply-To: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D6@tshuspaphimbx01.ERF.THOMSON.COM> References: <2213CAA79DA9094AB52517FDADEF07FA03D8C3D6@tshuspaphimbx01.ERF.THOMSON.COM> Message-ID: On Thu, 10 Aug 2006, Elizabeth.Brosch at Thomson.com wrote: > On Thu, 10 Aug 2006, Chip Shabazian wrote: > > > What version/update are you building from? > > > > Can you access the nfs mount if installing manually (not doing ks on > > the boot line)? > > I am building from RHEL v4 Update 3. Also, unfortunately I do not have > another server in this VLAN that I can test the NFS from. I have 12 > servers in this VLAN. The first one I manually installed then > configured for kickstart/dhcp/tftp. > > I am booting from CD #1 for the second server and entering "ks". I find HTTP much easier to use and debug. It might be slightly different now that NFS is using TCP. But only having to ask for tcp/80 to be opened in one direction and being able to check a log-file for troubleshooting is worth my time :) The same firewall rule is then used for installing updates using apt (although in some environments we set up a remote tunnel using SSH). Kind regards, -- dag wieers, dag at wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power] From Ajitabh.Pandey at bds.tv Fri Aug 11 09:22:25 2006 From: Ajitabh.Pandey at bds.tv (Ajitabh Pandey) Date: Fri, 11 Aug 2006 10:22:25 +0100 Subject: Kickstart fails mounting nfs source Message-ID: RHEL4 update 3 (both amd-64 and i386) works fine using kickstart over http. Installed several servers so far. But you need to keep one thing in mind if you are dealing with HP servers with RAID controllers (though DL140s I don't think have RAID controllers), refer to the following bug: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=187093 This is still true and reproducable every time I want to reproduce it. Surprisingly there is no update on it so far. Regards. Ajitabh Pandey -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Dag Wieers Sent: 11 August 2006 06:48 To: Discussion list about Kickstart Subject: RE: Kickstart fails mounting nfs source On Thu, 10 Aug 2006, Elizabeth.Brosch at Thomson.com wrote: > On Thu, 10 Aug 2006, Chip Shabazian wrote: > > > What version/update are you building from? > > > > Can you access the nfs mount if installing manually (not doing ks on > > the boot line)? > > I am building from RHEL v4 Update 3. Also, unfortunately I do not > have another server in this VLAN that I can test the NFS from. I have > 12 servers in this VLAN. The first one I manually installed then > configured for kickstart/dhcp/tftp. > > I am booting from CD #1 for the second server and entering "ks". I find HTTP much easier to use and debug. It might be slightly different now that NFS is using TCP. But only having to ask for tcp/80 to be opened in one direction and being able to check a log-file for troubleshooting is worth my time :) The same firewall rule is then used for installing updates using apt (although in some environments we set up a remote tunnel using SSH). Kind regards, -- dag wieers, dag at wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power] _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list This email is confidential and intended solely for the use of the individual to whom it is addressed. Any views or opinions presented are solely those of the author and do not necessarily represent those of Broadcasting Dataservices. If you are not the intended recipient, be advised that you have received this email in error and that any use, dissemination, forwarding, printing, or copying of this email is strictly prohibited. If you have received this email in error please notify the sender From Elizabeth.Brosch at thomson.com Fri Aug 11 15:07:09 2006 From: Elizabeth.Brosch at thomson.com (Elizabeth.Brosch at thomson.com) Date: Fri, 11 Aug 2006 11:07:09 -0400 Subject: Kickstart fails mounting nfs source In-Reply-To: Message-ID: <2213CAA79DA9094AB52517FDADEF07FA03D8C3E8@tshuspaphimbx01.ERF.THOMSON.COM> Thanks to everyone who took the time to reply. Seems I didn't have nfs running on server!!! Thanks for the tips from everyone. From tmafs at us.ibm.com Mon Aug 21 19:05:37 2006 From: tmafs at us.ibm.com (Tom Miller) Date: Mon, 21 Aug 2006 14:05:37 -0500 Subject: Anaconda kickstart options and lowlevel netboot.img input options. Message-ID: I am using a kickstart.cfg file to load power 5 lpars via network install. I use a bootp server to deliver the RedHat netboot.img to the victim machine, I set the boot-file SMS env variable to pass boot image parms to netboot.img. Here's what I pass: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/sift22104.cfg This lets the netboot.img config eth0 to access the ks file on the server. I have this network line in the /lpars/sift22104.cfg kickstart file: network --device eth0 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com This works just fine and the install completes. I need to make this work on lpars that have never been loaded before and these lpars can have several eth adapters so the user will not know what order the adatpers will be configured so he can't specify eth0 or eth1 in the options above. I have expect code that goes thru the SMS menus and attempts to ping the server on each eth adatper until it finds one that is connected: I can pull the mac address for that card and could at that point create the netboot.img boot parms line that includes the mac address, but there is no way I can know what ethX redhat will confgure this card as. I tried changing the netboot.img parms line to: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth-id-2a:ca:30:00:40:02 ks=nfs:9.3.80.16:/lpars/sift22104.cfg and the kickstart network line to this: network --device eth-id-2a:ca:30:00:40:02 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com But it did not work.. I got a syntax error back from anaconda complaining about this: self.setNetwork(id, bootProto, ip, netmask, ethtool, device=device, onboot=onboot, dhcpclass=dhcpclass, essid=essid, wepkey=wepkey) File "/usr/lib/anaconda/installclass.py", line 332, in setNetwork dev = devices[device] KeyError: 'eth-id-2a:ca:30:00:40:02' I have searched thru doc's till I am blue and can't find out if there is some way (any correct syntax) to pass the mac address to netboot.img and anaconda instead of the ethX name. Dose anyone know how this can be done, is it even in the netboot.img/anaconda code to all this and deal with the mac address? This is the last thing I need to fully automate these installs for my users with kickstart instead of using expect to fill out all those changing install menus in text mode. thanks tom miller IBM. -------------- next part -------------- An HTML attachment was scrubbed... URL: From Colin.Coe at woodside.com.au Tue Aug 22 00:50:11 2006 From: Colin.Coe at woodside.com.au (Coe, Colin C.) Date: Tue, 22 Aug 2006 08:50:11 +0800 Subject: Anaconda kickstart options and lowlevel netboot.img input options. Message-ID: <17EBC05307239C4896ED25244F42A80101367AD4@permls05.wde.woodside.com.au> If you know the MAC addr, could you use a DHCP server with an entry for each LPAR you want to install? Something like: allow booting; allow bootp; ... group { subnet 9.3.22.0 netmask 255.255.255.0 { option domain-name "ibm.com"; option subnet-mask 255.255.255.0; option broadcast-address 9.3.22.255; option domain-name-servers 9.0..7.1; option routers 9.3.22.1; } # PXE-specific configuration directives... next-server 9.3.22.253; filename "pxelinux.0"; ... host lpar02 { hardware ethernet 2A:CA:30:00:40:02; fixed-address 9.3.22.202; option host-name "lpar02"; } ... } Just a thought. Also, if you're using EL4, beware that sometimes the NIC that the installer decides is eth0 is not what the installed kernel thinks is eth0. CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Tuesday, 22 August 2006 3:06 AM To: kickstart-list at redhat.com Subject: Anaconda kickstart options and lowlevel netboot.img input options. I am using a kickstart.cfg file to load power 5 lpars via network install. I use a bootp server to deliver the RedHat netboot.img to the victim machine, I set the boot-file SMS env variable to pass boot image parms to netboot.img. Here's what I pass: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/sift22104.cfg This lets the netboot.img config eth0 to access the ks file on the server. I have this network line in the /lpars/sift22104.cfg kickstart file: network --device eth0 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com This works just fine and the install completes. I need to make this work on lpars that have never been loaded before and these lpars can have several eth adapters so the user will not know what order the adatpers will be configured so he can't specify eth0 or eth1 in the options above. I have expect code that goes thru the SMS menus and attempts to ping the server on each eth adatper until it finds one that is connected: I can pull the mac address for that card and could at that point create the netboot.img boot parms line that includes the mac address, but there is no way I can know what ethX redhat will confgure this card as. I tried changing the netboot.img parms line to: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth-id-2a:ca:30:00:40:02 ks=nfs:9.3.80.16:/lpars/sift22104.cfg and the kickstart network line to this: network --device eth-id-2a:ca:30:00:40:02 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com But it did not work.. I got a syntax error back from anaconda complaining about this: self.setNetwork(id, bootProto, ip, netmask, ethtool, device=device, onboot=onboot, dhcpclass=dhcpclass, essid=essid, wepkey=wepkey) File "/usr/lib/anaconda/installclass.py", line 332, in setNetwork dev = devices[device] KeyError: 'eth-id-2a:ca:30:00:40:02' I have searched thru doc's till I am blue and can't find out if there is some way (any correct syntax) to pass the mac address to netboot.img and anaconda instead of the ethX name. Dose anyone know how this can be done, is it even in the netboot.img/anaconda code to all this and deal with the mac address? This is the last thing I need to fully automate these installs for my users with kickstart instead of using expect to fill out all those changing install menus in text mode. thanks tom miller IBM. -------------- next part -------------- An HTML attachment was scrubbed... URL: From tmafs at us.ibm.com Tue Aug 22 18:09:19 2006 From: tmafs at us.ibm.com (Tom Miller) Date: Tue, 22 Aug 2006 13:09:19 -0500 Subject: Anaconda kickstart options and lowlevel netboot.img input options. Message-ID: Collin, Thanks for your reply. I would love to use dhcp, this would solve all my problems but IBM prevents me from doing that. They do not pass dhcp requests thru the site bridges/gateways/routes so I would have to have a dhcp server on each subnet that contained victims to install. I have 5 servers in Austin, one in Rochester, Mn and one in Poughkeepsie, NY and they install 100's of systems a week on many different subnets. I did not realize that the installer and kernel couldn't even agree on which nic was eth0. Thats all the more need for anacondo and the netboot.img to deal with mac addresses. You don't happen to be an anaconda coder do you? thanks tom Monday, August 21, 2006 7:51 PM To: "Discussion list about Kickstart" cc: From: "Coe, Colin C." Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. If you know the MAC addr, could you use a DHCP server with an entry for each LPAR you want to install? Something like: allow booting; allow bootp; ... group { subnet 9.3.22.0 netmask 255.255.255.0 { option domain-name "ibm.com"; option subnet-mask 255.255.255.0; option broadcast-address 9.3.22.255; option domain-name-servers 9.0..7.1; option routers 9.3.22.1; } # PXE-specific configuration directives... next-server 9.3.22.253; filename "pxelinux.0"; ... host lpar02 { hardware ethernet 2A:CA:30:00:40:02; fixed-address 9.3.22.202; option host-name "lpar02"; } ... } Just a thought. Also, if you're using EL4, beware that sometimes the NIC that the installer decides is eth0 is not what the installed kernel thinks is eth0. CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Tuesday, 22 August 2006 3:06 AM To: kickstart-list at redhat.com Subject: Anaconda kickstart options and lowlevel netboot.img input options. I am using a kickstart.cfg file to load power 5 lpars via network install. I use a bootp server to deliver the RedHat netboot.img to the victim machine, I set the boot-file SMS env variable to pass boot image parms to netboot.img. Here's what I pass: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/sift22104.cfg This lets the netboot.img config eth0 to access the ks file on the server. I have this network line in the /lpars/sift22104.cfg kickstart file: network --device eth0 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com This works just fine and the install completes. I need to make this work on lpars that have never been loaded before and these lpars can have several eth adapters so the user will not know what order the adatpers will be configured so he can't specify eth0 or eth1 in the options above. I have expect code that goes thru the SMS menus and attempts to ping the server on each eth adatper until it finds one that is connected: I can pull the mac address for that card and could at that point create the netboot.img boot parms line that includes the mac address, but there is no way I can know what ethX redhat will confgure this card as. I tried changing the netboot.img parms line to: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth-id-2a:ca:30:00:40:02 ks=nfs:9.3.80.16:/lpars/sift22104.cfg and the kickstart network line to this: network --device eth-id-2a:ca:30:00:40:02 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com But it did not work.. I got a syntax error back from anaconda complaining about this: self.setNetwork(id, bootProto, ip, netmask, ethtool, device=device, onboot=onboot, dhcpclass=dhcpclass, essid=essid, wepkey=wepkey) File "/usr/lib/anaconda/installclass.py", line 332, in setNetwork dev = devices[device] KeyError: 'eth-id-2a:ca:30:00:40:02' I have searched thru doc's till I am blue and can't find out if there is some way (any correct syntax) to pass the mac address to netboot.img and anaconda instead of the ethX name. Dose anyone know how this can be done, is it even in the netboot.img/anaconda code to all this and deal with the mac address? This is the last thing I need to fully automate these installs for my users with kickstart instead of using expect to fill out all those changing install menus in text mode. thanks tom miller IBM. -------------- next part -------------- An HTML attachment was scrubbed... URL: From Colin.Coe at woodside.com.au Tue Aug 22 23:29:07 2006 From: Colin.Coe at woodside.com.au (Coe, Colin C.) Date: Wed, 23 Aug 2006 07:29:07 +0800 Subject: Anaconda kickstart options and lowlevel netboot.img input options. Message-ID: <17EBC05307239C4896ED25244F42A80101367AE5@permls05.wde.woodside.com.au> Hi Tom No, I'm not an Anaconda coder. There are quirks in the installer but the issue with the swapped NICs I believe is a kernel compile config quirk rather than an Anaconda problem. Pity about the DHCP. As I said, I'm not an Anaconda or Python coder, but what about a custom install class? Is this were a custom install class would be used? Also, given your working with Power5 LPARs I can only assume you're working with a beastie along the lines of a P595, maybe you could query the hypervisor? I know these are pretty locked down but... Just out of curiosity, what boot method do these use (bootp/pxe/etc)? If its PXE and you're able to use PXELINUX (I don't know if you can), you might be able to make use of the IPAPPEND option. Or better yet, you could have your expect script create a symlink called the MAC address of the NIC you want to use which points to a standard pxelinux config file. For example 2A:CA:30:00:40:02 /tftpboot/pxelinux.cfg/2a-ca-30-00-40-02 -> el4_lpar02 /tftpboot/pxelinux.cfg/el4_lpar02 Where el4_lpar02 would contain something like label el4_lpar kernel el4u4/vmlinuz append ksdevice=eth0 ks=nfs:9.3.80.16:/export/install/ks.cfg initrd=el4u4/initrd.img ip=9.3.22.202 method=nfs:10.240.34.254:/export/install/el4u4_x86_64 CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Wednesday, 23 August 2006 2:09 AM To: Discussion list about Kickstart Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. Collin, Thanks for your reply. I would love to use dhcp, this would solve all my problems but IBM prevents me from doing that. They do not pass dhcp requests thru the site bridges/gateways/routes so I would have to have a dhcp server on each subnet that contained victims to install. I have 5 servers in Austin, one in Rochester, Mn and one in Poughkeepsie, NY and they install 100's of systems a week on many different subnets. I did not realize that the installer and kernel couldn't even agree on which nic was eth0. Thats all the more need for anacondo and the netboot.img to deal with mac addresses. You don't happen to be an anaconda coder do you? thanks tom Monday, August 21, 2006 7:51 PM To: "Discussion list about Kickstart" cc: From: "Coe, Colin C." Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. If you know the MAC addr, could you use a DHCP server with an entry for each LPAR you want to install? Something like: allow booting; allow bootp; ... group { subnet 9.3.22.0 netmask 255.255.255.0 { option domain-name "ibm.com"; option subnet-mask 255.255.255.0; option broadcast-address 9.3.22.255; option domain-name-servers 9.0..7.1; option routers 9.3.22.1; } # PXE-specific configuration directives... next-server 9.3.22.253; filename "pxelinux.0"; ... host lpar02 { hardware ethernet 2A:CA:30:00:40:02; fixed-address 9.3.22.202; option host-name "lpar02"; } ... } Just a thought. Also, if you're using EL4, beware that sometimes the NIC that the installer decides is eth0 is not what the installed kernel thinks is eth0. CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Tuesday, 22 August 2006 3:06 AM To: kickstart-list at redhat.com Subject: Anaconda kickstart options and lowlevel netboot.img input options. I am using a kickstart.cfg file to load power 5 lpars via network install. I use a bootp server to deliver the RedHat netboot.img to the victim machine, I set the boot-file SMS env variable to pass boot image parms to netboot.img. Here's what I pass: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/sift22104.cfg This lets the netboot.img config eth0 to access the ks file on the server. I have this network line in the /lpars/sift22104.cfg kickstart file: network --device eth0 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com This works just fine and the install completes. I need to make this work on lpars that have never been loaded before and these lpars can have several eth adapters so the user will not know what order the adatpers will be configured so he can't specify eth0 or eth1 in the options above. I have expect code that goes thru the SMS menus and attempts to ping the server on each eth adatper until it finds one that is connected: I can pull the mac address for that card and could at that point create the netboot.img boot parms line that includes the mac address, but there is no way I can know what ethX redhat will confgure this card as. I tried changing the netboot.img parms line to: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth-id-2a:ca:30:00:40:02 ks=nfs:9.3.80.16:/lpars/sift22104.cfg and the kickstart network line to this: network --device eth-id-2a:ca:30:00:40:02 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com But it did not work.. I got a syntax error back from anaconda complaining about this: self.setNetwork(id, bootProto, ip, netmask, ethtool, device=device, onboot=onboot, dhcpclass=dhcpclass, essid=essid, wepkey=wepkey) File "/usr/lib/anaconda/installclass.py", line 332, in setNetwork dev = devices[device] KeyError: 'eth-id-2a:ca:30:00:40:02' I have searched thru doc's till I am blue and can't find out if there is some way (any correct syntax) to pass the mac address to netboot.img and anaconda instead of the ethX name. Dose anyone know how this can be done, is it even in the netboot.img/anaconda code to all this and deal with the mac address? This is the last thing I need to fully automate these installs for my users with kickstart instead of using expect to fill out all those changing install menus in text mode. thanks tom miller IBM. -------------- next part -------------- An HTML attachment was scrubbed... URL: From tmafs at us.ibm.com Tue Aug 22 23:59:21 2006 From: tmafs at us.ibm.com (Tom Miller) Date: Tue, 22 Aug 2006 18:59:21 -0500 Subject: Anaconda kickstart options and lowlevel netboot.img input options. Message-ID: Colin, When the power 5 lpars boot up, a boot option menu is presented to the user on the console for that lpar. The user can choose to break into the boot and navigate the boot control menus. Where, he can pick a nic adapter to use to issure the (bootp) request to a pre setup network install server.. We load aix and linux the same way at this level in the open firmware. The user must supply, ip of victim, ip of server, mask and gateway, and he can test ping the server to see if that nic is really connected to the network and the server is up. Once the connected adapter is found, the user can set this nic as the boot device and off it goes, issuing the bootp request to the server. Bootp is the only option we have on these machines. My expect code does all of this automatically, stopping in the boot menus during boot, finding the nic that will ping and setting the boot list.. It also uses the low level open firmware menus to setevn boot-file text ip=9.3.218.145 netmask=255.255.255.0 gateway=9.3.218.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/unitedp02.cfg so that these parms are passed to whatever boot image is retrived by bootp. So first OF menus needs ip to do boop, then netboot.img needs ip stuff to get kickstart file. then adconda needs ip stuff in kickstart file to config network for install. I already use PXELINUX to install the big machinds Hardware Management Consoles. They run linux on X86 platform. And it works ok there in our local lab, but again not outside the subnet. The OF code for these lpars does not support PXELINUX. I think my only hope is to find out is there is some systax that will work to pass the mac address to anacondo instead of the ethX.. If power 5 keeps supporting linux RedHat I am hope the anacondo developers will get around accepting the mac address. I also load linux SuSE from my servers and I have already done succesfull autoyast installs with and autoinst.xml file. Now these files (autoinst.xml) will take input like this: eth-id-2a:ca:30:00:20:02 they define the device with mac address. I have not yet tried passing the suse netboot image a nic name of this form. But I am hopeing it will work since the autoinst file uses that format in SuSE installs. tom Tuesday, August 22, 2006 6:30 PM To: "Discussion list about Kickstart" cc: From: "Coe, Colin C." Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. Hi Tom No, I'm not an Anaconda coder. There are quirks in the installer but the issue with the swapped NICs I believe is a kernel compile config quirk rather than an Anaconda problem. Pity about the DHCP. As I said, I'm not an Anaconda or Python coder, but what about a custom install class? Is this were a custom install class would be used? Also, given your working with Power5 LPARs I can only assume you're working with a beastie along the lines of a P595, maybe you could query the hypervisor? I know these are pretty locked down but... Just out of curiosity, what boot method do these use (bootp/pxe/etc)? If its PXE and you're able to use PXELINUX (I don't know if you can), you might be able to make use of the IPAPPEND option. Or better yet, you could have your expect script create a symlink called the MAC address of the NIC you want to use which points to a standard pxelinux config file. For example 2A:CA:30:00:40:02 /tftpboot/pxelinux.cfg/2a-ca-30-00-40-02 -> el4_lpar02 /tftpboot/pxelinux.cfg/el4_lpar02 Where el4_lpar02 would contain something like label el4_lpar kernel el4u4/vmlinuz append ksdevice=eth0 ks=nfs:9.3.80.16:/export/install/ks.cfg initrd=el4u4/initrd.img ip=9.3.22.202 method=nfs:10.240.34.254:/export/install/el4u4_x86_64 CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Wednesday, 23 August 2006 2:09 AM To: Discussion list about Kickstart Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. Collin, Thanks for your reply. I would love to use dhcp, this would solve all my problems but IBM prevents me from doing that. They do not pass dhcp requests thru the site bridges/gateways/routes so I would have to have a dhcp server on each subnet that contained victims to install. I have 5 servers in Austin, one in Rochester, Mn and one in Poughkeepsie, NY and they install 100's of systems a week on many different subnets. I did not realize that the installer and kernel couldn't even agree on which nic was eth0. Thats all the more need for anacondo and the netboot.img to deal with mac addresses. You don't happen to be an anaconda coder do you? thanks tom Monday, August 21, 2006 7:51 PM To: "Discussion list about Kickstart" cc: From: "Coe, Colin C." Subject: RE: Anaconda kickstart options and lowlevel netboot.img input options. If you know the MAC addr, could you use a DHCP server with an entry for each LPAR you want to install? Something like: allow booting; allow bootp; ... group { subnet 9.3.22.0 netmask 255.255.255.0 { option domain-name "ibm.com"; option subnet-mask 255.255.255.0; option broadcast-address 9.3.22.255; option domain-name-servers 9.0..7.1; option routers 9.3.22.1; } # PXE-specific configuration directives... next-server 9.3.22.253; filename "pxelinux.0"; ... host lpar02 { hardware ethernet 2A:CA:30:00:40:02; fixed-address 9.3.22.202; option host-name "lpar02"; } ... } Just a thought. Also, if you're using EL4, beware that sometimes the NIC that the installer decides is eth0 is not what the installed kernel thinks is eth0. CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Tom Miller Sent: Tuesday, 22 August 2006 3:06 AM To: kickstart-list at redhat.com Subject: Anaconda kickstart options and lowlevel netboot.img input options. I am using a kickstart.cfg file to load power 5 lpars via network install. I use a bootp server to deliver the RedHat netboot.img to the victim machine, I set the boot-file SMS env variable to pass boot image parms to netboot.img. Here's what I pass: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth0 ks=nfs:9.3.80.16:/lpars/sift22104.cfg This lets the netboot.img config eth0 to access the ks file on the server. I have this network line in the /lpars/sift22104.cfg kickstart file: network --device eth0 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com This works just fine and the install completes. I need to make this work on lpars that have never been loaded before and these lpars can have several eth adapters so the user will not know what order the adatpers will be configured so he can't specify eth0 or eth1 in the options above. I have expect code that goes thru the SMS menus and attempts to ping the server on each eth adatper until it finds one that is connected: I can pull the mac address for that card and could at that point create the netboot.img boot parms line that includes the mac address, but there is no way I can know what ethX redhat will confgure this card as. I tried changing the netboot.img parms line to: text ip=9.3.22.104 netmask=255.255.255.128 gateway=9.3.22.1 nameserver=9.0.7.1 ksdevice=eth-id-2a:ca:30:00:40:02 ks=nfs:9.3.80.16:/lpars/sift22104.cfg and the kickstart network line to this: network --device eth-id-2a:ca:30:00:40:02 --bootproto static --ip 9.3.22.104 --netmask 255.255.255.128 --gateway 9.3.22.1 --nameserver 9.0.7.1 --hostname sift22104.austin.ibm.com But it did not work.. I got a syntax error back from anaconda complaining about this: self.setNetwork(id, bootProto, ip, netmask, ethtool, device=device, onboot=onboot, dhcpclass=dhcpclass, essid=essid, wepkey=wepkey) File "/usr/lib/anaconda/installclass.py", line 332, in setNetwork dev = devices[device] KeyError: 'eth-id-2a:ca:30:00:40:02' I have searched thru doc's till I am blue and can't find out if there is some way (any correct syntax) to pass the mac address to netboot.img and anaconda instead of the ethX name. Dose anyone know how this can be done, is it even in the netboot.img/anaconda code to all this and deal with the mac address? This is the last thing I need to fully automate these installs for my users with kickstart instead of using expect to fill out all those changing install menus in text mode. thanks tom miller IBM. -------------- next part -------------- An HTML attachment was scrubbed... URL: From art at caa.is Fri Aug 25 15:47:59 2006 From: art at caa.is (Arnar =?ISO-8859-1?Q?=DE=F3rarinsson?=) Date: Fri, 25 Aug 2006 15:47:59 +0000 Subject: why o why - ftp install method problem Message-ID: <1156520880.25297.9.camel@black148> Hello I'm having a "small" problem with my kickstart...it just won't work ;) Using RHEL 4.0 WS 32bit I installed a machine using cdrom and then took the anaconda-ks.cfg file generated by that install and used that for a kickstart base using ftp install method. Now when I use that kickstart file I get the following error: "You are trying to install on a machine which isn't supported by this release of Red Hat Enterprise Linux WS." ... ok fine... but I just installed RHEL on the same machine using CDROM install method, why can't I install the same RHEL on the same maching using ftp install method ??? Anybody know why this happens and how to get around this ?? /Arnar -- -- Regards Arnar Thorarinsson Software Programmer / System Administrator Systems Department, Air Traffic Services Phone : +354 569 4289 ------------------------------------------ Icelandic Civil Aviation Administration Reykjavik Airport IS 121 Reykjavik Iceland http://www.caa.is ------------------------------------------ Thessi tolvupostur og vidhengi hans geta innihaldid trunadarupplysingar og er eingongu aetladur theim sem hann er stiladur a. Oheimil medferd tolvuposts thessa og vidhengja hans getur vardad skadabota- og refsiabyrgd samkvaemt logum um fjarskipti. Efni tolvupostsins og vidhengja er a abyrgd sendanda ef thad tengist ekki starfsemi Flugmalastjornar Islands. Ef Thu ert ekki skradur mottakandi og hefur fengid skeytid vegna mistaka, vinsamlegast hafdu strax samband vid sendanda. ------------ This e-mail and its attachments is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer. From Chip.Shabazian at bankofamerica.com Fri Aug 25 16:05:04 2006 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Fri, 25 Aug 2006 09:05:04 -0700 Subject: why o why - ftp install method problem In-Reply-To: <1156520880.25297.9.camel@black148> Message-ID: Sounds like your install tree that you are accessing via ftp may contain the wrong images. -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Arnar ??rarinsson Sent: Friday, August 25, 2006 8:48 AM To: kickstart-list at redhat.com Subject: why o why - ftp install method problem Hello I'm having a "small" problem with my kickstart...it just won't work ;) Using RHEL 4.0 WS 32bit I installed a machine using cdrom and then took the anaconda-ks.cfg file generated by that install and used that for a kickstart base using ftp install method. Now when I use that kickstart file I get the following error: "You are trying to install on a machine which isn't supported by this release of Red Hat Enterprise Linux WS." ... ok fine... but I just installed RHEL on the same machine using CDROM install method, why can't I install the same RHEL on the same maching using ftp install method ??? Anybody know why this happens and how to get around this ?? /Arnar -- -- Regards Arnar Thorarinsson Software Programmer / System Administrator Systems Department, Air Traffic Services Phone : +354 569 4289 ------------------------------------------ Icelandic Civil Aviation Administration Reykjavik Airport IS 121 Reykjavik Iceland http://www.caa.is ------------------------------------------ Thessi tolvupostur og vidhengi hans geta innihaldid trunadarupplysingar og er eingongu aetladur theim sem hann er stiladur a. Oheimil medferd tolvuposts thessa og vidhengja hans getur vardad skadabota- og refsiabyrgd samkvaemt logum um fjarskipti. Efni tolvupostsins og vidhengja er a abyrgd sendanda ef thad tengist ekki starfsemi Flugmalastjornar Islands. Ef Thu ert ekki skradur mottakandi og hefur fengid skeytid vegna mistaka, vinsamlegast hafdu strax samband vid sendanda. ------------ This e-mail and its attachments is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer. _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From art at caa.is Fri Aug 25 17:50:21 2006 From: art at caa.is (Arnar =?ISO-8859-1?Q?=DE=F3rarinsson?=) Date: Fri, 25 Aug 2006 17:50:21 +0000 Subject: why o why - ftp install method problem In-Reply-To: References: Message-ID: <1156528221.25297.19.camel@black148> Yup... and I know why... seems SOMEONE ;) copied the extras disk also into the base OS dir on the ftp site... different hdlist and more ... bugger ;) thx. /Arnar On Fri, 2006-08-25 at 09:05 -0700, Shabazian, Chip wrote: > Sounds like your install tree that you are accessing via ftp may contain the wrong images. > > -----Original Message----- > From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Arnar ??rarinsson > Sent: Friday, August 25, 2006 8:48 AM > To: kickstart-list at redhat.com > Subject: why o why - ftp install method problem > > Hello > I'm having a "small" problem with my kickstart...it just won't work ;) Using RHEL 4.0 WS 32bit > > I installed a machine using cdrom and then took the anaconda-ks.cfg file generated by that install and used that for a kickstart base using ftp install method. Now when I use that kickstart file I get the following > error: > "You are trying to install on a machine which isn't supported by this release of Red Hat Enterprise Linux WS." > ... ok fine... but I just installed RHEL on the same machine using CDROM install method, why can't I install the same RHEL on the same maching using ftp install method ??? > > Anybody know why this happens and how to get around this ?? > > /Arnar > > -- > -- > Regards > Arnar Thorarinsson > > Software Programmer / System Administrator Systems Department, Air Traffic Services Phone : +354 569 4289 > ------------------------------------------ > Icelandic Civil Aviation Administration > Reykjavik Airport > IS 121 Reykjavik > Iceland > http://www.caa.is > ------------------------------------------ > > > Thessi tolvupostur og vidhengi hans geta innihaldid trunadarupplysingar og er eingongu aetladur theim sem hann er stiladur a. > Oheimil medferd tolvuposts thessa og vidhengja hans getur vardad > skadabota- og refsiabyrgd samkvaemt logum um fjarskipti. Efni tolvupostsins og vidhengja er a abyrgd sendanda ef thad tengist ekki starfsemi Flugmalastjornar Islands. Ef Thu ert ekki skradur mottakandi og hefur fengid skeytid vegna mistaka, vinsamlegast hafdu strax samband vid sendanda. > > ------------ > > > This e-mail and its attachments is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. > If you received this in error, please contact the sender and delete the material from any computer. > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list Thessi tolvupostur og vidhengi hans geta innihaldid trunadarupplysingar og er eingongu aetla?ur theim sem hann er stiladur a. Oheimil medferd tolvuposts thessa og vidhengja hans getur vardad skadabota- og refsiabyrgd samkvaemt logum um fjarskipti. Efni tolvupostsins og vi?hengja er a abyrgd sendanda ef thad tengist ekki starfsemi Flugmalastjornar Islands. Ef Thu ert ekki skradur mottakandi og hefur fengid skeytid vegna mistaka, vinsamlegast hafdu strax samband vid sendanda. ------------ This e-mail and its attachments is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer. From jbrussard at netnumber.net Tue Aug 29 15:09:06 2006 From: jbrussard at netnumber.net (John Brussard) Date: Tue, 29 Aug 2006 11:09:06 -0400 Subject: trying to mount the dvd during post install. In-Reply-To: <1156528221.25297.19.camel@black148> Message-ID: <200608291609.k7TG93mR032061@mx3.redhat.com> Hi all, After the Kickstart is complete and the %post install is running, I am unable to mount the dvd and copy extra files from the dvd to the system. The %post makes the /mnt/dvdrom directory but does not mount the /dev/dvd and copy the customer files, IT DOES however continue on with the %post. Any ideas? %post groupadd oinstall groupadd dba . . . mkdir /var/spool/rpms/ mkdir /mnt/dvdrom mount -oro /dev/dvd /mnt/dvdrom cp /mnt/dvdrom/custom_files/* /var/spool/rpms/ chkconfig --del autofs chkconfig --del atd chkconfig --del anacron . . . Thanks in advance for your assistance. John From ebrown at lanl.gov Tue Aug 29 18:24:18 2006 From: ebrown at lanl.gov (Edward F. Brown) Date: Tue, 29 Aug 2006 12:24:18 -0600 (MDT) Subject: trying to mount the dvd during post install. In-Reply-To: <200608291609.k7TG93mR032061@mx3.redhat.com> References: <1156528221.25297.19.camel@black148> <200608291609.k7TG93mR032061@mx3.redhat.com> Message-ID: <46108.128.165.253.115.1156875858.squirrel@webmail.lanl.gov> On Tue, August 29, 2006 9:09 am, John Brussard wrote: > Hi all, > > After the Kickstart is complete and the %post install is running, I am > unable to mount the dvd and copy extra files from the dvd to the system. > > The %post makes the /mnt/dvdrom directory but does not mount the /dev/dvd > and copy the customer files, IT DOES however continue on with the %post. > > Any ideas? Someone might have the exact answer to your question, but if not, I find that there is no better %post debug and devel environment than to simply open a shell and see what is going on. (In fact, I would very much like to see this capability built into kickstart as a simple option: specify a DEBUG parameter, and, on error, a shell prompt is provided.) Here's how to do it for yourself: %post ... some command that deserves an error check... if [ $? -ne 0 ]; then chvt 3 exec < /dev/tty3 > /dev/tty3 2>/dev/tty3 sh fi Now try your commands manually, see what the errors are, try to find the workaround, and possibly finish successfully, without lots of starting completely over to test fixes. This is helpful when developing %post scripts, and it's very useful even in working scripts, to offer the ability to workaround onetime issues, like forgetting to put the cd or a network path down or a file on a remote server missing: you can almost always finish successfully without kickstarting repeatedly to work it out. Of course, edit the kickstart file when the problem is discovered. I actually put the chvt/exec at the top of %post, and the error test/shell command after every command that I want to be sure executes successfully. My whole %post runs in this chvt'ed window, so I can see what is going on. Also, my install runs in text mode, not sure how this would work if the install is in graphical mode. hth, Ed From jerlyon at gmail.com Tue Aug 29 21:53:04 2006 From: jerlyon at gmail.com (Jeremy Lyon) Date: Tue, 29 Aug 2006 15:53:04 -0600 Subject: trying to mount the dvd during post install. In-Reply-To: <46108.128.165.253.115.1156875858.squirrel@webmail.lanl.gov> References: <1156528221.25297.19.camel@black148> <200608291609.k7TG93mR032061@mx3.redhat.com> <46108.128.165.253.115.1156875858.squirrel@webmail.lanl.gov> Message-ID: <779919740608291453m66342a3aofea3dd3c61892ad9@mail.gmail.com> > that there is no better %post debug and devel environment than to simply > open a shell and see what is going on. (In fact, I would very much like > to see this capability built into kickstart as a simple option: specify a > DEBUG parameter, and, on error, a shell prompt is provided.) > There is no need for a DEBUG parameter. You can get to a shell for debugging by.. 1. not putting reboot in your ks.cfg 2. Crtl-Alt-F2 #2 will get you into a shell prompt that will allow you to do your debugging. If you are running your %post in a chroot fashion (default), the you will want to run chroot /mnt/sysimage and then do your debugging. -Jeremy, RHCE -------------- next part -------------- An HTML attachment was scrubbed... URL: From ebrown at lanl.gov Tue Aug 29 22:53:16 2006 From: ebrown at lanl.gov (Edward F. Brown) Date: Tue, 29 Aug 2006 16:53:16 -0600 (MDT) Subject: trying to mount the dvd during post install. In-Reply-To: <779919740608291453m66342a3aofea3dd3c61892ad9@mail.gmail.com> References: <1156528221.25297.19.camel@black148> <200608291609.k7TG93mR032061@mx3.redhat.com> <46108.128.165.253.115.1156875858.squirrel@webmail.lanl.gov> <779919740608291453m66342a3aofea3dd3c61892ad9@mail.gmail.com> Message-ID: <56451.128.165.0.81.1156891996.squirrel@webmail.lanl.gov> On Tue, August 29, 2006 3:53 pm, Jeremy Lyon wrote: > There is no need for a DEBUG parameter. You can get to a shell for > debugging by.. > > 1. not putting reboot in your ks.cfg > 2. Crtl-Alt-F2 That's a good point, and somewhat workable, especially if your %post is short and simple. But it's just basic good programming practice to check for errors as you proceed, and preferably offer useful error messages, such as echoing the command that failed. If you wait till the end, you may have well missed an error, not know how far back you need to check for problems, and if you want to finish without restarting, then you have to manually enter the rest of your script. Stopping at the first [critical - one you can't or don't want to proceed past] error, allowing a chance to fix it, and then resume, also can prevent many subsequent errors, besides being efficient and logical. I understand that the goal of kickstart is to run without intervention. But it takes time to develop robust %post scripts, and everything can and will go wrong in the future with scripts that work well now. It's such a simple idea really, it surprises me the resistance it meets. If your script is a few lines, then you might not need a DEBUG parameter. Our goal, frequently met, is that a server is completely configured, even online and in-service if it's a rebuild, as soon as it reboots the first time after being kickstarted, thanks to kickstart and cfengine. -Ed From forltran at yahoo.com Wed Aug 30 21:25:45 2006 From: forltran at yahoo.com (Lambert Tran) Date: Wed, 30 Aug 2006 14:25:45 -0700 (PDT) Subject: trying to mount the dvd during post install. In-Reply-To: <200608291609.k7TG93mR032061@mx3.redhat.com> Message-ID: <20060830212545.63199.qmail@web83207.mail.mud.yahoo.com> >From the RedHat Kickstart manual: Note "Note that the post-install script is run in a chroot environment; therefore, performing tasks such as copying scripts or RPMs from the installation media will not work." HOWEVER, we were able to "remount" the installation CD (your is DVD), and copy files from it, using --nochroot. %post --nochroot --interpreter /mnt/sysimage/bin/bash #Loading data files echo "Copying server files....." mount /tmp/cdrom /mnt/sysimage/mnt cp -r /mnt/sysimage/mnt/mycdmydir/* /mnt/sysimage/ .... umount /mnt/sysimage/mnt .... The trick at time was how to know the installation CD was actually is /tmp/cdrom... (your may be /tmp/dvd). To find out, we had a sleep 600 at the %post section, and went look around. Good Luck! -Lambert Tran --- John Brussard wrote: > Hi all, > > After the Kickstart is complete and the %post install is running, I am > unable to mount the dvd and copy extra files from the dvd to the system. > > The %post makes the /mnt/dvdrom directory but does not mount the /dev/dvd > and copy the customer files, IT DOES however continue on with the %post. > > Any ideas? > > > %post > groupadd oinstall > groupadd dba > . > . > . > mkdir /var/spool/rpms/ > mkdir /mnt/dvdrom > mount -oro /dev/dvd /mnt/dvdrom > cp /mnt/dvdrom/custom_files/* /var/spool/rpms/ > chkconfig --del autofs > chkconfig --del atd > chkconfig --del anacron > . > . > . > > > > Thanks in advance for your assistance. > > John > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From Zack.Bartel at Tectura.com Wed Aug 30 22:06:41 2006 From: Zack.Bartel at Tectura.com (Bartel, Zack) Date: Wed, 30 Aug 2006 15:06:41 -0700 Subject: xconfig option bluescreen error Message-ID: <5303BDF3F253554E889BE362FA0C604E016A6784@MAIL1.TecturaCorp.net> Hi there, I am trying to have kickstart auto configure Xorg during a custom FC5 install. I say custom because I am hand selecting the RPMs that get included. My current ks.cfg works fine until I add the "xconfig" directive (with any or no arguments). During the boot process immediately after anaconda starts the screen goes all blue with (what looks like python) error messages strewn across the screen. The messages fly by but at the end I get something like: ImportError: No module named x86config I am wondering what I am missing (like and RPM). Any ideas? Thanks ahead of time! Here is my ks.cfg: lang en_US keyboard us timezone --utc GMT rootpw password selinux --disabled # text graphical # skipx ###THIS WORKS #### NONE OF THESE DO # xconfig --driver "vmware" --videoram 16384 --resolution 800x600 --depth 16 # xconfig --depth=16 --resolution=800x600 # xconfig install cdrom bootloader --location=mbr zerombr clearpart --all --initlabel partition / --fstype ext3 --size 5000 partition swap --size 2000 partition /var --fstype ext3 --size 1 --grow authconfig --useshadow --enablemd5 %packages @Core @Base @base-x From ebrown at lanl.gov Wed Aug 30 23:32:25 2006 From: ebrown at lanl.gov (Ed Brown) Date: Wed, 30 Aug 2006 17:32:25 -0600 Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) In-Reply-To: <20060830212545.63199.qmail@web83207.mail.mud.yahoo.com> References: <20060830212545.63199.qmail@web83207.mail.mud.yahoo.com> Message-ID: <44F62009.1020104@lanl.gov> Lambert Tran wrote: > > The trick at time was how to know the installation CD was actually is > /tmp/cdrom... (your may be /tmp/dvd). To find out, we had a sleep 600 at the > %post section, and went look around. %post (and %pre) scripts are versatile and powerful tools, so why does the kickstart environment have a reputation of being difficult to work in? How many questions to this list revolve around figuring out how to determine what the heck is going on in %pre and %post scripts? Who has chosen to do things with scripts that execute on firstboot, rather than try to get something done in %post? How many of us resort to inserting sleeps or other tricks to buy time at a commandline to debug problems? How hard would it be to provide a debug mode, where non-zero return codes for commands (either all commands, or possibly only for specially-indicated commands) results in seeing the command that failed, and the reason/output if any, followed by a shell prompt? Would this be useful for anyone here? (Is this the wrong list, should it be posted to anaconda-devel instead (or also)?) -Ed? From Colin.Coe at woodside.com.au Wed Aug 30 23:36:15 2006 From: Colin.Coe at woodside.com.au (Coe, Colin C.) Date: Thu, 31 Aug 2006 07:36:15 +0800 Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) Message-ID: <17EBC05307239C4896ED25244F42A80101367B2F@permls05.wde.woodside.com.au> We have a minimal %post section that only copies our script ks_post.sh to /var/tmp and inserts 'sh /var/tmp/ks_post.sh' into /etc/rc.d/rc.local with a clean up after it has run. As far as debugging goes, whats wrong with: %post set -x cmd1 && cmd2 && cmd3 && cmd4 || fixit ??? All the debugging goes to vt3. CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Ed Brown Sent: Thursday, 31 August 2006 7:32 AM To: Discussion list about Kickstart Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) Lambert Tran wrote: > > The trick at time was how to know the installation CD was actually is > /tmp/cdrom... (your may be /tmp/dvd). To find out, we had a sleep 600 at the > %post section, and went look around. %post (and %pre) scripts are versatile and powerful tools, so why does the kickstart environment have a reputation of being difficult to work in? How many questions to this list revolve around figuring out how to determine what the heck is going on in %pre and %post scripts? Who has chosen to do things with scripts that execute on firstboot, rather than try to get something done in %post? How many of us resort to inserting sleeps or other tricks to buy time at a commandline to debug problems? How hard would it be to provide a debug mode, where non-zero return codes for commands (either all commands, or possibly only for specially-indicated commands) results in seeing the command that failed, and the reason/output if any, followed by a shell prompt? Would this be useful for anyone here? (Is this the wrong list, should it be posted to anaconda-devel instead (or also)?) -Ed? _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From ebrown at lanl.gov Thu Aug 31 03:20:10 2006 From: ebrown at lanl.gov (Edward F. Brown) Date: Wed, 30 Aug 2006 21:20:10 -0600 (MDT) Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) In-Reply-To: <17EBC05307239C4896ED25244F42A80101367B2F@permls05.wde.woodside.com.au > References: <17EBC05307239C4896ED25244F42A80101367B2F@permls05.wde.woodside.com.au> Message-ID: <39380.128.165.253.27.1156994410.squirrel@webmail.lanl.gov> And WHY is having a "minimal %post section" a good thing? Why not run the commands in ks_post.sh in ks? On Wed, August 30, 2006 5:36 pm, Coe, Colin C. wrote: > > We have a minimal %post section that only copies our script ks_post.sh > to /var/tmp and inserts 'sh /var/tmp/ks_post.sh' into /etc/rc.d/rc.local > with a clean up after it has run. > > As far as debugging goes, whats wrong with: > %post > set -x > cmd1 && > cmd2 && > cmd3 && > cmd4 || > fixit > > ??? > > All the debugging goes to vt3. > > CC > > -----Original Message----- > From: kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Ed Brown > Sent: Thursday, 31 August 2006 7:32 AM > To: Discussion list about Kickstart > Subject: error handling in %post (was: Re: trying to mount the dvd > during post install.) > > > Lambert Tran wrote: >> >> The trick at time was how to know the installation CD was actually is >> /tmp/cdrom... (your may be /tmp/dvd). To find out, we had a sleep 600 > at the >> %post section, and went look around. > > %post (and %pre) scripts are versatile and powerful tools, so why does > the kickstart environment have a reputation of being difficult to work > in? How many questions to this list revolve around figuring out how > to determine what the heck is going on in %pre and %post scripts? Who > has chosen to do things with scripts that execute on firstboot, rather > than try to get something done in %post? How many of us resort to > inserting sleeps or other tricks to buy time at a commandline to debug > problems? How hard would it be to provide a debug mode, where > non-zero return codes for commands (either all commands, or possibly > only for specially-indicated commands) results in seeing the command > that failed, and the reason/output if any, followed by a shell prompt? > Would this be useful for anyone here? (Is this the wrong list, > should it be posted to anaconda-devel instead (or also)?) > > -Ed? > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > From Colin.Coe at woodside.com.au Thu Aug 31 03:51:55 2006 From: Colin.Coe at woodside.com.au (Coe, Colin C.) Date: Thu, 31 Aug 2006 11:51:55 +0800 Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) Message-ID: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> Because the INSTALLER kernel is different to the INSTALLED kernel. Its compiled with different options (especially evident under EL4) and it has different modules. chroot ain't everything... CC -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Edward F. Brown Sent: Thursday, 31 August 2006 11:20 AM To: Discussion list about Kickstart Subject: RE: error handling in %post (was: Re: trying to mount the dvd during post install.) And WHY is having a "minimal %post section" a good thing? Why not run the commands in ks_post.sh in ks? On Wed, August 30, 2006 5:36 pm, Coe, Colin C. wrote: > > We have a minimal %post section that only copies our script ks_post.sh > to /var/tmp and inserts 'sh /var/tmp/ks_post.sh' into /etc/rc.d/rc.local > with a clean up after it has run. > > As far as debugging goes, whats wrong with: > %post > set -x > cmd1 && > cmd2 && > cmd3 && > cmd4 || > fixit > > ??? > > All the debugging goes to vt3. > > CC > > -----Original Message----- > From: kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Ed Brown > Sent: Thursday, 31 August 2006 7:32 AM > To: Discussion list about Kickstart > Subject: error handling in %post (was: Re: trying to mount the dvd > during post install.) > > > Lambert Tran wrote: >> >> The trick at time was how to know the installation CD was actually is >> /tmp/cdrom... (your may be /tmp/dvd). To find out, we had a sleep 600 > at the >> %post section, and went look around. > > %post (and %pre) scripts are versatile and powerful tools, so why does > the kickstart environment have a reputation of being difficult to work > in? How many questions to this list revolve around figuring out how > to determine what the heck is going on in %pre and %post scripts? Who > has chosen to do things with scripts that execute on firstboot, rather > than try to get something done in %post? How many of us resort to > inserting sleeps or other tricks to buy time at a commandline to debug > problems? How hard would it be to provide a debug mode, where > non-zero return codes for commands (either all commands, or possibly > only for specially-indicated commands) results in seeing the command > that failed, and the reason/output if any, followed by a shell prompt? > Would this be useful for anyone here? (Is this the wrong list, > should it be posted to anaconda-devel instead (or also)?) > > -Ed? > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From jkeating at j2solutions.net Thu Aug 31 03:55:18 2006 From: jkeating at j2solutions.net (Jesse Keating) Date: Wed, 30 Aug 2006 23:55:18 -0400 Subject: error handling in %post (was: Re: trying to mount the dvd during post install.) In-Reply-To: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> References: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> Message-ID: <200608302355.21815.jkeating@j2solutions.net> On Wednesday 30 August 2006 23:51, Coe, Colin C. wrote: > Because the INSTALLER kernel is different to the INSTALLED kernel. ?Its > compiled with different options (especially evident under EL4) and it > has different modules. ?chroot ain't everything... This isn't true as of recent Fedora releases, and the upcoming RHEL5 release. The installer kernel is the same as the installed kernel, unless of course you've enabled the updates repo at install time and thus install a newer kernel than that which is in the installer.... -- Jesse Keating RHCE (geek.j2solutions.net) Fedora Legacy Team (www.fedoralegacy.org) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From terjekv at math.uio.no Thu Aug 31 12:45:08 2006 From: terjekv at math.uio.no (Terje Kvernes) Date: Thu, 31 Aug 2006 14:45:08 +0200 Subject: error handling in %post In-Reply-To: <200608302355.21815.jkeating@j2solutions.net> (Jesse Keating's message of "Wed, 30 Aug 2006 23:55:18 -0400") References: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> <200608302355.21815.jkeating@j2solutions.net> Message-ID: Jesse Keating writes: > On Wednesday 30 August 2006 23:51, Coe, Colin C. wrote: > > Because the INSTALLER kernel is different to the INSTALLED kernel. ?Its > > compiled with different options (especially evident under EL4) and it > > has different modules. ?chroot ain't everything... > > This isn't true as of recent Fedora releases, and the upcoming RHEL5 > release. this'll be nice. right now a lot of our machines need a secondary network card while installing while the integrated card gets used by the installed kernel. bloody annoying to say the very least. of course, the general lack of support for PCI-E versions of the sk98-based stuff isn't exactly funny either. :-/ [ ... ] -- Terje From ebrown at lanl.gov Thu Aug 31 14:28:22 2006 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 31 Aug 2006 08:28:22 -0600 Subject: error handling in %post In-Reply-To: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> References: <17EBC05307239C4896ED25244F42A80101367B30@permls05.wde.woodside.com.au> Message-ID: <44F6F206.8060708@lanl.gov> Coe, Colin C. wrote: > Because the INSTALLER kernel is different to the INSTALLED kernel. Its > compiled with different options (especially evident under EL4) and it > has different modules. chroot ain't everything... And even with a different kernel, when is this is a reason not to do post install configuration in %post? The only situation that I can think of is if you have to install third party software that involves building a kernel module, in which case there needs to be a mechanism to update the module after every kernel change anyway. I'm not saying that installing a ks_post.sh script that runs from /etc/rc.d/rc.local after rebooting and then cleans up after itself is a bad idea: it's a GOOD solution to the situation that exists now, a solution that many have chosen, because doing things in %post is awkward and problematic. A 'debug mode' or error-handling framework for %pre and %post scripts like I'm suggesting, would allow for far easier development, and a more robust recovery capability, and allow %post to be more fully utilized for what is was intended for, instead of resorting to post-post scripts. -Ed