From erik.niessen at gmail.com Thu Mar 3 10:59:05 2005 From: erik.niessen at gmail.com (Erik Niessen) Date: Thu, 3 Mar 2005 11:59:05 +0100 Subject: Adding a label to isolinux.cfg doesn't seem to work under fc1 Message-ID: Hi, I am trying to add a label for my custum kickstart file to isolinux.cfg. I copied my ks.cfg file to the top level dir and also to isolinux dir to be safe. Then added a label to isolinux.cfg. Below you can find my isolinux.cfg Then I do a mkisofs and boot from my new cd. But when I do linux ks or linux kickstart it starts the normal installation not my kickstart. But when I type linux ks=cdrom:/ks.cfg then it works??? Any ideas? Is there a way to debug this? Thanks, Erik default linux prompt 1 timeout 600 display boot.msg F1 boot.msg F2 options.msg F3 general.msg F4 param.msg F5 rescue.msg F7 snake.msg F8 kickstart.msg label linux kernel vmlinuz append initrd=initrd.img ramdisk_size=8192 label text kernel vmlinuz append initrd=initrd.img text ramdisk_size=8192 label expert kernel vmlinuz append expert initrd=initrd.img ramdisk_size=8192 label ks kernel vmlinuz append ks=cdrom:/ks.cfg initrd=initrd.img ramdisk_size=8192 label lowres kernel vmlinuz append initrd=initrd.img lowres ramdisk_size=8192 label memtest86 kernel memtest append - label kickstart kernel vmlinuz append initrd=initrd.img ramdisk_size=8192 ks=cdrom:/ks.cfg From samuli.jarvinen at hp.com Thu Mar 3 11:19:07 2005 From: samuli.jarvinen at hp.com (Samuli Jarvinen) Date: Thu, 03 Mar 2005 13:19:07 +0200 Subject: Adding a label to isolinux.cfg doesn't seem to work under fc1 In-Reply-To: References: Message-ID: <1109848747.3832.25.camel@sajarnb> Try typing only kickstart. -Samuli On Thu, 2005-03-03 at 11:59 +0100, Erik Niessen wrote: > Hi, > > I am trying to add a label for my custum kickstart file to > isolinux.cfg. I copied my ks.cfg file to the top level dir and also to > isolinux dir to be safe. Then added a label to isolinux.cfg. Below you > can find my isolinux.cfg > > Then I do a mkisofs and boot from my new cd. But when I do linux ks or > linux kickstart > it starts the normal installation not my kickstart. But when I type > linux ks=cdrom:/ks.cfg then it works??? Any ideas? Is there a way to debug this? > > Thanks, > > Erik > > default linux > prompt 1 > timeout 600 > display boot.msg > F1 boot.msg > F2 options.msg > F3 general.msg > F4 param.msg > F5 rescue.msg > F7 snake.msg > F8 kickstart.msg > label linux > kernel vmlinuz > append initrd=initrd.img ramdisk_size=8192 > label text > kernel vmlinuz > append initrd=initrd.img text ramdisk_size=8192 > label expert > kernel vmlinuz > append expert initrd=initrd.img ramdisk_size=8192 > label ks > kernel vmlinuz > append ks=cdrom:/ks.cfg initrd=initrd.img ramdisk_size=8192 > label lowres > kernel vmlinuz > append initrd=initrd.img lowres ramdisk_size=8192 > label memtest86 > kernel memtest > append - > label kickstart > kernel vmlinuz > append initrd=initrd.img ramdisk_size=8192 ks=cdrom:/ks.cfg > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From iain at uk.colt.net Thu Mar 3 11:51:28 2005 From: iain at uk.colt.net (Iain Conochie) Date: Thu, 03 Mar 2005 11:51:28 +0000 Subject: Adding a label to isolinux.cfg doesn't seem to work under fc1 In-Reply-To: References: Message-ID: <4226FA40.3090902@uk.colt.net> Erik, To debug it you can, once the system has booted from the CDROM, and the install is away to setup the partitions, pressing F2 will give you a shell. Type cat /proc/cmdline This will tell you the options the system was booted with. Cheers Iain Erik Niessen wrote: >Hi, > >I am trying to add a label for my custum kickstart file to >isolinux.cfg. I copied my ks.cfg file to the top level dir and also to >isolinux dir to be safe. Then added a label to isolinux.cfg. Below you >can find my isolinux.cfg > >Then I do a mkisofs and boot from my new cd. But when I do linux ks or >linux kickstart >it starts the normal installation not my kickstart. But when I type >linux ks=cdrom:/ks.cfg then it works??? Any ideas? Is there a way to debug this? > >Thanks, > > Erik > >default linux >prompt 1 >timeout 600 >display boot.msg >F1 boot.msg >F2 options.msg >F3 general.msg >F4 param.msg >F5 rescue.msg >F7 snake.msg >F8 kickstart.msg >label linux >kernel vmlinuz >append initrd=initrd.img ramdisk_size=8192 >label text >kernel vmlinuz >append initrd=initrd.img text ramdisk_size=8192 >label expert >kernel vmlinuz >append expert initrd=initrd.img ramdisk_size=8192 >label ks >kernel vmlinuz >append ks=cdrom:/ks.cfg initrd=initrd.img ramdisk_size=8192 >label lowres >kernel vmlinuz >append initrd=initrd.img lowres ramdisk_size=8192 >label memtest86 >kernel memtest >append - >label kickstart >kernel vmlinuz >append initrd=initrd.img ramdisk_size=8192 ks=cdrom:/ks.cfg > >_______________________________________________ >Kickstart-list mailing list >Kickstart-list at redhat.com >https://www.redhat.com/mailman/listinfo/kickstart-list > > From email at jasonkohles.com Thu Mar 3 13:39:52 2005 From: email at jasonkohles.com (Jason Kohles) Date: Thu, 3 Mar 2005 08:39:52 -0500 Subject: Adding a label to isolinux.cfg doesn't seem to work under fc1 In-Reply-To: References: Message-ID: <20050303133952.GA25493@mail.jasonkohles.com> On Thu, Mar 03, 2005 at 11:59:05AM +0100, Erik Niessen wrote: > > Then I do a mkisofs and boot from my new cd. But when I do linux ks or > linux kickstart > it starts the normal installation not my kickstart. But when I type > linux ks=cdrom:/ks.cfg then it works??? Any ideas? Is there a way to debug this? > By typing 'linux kickstart', you are telling it to boot the linux label, and to add kickstart to the append= list. With the configuration you have here you need to type just 'kickstart' to boot it. > Thanks, > > Erik > > default linux > prompt 1 > timeout 600 > display boot.msg > F1 boot.msg > F2 options.msg > F3 general.msg > F4 param.msg > F5 rescue.msg > F7 snake.msg > F8 kickstart.msg > label linux > kernel vmlinuz > append initrd=initrd.img ramdisk_size=8192 > label text > kernel vmlinuz > append initrd=initrd.img text ramdisk_size=8192 > label expert > kernel vmlinuz > append expert initrd=initrd.img ramdisk_size=8192 > label ks > kernel vmlinuz > append ks=cdrom:/ks.cfg initrd=initrd.img ramdisk_size=8192 > label lowres > kernel vmlinuz > append initrd=initrd.img lowres ramdisk_size=8192 > label memtest86 > kernel memtest > append - > label kickstart > kernel vmlinuz > append initrd=initrd.img ramdisk_size=8192 ks=cdrom:/ks.cfg > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -- Jason Kohles A witty saying proves nothing. email at jasonkohles.com -- Voltaire (1694 - 1778) http://www.jasonkohles.com/ From davidbartlett2003 at yahoo.co.uk Thu Mar 3 15:10:50 2005 From: davidbartlett2003 at yahoo.co.uk (David Bartlett) Date: Thu, 3 Mar 2005 15:10:50 +0000 (GMT) Subject: Display progress during long kickstart %post script Message-ID: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> Hi all, I run a quite lengthy %post script (runtime = 5-10 minutes) as part of my kickstart build and would like to include some useful output to reassure my users that the build process is running normally and has not hung. A progress bar to match the [text interface] anaconda style would be ideal, but lines of text simply echoed to screen would be ok in the meantime. It looks as though the %post is spawned out of anaconda and I'm not sure how to display messages to screen from within my %post script (as anaconda still owns the screen). I am reluctant to change the anaconda python code as I really want to keep the vendor tools standard. I use the text-only version of anaconda and have no interest in the GUI component. Has anyone else successfully controlled the screen whilst anaconda is running, or have any thoughts on this? Thanks in advance. Rgds, David Bartlett. Send instant messages to your online friends http://uk.messenger.yahoo.com From ajaymulwani at gmail.com Thu Mar 3 15:30:38 2005 From: ajaymulwani at gmail.com (Ajay Mulwani) Date: Thu, 3 Mar 2005 21:00:38 +0530 Subject: Display progress during long kickstart %post script In-Reply-To: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> Message-ID: <5520b2790503030730404f6d4@mail.gmail.com> One of the virtual consoles, I think Ctrl+Alt+F3 already displays the progress of %post. You will probably not require any such script but just to educate the users. Regards, Ajay On Thu, 3 Mar 2005 15:10:50 +0000 (GMT), David Bartlett wrote: > Hi all, > > I run a quite lengthy %post script (runtime = 5-10 > minutes) as part of my kickstart build and would like > to include some useful output to reassure my users > that the build process is running normally and has not > hung. A progress bar to match the [text interface] > anaconda style would be ideal, but lines of text > simply echoed to screen would be ok in the meantime. > > It looks as though the %post is spawned out of > anaconda and I'm not sure how to display messages to > screen from within my %post script (as anaconda still > owns the screen). I am reluctant to change the > anaconda python code as I really want to keep the > vendor tools standard. I use the text-only version of > anaconda and have no interest in the GUI component. > > Has anyone else successfully controlled the screen > whilst anaconda is running, or have any thoughts on > this? > > Thanks in advance. > > Rgds, David Bartlett. > > Send instant messages to your online friends http://uk.messenger.yahoo.com > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > From mshuler at rackspace.com Thu Mar 3 15:37:10 2005 From: mshuler at rackspace.com (Michael Shuler) Date: Thu, 03 Mar 2005 09:37:10 -0600 Subject: Display progress during long kickstart %post script In-Reply-To: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> Message-ID: <42272F26.7080502@rackspace.com> David Bartlett wrote: > I run a quite lengthy %post script (runtime = 5-10 > minutes) as part of my kickstart build and would like > to include some useful output to reassure my users > that the build process is running normally and has not > hung. We use the following method as a beginning point, in order to have interactive post-install scripting, as well as logging the complete output of post-install to the terminal and to a text file, which is uploaded back to the kickstart server for documentation, as the last script called in %post. %post # change to new vt and set stout/stdin exec < /dev/tty6 > /dev/tty6 chvt 6 # run post-install # write/call scripts here # change back to first vt chvt 1 -- Kind Regards, Michael Shuler Rackspace Managed Hosting From bdoctor at ps-ax.com Thu Mar 3 15:43:26 2005 From: bdoctor at ps-ax.com (Brad Doctor) Date: Thu, 3 Mar 2005 08:43:26 -0700 (MST) Subject: Display progress during long kickstart %post script In-Reply-To: <5520b2790503030730404f6d4@mail.gmail.com> Message-ID: <20050303154326.F13F13A2A@bigbox.ps-ax.com> If you want to get into the code, kickstart.py is where to look. For example: # added intf as the final call argument def postAction(self, rootPath, serial, intf): mylen = len(self.postScripts) if mylen <= 1: waitwin = intf.waitWindow(_("Running post configuration scripts"),_("Please wait, this may take a few moments.. %s script to execute") %mylen ) else: waitwin = intf.waitWindow(_("Running post configuration scripts"),_("Please wait, this may take a few moments.. %s scripts to execute") %mylen ) log("Running kickstart %%post script(s)") for script in self.postScripts: script.run(rootPath, serial) log("All kickstart %%post script(s) have been run") # added this waitwin.pop() waitwin.pop() You have to also modify packages.py to provide the intf, which provides the window functionality as well: # Added intf=None in call so we can provide a gui def doPostAction(id, instPath, intf=None): id.instClass.postAction(instPath, flags.serial,intf) Having 'intf=None' maintains compatibility with other callers to this method, minimizing the ripple effect - if it is provided, cool, if not, cool. I think that is the extent of it. You then get a little window letting the user know how many scripts there are. No progress bar, but extending it should be simple -- however how will you update it's progress? I don't think you can do callbacks from the post interpreters. That was my issue and why I didn't do that. -brad > One of the virtual consoles, I think Ctrl+Alt+F3 already displays the > progress of %post. You will probably not require any such script but > just to educate the users. > > Regards, > Ajay > > > On Thu, 3 Mar 2005 15:10:50 +0000 (GMT), David Bartlett > wrote: > > Hi all, > > > > I run a quite lengthy %post script (runtime = 5-10 > > minutes) as part of my kickstart build and would like > > to include some useful output to reassure my users > > that the build process is running normally and has not > > hung. A progress bar to match the [text interface] > > anaconda style would be ideal, but lines of text > > simply echoed to screen would be ok in the meantime. > > > > It looks as though the %post is spawned out of > > anaconda and I'm not sure how to display messages to > > screen from within my %post script (as anaconda still > > owns the screen). I am reluctant to change the > > anaconda python code as I really want to keep the > > vendor tools standard. I use the text-only version of > > anaconda and have no interest in the GUI component. > > > > Has anyone else successfully controlled the screen > > whilst anaconda is running, or have any thoughts on > > this? > > > > Thanks in advance. > > > > Rgds, David Bartlett. > > > > Send instant messages to your online friends http://uk.messenger.yahoo.com > > > > _______________________________________________ > > 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 > -- Brad Doctor, CISSP From dan at half-asleep.com Thu Mar 3 15:55:58 2005 From: dan at half-asleep.com (Daniel Segall) Date: Thu, 3 Mar 2005 10:55:58 -0500 (EST) Subject: Display progress during long kickstart %post script In-Reply-To: <42272F26.7080502@rackspace.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> <42272F26.7080502@rackspace.com> Message-ID: <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> Hey Michael, Speaking of this,... have you seen any problems doing this method with RHES4 or FC3? I use a similar method (tty3), and ever since FC3, the display doesn't show up automatically. You have to hit enter blindly for my first (second really) dialog prompt to show up. Do you think that by using tty6 or another unused tty would resolve this? Hope all is well at the rack. I miss you guys... Thanks! -Dan > David Bartlett wrote: >> I run a quite lengthy %post script (runtime = 5-10 >> minutes) as part of my kickstart build and would like >> to include some useful output to reassure my users >> that the build process is running normally and has not >> hung. > > We use the following method as a beginning point, in order to have > interactive post-install scripting, as well as logging the complete > output of post-install to the terminal and to a text file, which is > uploaded back to the kickstart server for documentation, as the last > script called in %post. > > %post > # change to new vt and set stout/stdin > exec < /dev/tty6 > /dev/tty6 > chvt 6 > # run post-install > > # write/call scripts here > > # change back to first vt > chvt 1 > > -- > Kind Regards, > Michael Shuler > Rackspace Managed Hosting > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > From mshuler at rackspace.com Thu Mar 3 16:13:00 2005 From: mshuler at rackspace.com (Michael Shuler) Date: Thu, 03 Mar 2005 10:13:00 -0600 Subject: Display progress during long kickstart %post script In-Reply-To: <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> <42272F26.7080502@rackspace.com> <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> Message-ID: <4227378C.8000208@rackspace.com> Daniel Segall wrote: > Hey Michael, Hi Dan! > Speaking of this,... have you seen any problems doing this method with > RHES4 or FC3? I use a similar method (tty3), and ever since FC3, the > display doesn't show up automatically. You have to hit enter blindly for > my first (second really) dialog prompt to show up. Do you think that by > using tty6 or another unused tty would resolve this? Using tty6 works perfectly in RHEL-4, and I would expect that FC-3 should work fine, although we do not maintain a Fedora kickstart. > Hope all is well at the rack. I miss you guys... The rack is very well, and you are missed, too ;) Warm Regards, Michael Shuler Rackspace Managed Hosting From davidbartlett2003 at yahoo.co.uk Thu Mar 3 16:34:27 2005 From: davidbartlett2003 at yahoo.co.uk (David Bartlett) Date: Thu, 3 Mar 2005 16:34:27 +0000 (GMT) Subject: Display progress during long kickstart %post script In-Reply-To: <42272F26.7080502@rackspace.com> Message-ID: <20050303163427.66959.qmail@web26304.mail.ukl.yahoo.com> Hi Michael, I just wanted to pop you a THANK YOU e-mail for your prompt & superb reply. Your solution is spot-on and exactly what I need to get the ball rolling. The people in my office have been left wondering why I've been whooping and punching the air for the last five minutes! Thanks a lot man -- really good solution. Rgds, David Bartlett. --- Michael Shuler wrote: > David Bartlett wrote: > > I run a quite lengthy %post script (runtime = 5-10 > > minutes) as part of my kickstart build and would > like > > to include some useful output to reassure my users > > that the build process is running normally and has > not > > hung. > > We use the following method as a beginning point, in > order to have > interactive post-install scripting, as well as > logging the complete > output of post-install to the terminal and to a text > file, which is > uploaded back to the kickstart server for > documentation, as the last > script called in %post. > > %post > # change to new vt and set stout/stdin > exec < /dev/tty6 > /dev/tty6 > chvt 6 > # run post-install > > # write/call scripts here > > # change back to first vt > chvt 1 > > -- > Kind Regards, > Michael Shuler > Rackspace Managed Hosting > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > Send instant messages to your online friends http://uk.messenger.yahoo.com From rreed at ops.sgp.arm.gov Thu Mar 3 21:14:19 2005 From: rreed at ops.sgp.arm.gov (Ronald Reed) Date: Thu, 03 Mar 2005 15:14:19 -0600 Subject: initrd.img module loading In-Reply-To: References: Message-ID: <1109884458.2982.56.camel@ronnote> I created a new initrd and tried to use it as part of the install. But this didn't work. I understand what your steps are doing, but that doesn't work for what I need. I have to make sure that the aacraid module gets loaded first so that it will be sda. Your ideal works, but it is after the fact, which means that if I do it in the %post like you said, the machine would not be bootable, since the drives that were attached to the megaraid card were formated and setup as the root device and the new initrd will load the aacraid first, which will having nothing on it. Ron On Tue, 2005-02-22 at 13:39, Peter Eisch wrote: > Did you try rebuilding the initrd.img using something like: > > /sbin/mkinitrd --preload scsi_mod --preload sd_mod --with aacraid \ > --with megaraid --with aic7xxx --with eepro100 --with bcm5700 \ > /boot/ninitrd-%{KVERREL}.img %{KVERREL} > /bin/mv /boot/initrd-%{KVERREL}.img /boot/oinitrd-%{KVERREL}.img > /bin/mv /boot/ninitrd-%{KVERREL}.img /boot/initrd-%{KVERREL}.img > > I put this in my kernel's %post so it gets done at boot, but do the same > kind of thing with the initrd.img on the install media. > > peter > > On 2/22/05 12:38 PM, "Ronald Reed" wrote: > > > I have found that the scsihosts= boot line command is only for kernels > > that have the modules built into the kernel, but no other resolution has > > been found. > > > > I have tried the device parameter in the kickstart file, but anaconda > > seems to ignore them. Anyone have any other ideals? Here is a portion of > > my kickstart file: > > > > #kickstart file (partial) > > install > > cdrom > > lang en_US.UTF-8 > > langsupport --default en_US.UTF-8 en_US.UTF-8 > > device scsi aacraid > > device scsi megaraid > > device scsi aic7xxx > > device eth eepro100 > > device eth bcm5700 > > keyboard us > > mouse genericwheelps/2 --device psaux > > skipx > > network --device eth0 --bootproto dhcp --hostname nfs-basic > > > > Ron > > > > On Tue, 2005-02-08 at 12:27, Ronald Reed wrote: > >> No, I need the drive letter assignments to be different. > >> > >> Ron > >> > >> On Tue, 2005-02-08 at 11:58, Ed Brown wrote: > >>> The --driveorder option won't change the drive letter assignments, but > >>> it will change the drive that grub is installed to. Isn't that what you > >>> are after? In your case, you probably want: > >>> > >>> bootloader --driveorder=sdb,sda --location=mbr > >>> > >>> -Ed > >>> > >>> > >>> On Tue, 2005-02-08 at 10:53, Ronald Reed wrote: > >>>> Yes, the aacraid device get set to /dev/sdb. The accraid device is a > >>>> mirrored pair of 36GB drives that are inside the hardware. The megaraid > >>>> device is a Raid 5 group of 4 73GB drives (3 in the array, and 1 hot > >>>> spare). The megaraid is to be used for NFS file storage and the accraid > >>>> is to be used for the system drive. > >>>> > >>>> Ron > >>>> > >>>> On Tue, 2005-02-08 at 11:47, Brian Long wrote: > >>>> > >>>>> What does it detect the aacraid device as? /dev/sdb? If so, why don't > >>>>> you use the bootloader option mentioned previously to get grub to use > >>>>> sdb as the boot drive? > >>>>> > >>>>> /Brian/ > >>> > >>> _______________________________________________ > >>> 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 -- =========================== Ron Reed RedHat Certified Engineer SGP Computer Department Manager Unix Systems Administrator ARM SGP CART Site (580)388-4053 ron.reed at arm.gov From ebrown at lanl.gov Thu Mar 3 22:44:09 2005 From: ebrown at lanl.gov (Edward F. Brown) Date: Thu, 3 Mar 2005 15:44:09 -0700 (MST) Subject: initrd.img module loading In-Reply-To: <1109884458.2982.56.camel@ronnote> References: <1109884458.2982.56.camel@ronnote> Message-ID: <32808.128.165.7.35.1109889849.squirrel@webmail.lanl.gov> Is there a particular reason "sda' has to be assigned to this drive? I understand that it would be great if linux allowed more determination over drive letter assignments, I've wrestled with this too. But is there an issue for which a workaround hasn't been suggested? -Ed > I created a new initrd and tried to use it as part of the install. But > this didn't work. I understand what your steps are doing, but that > doesn't work for what I need. > > I have to make sure that the aacraid module gets loaded first so that it > will be sda. Your ideal works, but it is after the fact, which means > that if I do it in the %post like you said, the machine would not be > bootable, since the drives that were attached to the megaraid card were > formated and setup as the root device and the new initrd will load the > aacraid first, which will having nothing on it. > > Ron > > On Tue, 2005-02-22 at 13:39, Peter Eisch wrote: >> Did you try rebuilding the initrd.img using something like: >> >> /sbin/mkinitrd --preload scsi_mod --preload sd_mod --with aacraid \ >> --with megaraid --with aic7xxx --with eepro100 --with bcm5700 \ >> /boot/ninitrd-%{KVERREL}.img %{KVERREL} >> /bin/mv /boot/initrd-%{KVERREL}.img /boot/oinitrd-%{KVERREL}.img >> /bin/mv /boot/ninitrd-%{KVERREL}.img /boot/initrd-%{KVERREL}.img >> >> I put this in my kernel's %post so it gets done at boot, but do the same >> kind of thing with the initrd.img on the install media. >> >> peter >> >> On 2/22/05 12:38 PM, "Ronald Reed" wrote: >> >> > I have found that the scsihosts= boot line command is only for kernels >> > that have the modules built into the kernel, but no other resolution >> has >> > been found. >> > >> > I have tried the device parameter in the kickstart file, but anaconda >> > seems to ignore them. Anyone have any other ideals? Here is a portion >> of >> > my kickstart file: >> > >> > #kickstart file (partial) >> > install >> > cdrom >> > lang en_US.UTF-8 >> > langsupport --default en_US.UTF-8 en_US.UTF-8 >> > device scsi aacraid >> > device scsi megaraid >> > device scsi aic7xxx >> > device eth eepro100 >> > device eth bcm5700 >> > keyboard us >> > mouse genericwheelps/2 --device psaux >> > skipx >> > network --device eth0 --bootproto dhcp --hostname nfs-basic >> > >> > Ron >> > >> > On Tue, 2005-02-08 at 12:27, Ronald Reed wrote: >> >> No, I need the drive letter assignments to be different. >> >> >> >> Ron >> >> >> >> On Tue, 2005-02-08 at 11:58, Ed Brown wrote: >> >>> The --driveorder option won't change the drive letter assignments, >> but >> >>> it will change the drive that grub is installed to. Isn't that what >> you >> >>> are after? In your case, you probably want: >> >>> >> >>> bootloader --driveorder=sdb,sda --location=mbr >> >>> >> >>> -Ed >> >>> >> >>> >> >>> On Tue, 2005-02-08 at 10:53, Ronald Reed wrote: >> >>>> Yes, the aacraid device get set to /dev/sdb. The accraid device is >> a >> >>>> mirrored pair of 36GB drives that are inside the hardware. The >> megaraid >> >>>> device is a Raid 5 group of 4 73GB drives (3 in the array, and 1 >> hot >> >>>> spare). The megaraid is to be used for NFS file storage and the >> accraid >> >>>> is to be used for the system drive. >> >>>> >> >>>> Ron >> >>>> >> >>>> On Tue, 2005-02-08 at 11:47, Brian Long wrote: >> >>>> >> >>>>> What does it detect the aacraid device as? /dev/sdb? If so, why >> don't >> >>>>> you use the bootloader option mentioned previously to get grub to >> use >> >>>>> sdb as the boot drive? >> >>>>> >> >>>>> /Brian/ >> >>> >> >>> _______________________________________________ >> >>> 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 > -- > =========================== > Ron Reed > RedHat Certified Engineer > SGP Computer Department Manager > Unix Systems Administrator > ARM SGP CART Site > (580)388-4053 > ron.reed at arm.gov > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > From rreed at ops.sgp.arm.gov Fri Mar 4 15:07:09 2005 From: rreed at ops.sgp.arm.gov (Ronald Reed) Date: Fri, 04 Mar 2005 09:07:09 -0600 Subject: initrd.img module loading In-Reply-To: <32808.128.165.7.35.1109889849.squirrel@webmail.lanl.gov> References: <1109884458.2982.56.camel@ronnote> <32808.128.165.7.35.1109889849.squirrel@webmail.lanl.gov> Message-ID: <1109948829.3007.16.camel@ronnote> I can make it work with the workarounds that have been presented. I have to change the install so that it installs to sdb and then does Pete Eisch's steps to create a new initrd that will move sdb to sda. But it all seems like such a pain step to have to invoke when you should be able to specify the devices in the kickstart file and have it follow them. I would like to thank everyone who has responded to my problem. Ron On Thu, 2005-03-03 at 16:44, Edward F. Brown wrote: > Is there a particular reason "sda' has to be assigned to this drive? I > understand that it would be great if linux allowed more determination over > drive letter assignments, I've wrestled with this too. But is there an > issue for which a workaround hasn't been suggested? > > -Ed > > > > I created a new initrd and tried to use it as part of the install. But > > this didn't work. I understand what your steps are doing, but that > > doesn't work for what I need. > > > > I have to make sure that the aacraid module gets loaded first so that it > > will be sda. Your ideal works, but it is after the fact, which means > > that if I do it in the %post like you said, the machine would not be > > bootable, since the drives that were attached to the megaraid card were > > formated and setup as the root device and the new initrd will load the > > aacraid first, which will having nothing on it. > > > > Ron > > > > On Tue, 2005-02-22 at 13:39, Peter Eisch wrote: > >> Did you try rebuilding the initrd.img using something like: > >> > >> /sbin/mkinitrd --preload scsi_mod --preload sd_mod --with aacraid \ > >> --with megaraid --with aic7xxx --with eepro100 --with bcm5700 \ > >> /boot/ninitrd-%{KVERREL}.img %{KVERREL} > >> /bin/mv /boot/initrd-%{KVERREL}.img /boot/oinitrd-%{KVERREL}.img > >> /bin/mv /boot/ninitrd-%{KVERREL}.img /boot/initrd-%{KVERREL}.img > >> > >> I put this in my kernel's %post so it gets done at boot, but do the same > >> kind of thing with the initrd.img on the install media. > >> > >> peter > >> > >> On 2/22/05 12:38 PM, "Ronald Reed" wrote: > >> > >> > I have found that the scsihosts= boot line command is only for kernels > >> > that have the modules built into the kernel, but no other resolution > >> has > >> > been found. > >> > > >> > I have tried the device parameter in the kickstart file, but anaconda > >> > seems to ignore them. Anyone have any other ideals? Here is a portion > >> of > >> > my kickstart file: > >> > > >> > #kickstart file (partial) > >> > install > >> > cdrom > >> > lang en_US.UTF-8 > >> > langsupport --default en_US.UTF-8 en_US.UTF-8 > >> > device scsi aacraid > >> > device scsi megaraid > >> > device scsi aic7xxx > >> > device eth eepro100 > >> > device eth bcm5700 > >> > keyboard us > >> > mouse genericwheelps/2 --device psaux > >> > skipx > >> > network --device eth0 --bootproto dhcp --hostname nfs-basic > >> > > >> > Ron > >> > > >> > On Tue, 2005-02-08 at 12:27, Ronald Reed wrote: > >> >> No, I need the drive letter assignments to be different. > >> >> > >> >> Ron > >> >> > >> >> On Tue, 2005-02-08 at 11:58, Ed Brown wrote: > >> >>> The --driveorder option won't change the drive letter assignments, > >> but > >> >>> it will change the drive that grub is installed to. Isn't that what > >> you > >> >>> are after? In your case, you probably want: > >> >>> > >> >>> bootloader --driveorder=sdb,sda --location=mbr > >> >>> > >> >>> -Ed > >> >>> > >> >>> > >> >>> On Tue, 2005-02-08 at 10:53, Ronald Reed wrote: > >> >>>> Yes, the aacraid device get set to /dev/sdb. The accraid device is > >> a > >> >>>> mirrored pair of 36GB drives that are inside the hardware. The > >> megaraid > >> >>>> device is a Raid 5 group of 4 73GB drives (3 in the array, and 1 > >> hot > >> >>>> spare). The megaraid is to be used for NFS file storage and the > >> accraid > >> >>>> is to be used for the system drive. > >> >>>> > >> >>>> Ron > >> >>>> > >> >>>> On Tue, 2005-02-08 at 11:47, Brian Long wrote: > >> >>>> > >> >>>>> What does it detect the aacraid device as? /dev/sdb? If so, why > >> don't > >> >>>>> you use the bootloader option mentioned previously to get grub to > >> use > >> >>>>> sdb as the boot drive? > >> >>>>> > >> >>>>> /Brian/ > >> >>> > >> >>> _______________________________________________ > >> >>> 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 > > -- > > =========================== > > Ron Reed > > RedHat Certified Engineer > > SGP Computer Department Manager > > Unix Systems Administrator > > ARM SGP CART Site > > (580)388-4053 > > ron.reed at arm.gov > > > > _______________________________________________ > > 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 -- =========================== Ron Reed RedHat Certified Engineer SGP Computer Department Manager Unix Systems Administrator ARM SGP CART Site (580)388-4053 ron.reed at arm.gov From dan at half-asleep.com Fri Mar 4 15:20:25 2005 From: dan at half-asleep.com (Daniel Segall) Date: Fri, 4 Mar 2005 10:20:25 -0500 (EST) Subject: Display progress during long kickstart %post script In-Reply-To: <4227378C.8000208@rackspace.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> <42272F26.7080502@rackspace.com> <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> <4227378C.8000208@rackspace.com> Message-ID: <54104.192.80.55.74.1109949625.squirrel@webmail.half-asleep.com> It's really odd, but I can't get this to work properly with my dialog scripts. The first part of my %post does some basic NFS mounting, & copying of files. I see all of this fine on the console. But when I call my external post install script, a dialog is supposed to pop up saying "is this server going into nis?", and you answer yes or no. This dialog does not appear. By default it is at yes, so if I hit enter at this point, the next dialog will show up as it should. Anyone know if there is some special argument that needs to be passed prior to launching my external script? Mayber something in the script itself? Below are the sniplets of code at the beginning of the post and the end. Please excuse my sloppy code, I'm an admin not a programmer ;) Thanks, -Dan --------------- snip -------------------- %post --interpreter /bin/sh exec < /dev/tty3 > /dev/tty3 chvt 3 cat <>/etc/motd ####################################### ## KICKSTART STILL IN PROGRESS!!!!!! ## ####################################### `cat /etc/redhat-release` - Installed `date` EOF ------------------------------------------ --------------- snip -------------------- # Execute our post install echo " " >>/root/ks.log if [ -f /root/linux-postinstall.sh ]; then echo " * Executing Post-Install..." >>/root/ks.log echo " * Executing Post-Install..." chmod +x /root/linux-postinstall.sh exec /root/linux-postinstall.sh >>/root/ks.log else echo "Can't find linux-postinstall.sh." >>/root/ks.log echo "Can't find linux-postinstall.sh." fi # We done. echo " * Post-Install complete. System rebooting." >>/root/ks.log echo " * Post-Install complete. System rebooting." chvt 1 sleep 10 exit 0 #EOF -------------------------------------------- > Daniel Segall wrote: >> Hey Michael, > > Hi Dan! > >> Speaking of this,... have you seen any problems doing this method with >> RHES4 or FC3? I use a similar method (tty3), and ever since FC3, the >> display doesn't show up automatically. You have to hit enter blindly for >> my first (second really) dialog prompt to show up. Do you think that by >> using tty6 or another unused tty would resolve this? > > Using tty6 works perfectly in RHEL-4, and I would expect that FC-3 > should work fine, although we do not maintain a Fedora kickstart. > >> Hope all is well at the rack. I miss you guys... > > The rack is very well, and you are missed, too ;) > > Warm Regards, > Michael Shuler > Rackspace Managed Hosting > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > From iain at uk.colt.net Fri Mar 4 16:31:19 2005 From: iain at uk.colt.net (Iain Conochie) Date: Fri, 04 Mar 2005 16:31:19 +0000 Subject: Static route setup Message-ID: <42288D57.9000408@uk.colt.net> Hello, I am at present upgrading some IBM blades from RedHat AS2.1 to RedHat AS3.0. I am using basically the same ks.cfg file that was used for the 2.1 install, and all seems to progress smoothly. However, when the server reboots, I am finding that the static routes I am attempting to configure are broken. for example cat /etc/sysconfig/static-routes eth1 host 10.170.127.10/32 gw 10.170.67.121 netstat -rn Kernel IP routing table Destination Gateway Genmask Flags MSS Window irtt Iface 10.170.67.128 0.0.0.0 255.255.255.192 U 0 0 0 eth0 10.170.67.0 0.0.0.0 255.255.255.128 U 0 0 0 eth1 169.254.0.0 0.0.0.0 255.255.0.0 U 0 0 0 eth1 127.0.0.0 0.0.0.0 255.0.0.0 U 0 0 0 lo 0.0.0.0 10.170.67.185 0.0.0.0 UG 0 0 0 eth0 So it looks like the /etc/sysconfig/static-routes file is not in the format expected, as I have a bogus route for 169.254.0.0 network. Can anyone point out where I am going wrong in the config? Cheers Iain Iain Conochie UNIX Systems Administrator COLT Telecommunications PLC From email at jasonkohles.com Fri Mar 4 19:16:01 2005 From: email at jasonkohles.com (Jason Kohles) Date: Fri, 4 Mar 2005 14:16:01 -0500 Subject: Static route setup In-Reply-To: <42288D57.9000408@uk.colt.net> References: <42288D57.9000408@uk.colt.net> Message-ID: <20050304191600.GA10497@mail.jasonkohles.com> On Fri, Mar 04, 2005 at 04:31:19PM +0000, Iain Conochie wrote: > Hello, > > I am at present upgrading some IBM blades from RedHat AS2.1 to RedHat > AS3.0. I am using basically the same ks.cfg file that was used for the 2.1 > install, and all seems to progress smoothly. However, when the server > reboots, I am finding that the static routes I am attempting to configure > are broken. for example > This bit me too, the static route configuration changed between AS2.1 and AS3, so you can't do them the same way... This is how I solved the problem... # kickstart-post-static-routes TAG="# Static routes added by kickstart" function static_route() { DEST=$1 GW=$2 DEV=$3 case `rpm -qf /etc/redhat-release` in *3AS*) FILE="/etc/sysconfig/network-scripts/route-$DEV" ROUTE="$DEST via $GW dev $INT" ;; *2.1AS*) FILE="/etc/sysconfig/static-routes" ROUTE="$DEV host $DEST gw $GW" ;; esac grep -q $TAG $FILE || echo $TAG >> $FILE echo $ROUTE >> $FILE } static_route 10.170.127.10/32 10.170.67.121 eth1 static_route 192.168.42.1/24 10.170.67.122 eth1 # ... etc ... > So it looks like the /etc/sysconfig/static-routes file is not in the > format expected, as I have a bogus route for 169.254.0.0 network. Can > anyone point out where I am going wrong in the config? > AS3 doesn't even make any attempt to read /etc/sysconfig/static-routes as far as I can tell, but so far I haven't tracked down where the 168.245.0.0 route is coming from, I have it on most of my boxes as well. -- Jason Kohles A witty saying proves nothing. email at jasonkohles.com -- Voltaire (1694 - 1778) http://www.jasonkohles.com/ From nicolas.catania at hp.com Fri Mar 4 19:31:18 2005 From: nicolas.catania at hp.com (niko) Date: Fri, 4 Mar 2005 11:31:18 -0800 Subject: Post reboot commands (howto?) Message-ID: <16936.46982.639620.403752@localhost.localdomain> Hello, I am using kickstart (EL3) to setup machines over the network. I would like to be able to run a script only once when the machine reboots after a successful installation (script must run with the right kernel and so on...) The %post sequence is executed before the reboot so that is not doing it. I was thinking of adding a 'runonce' file in /etc/init.d to do the work and remove itself. Has anyone a better solution? Niko -- Nicolas Catania HP Change & Configuration Software Operation +1 408 447-4564 nicolas.catania at hp.com From jkeating at j2solutions.net Fri Mar 4 19:39:13 2005 From: jkeating at j2solutions.net (Jesse Keating) Date: Fri, 04 Mar 2005 11:39:13 -0800 Subject: Post reboot commands (howto?) In-Reply-To: <16936.46982.639620.403752@localhost.localdomain> References: <16936.46982.639620.403752@localhost.localdomain> Message-ID: <1109965153.12887.62.camel@jkeating2.hq.pogolinux.com> On Fri, 2005-03-04 at 11:31 -0800, niko wrote: > I am using kickstart (EL3) to setup machines over the network. I would > like to > be able to run a script only once when the machine reboots after a > successful > installation (script must run with the right kernel and so on...) > > The %post sequence is executed before the reboot so that is not doing > it. > I was thinking of adding a 'runonce' file in /etc/init.d to do the > work and > remove itself. Has anyone a better solution? I do pretty close to this. I'll backup the original /etc/rc.local file in the kickstart %post, and drop a new rc.local file in that points to the script I want ran after boot. The end of my script will replace the backed up rc.local file. -- Jesse Keating RHCE (geek.j2solutions.net) Fedora Legacy Team (www.fedoralegacy.org) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) Was I helpful? Let others know: http://svcs.affero.net/rm.php?r=jkeating From daniel.spratlen at cox.net Fri Mar 4 20:02:42 2005 From: daniel.spratlen at cox.net (Daniel Spratlen) Date: Fri, 04 Mar 2005 15:02:42 -0500 Subject: Static route setup In-Reply-To: <20050304191600.GA10497@mail.jasonkohles.com> References: <42288D57.9000408@uk.colt.net> <20050304191600.GA10497@mail.jasonkohles.com> Message-ID: <1109966562.4509.20.camel@boo> On Fri, 2005-03-04 at 14:16 -0500, Jason Kohles wrote: > > So it looks like the /etc/sysconfig/static-routes file is not in the > > format expected, as I have a bogus route for 169.254.0.0 network. Can > > anyone point out where I am going wrong in the config? > > > AS3 doesn't even make any attempt to read /etc/sysconfig/static-routes > as far as I can tell, but so far I haven't tracked doNOZEROCONF=yeswn where the > 168.245.0.0 route is coming from, I have it on most of my boxes as well. > THe 169.254.0.0 route comes from zeroconf networking. To disable this put NOZEROCONF=yes in /etc/sysconfig/network. http://www.linux-mag.com/2003-08/zeroconf_01.html thanks, Daniel From phr at doc.ic.ac.uk Fri Mar 4 22:49:30 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Fri, 4 Mar 2005 22:49:30 +0000 (GMT) Subject: Display progress during long kickstart %post script In-Reply-To: <54104.192.80.55.74.1109949625.squirrel@webmail.half-asleep.com> References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> <42272F26.7080502@rackspace.com> <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> <4227378C.8000208@rackspace.com> <54104.192.80.55.74.1109949625.squirrel@webmail.half-asleep.com> Message-ID: On Fri, 4 Mar 2005, Daniel Segall wrote: >--------------- snip -------------------- > ># Execute our post install >echo " " >>/root/ks.log >if [ -f /root/linux-postinstall.sh ]; then > echo " * Executing Post-Install..." >>/root/ks.log > echo " * Executing Post-Install..." > chmod +x /root/linux-postinstall.sh > exec /root/linux-postinstall.sh >>/root/ks.log ^^^^ What's this? >else > echo "Can't find linux-postinstall.sh." >>/root/ks.log > echo "Can't find linux-postinstall.sh." >fi > ># We done. >echo " * Post-Install complete. System rebooting." >>/root/ks.log >echo " * Post-Install complete. System rebooting." >chvt 1 >sleep 10 >exit 0 >#EOF When you hit that exec, you're not in Kansas any more, and specifically the final "Post-Install complete" section will never execute. The original question was about dialog, but there's no mention in the script?? Cheers, Phil From dan at half-asleep.com Fri Mar 4 23:06:32 2005 From: dan at half-asleep.com (Daniel Segall) Date: Fri, 4 Mar 2005 18:06:32 -0500 (EST) Subject: Display progress during long kickstart %post script In-Reply-To: References: <20050303151050.57971.qmail@web26306.mail.ukl.yahoo.com> <42272F26.7080502@rackspace.com> <55205.192.80.55.74.1109865358.squirrel@webmail.half-asleep.com> <4227378C.8000208@rackspace.com> <54104.192.80.55.74.1109949625.squirrel@webmail.half-asleep.com> Message-ID: <49428.192.80.55.73.1109977592.squirrel@webmail.half-asleep.com> Thanks for the reply Phil. The dialog commands are all in the linux-postinstall.sh script that is being executed (see sniplet below). This has been working as expected with FC1 & RHES3. The main reason this is done in an external script, rather than within the kickstart file itself, is because we use the same post-instal script on machines that were not kicked, and I don't want to have to maintain different code for each kickstart as well. --------------- snip -------------------- dialog --backtitle "$BT" --title "NIS" --yesno "Is this system going into NIS?" 8 60 if [ $? -eq 0 ] then IPADDR=`dialog --backtitle "$BT" --title "IP Address?" --stdout --inputbox "Enter your IP address" 8 60` HNAME=`dialog --backtitle "$BT" --title "Hostname?" --stdout --inputbox "Enter yourhostname (without domain)" 8 60` else dialog --backtitle "$BT" --title "See Ya!" --msgbox "Your on your own, Big Admin" 860 exit fi ----------------------------------------- -Dan > On Fri, 4 Mar 2005, Daniel Segall wrote: > >>--------------- snip -------------------- >> >># Execute our post install >>echo " " >>/root/ks.log >>if [ -f /root/linux-postinstall.sh ]; then >> echo " * Executing Post-Install..." >>/root/ks.log >> echo " * Executing Post-Install..." >> chmod +x /root/linux-postinstall.sh >> exec /root/linux-postinstall.sh >>/root/ks.log > ^^^^ What's this? >>else >> echo "Can't find linux-postinstall.sh." >>/root/ks.log >> echo "Can't find linux-postinstall.sh." >>fi >> >># We done. >>echo " * Post-Install complete. System rebooting." >>/root/ks.log >>echo " * Post-Install complete. System rebooting." >>chvt 1 >>sleep 10 >>exit 0 >>#EOF > > When you hit that exec, you're not in Kansas any more, and specifically > the final "Post-Install complete" section will never execute. The > original question was about dialog, but there's no mention in the > script?? > > > Cheers, > Phil > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > From athos10 at altern.org Sat Mar 5 10:31:09 2005 From: athos10 at altern.org (athos10) Date: Sat, 05 Mar 2005 11:31:09 +0100 Subject: Post reboot commands (howto?) In-Reply-To: <16936.46982.639620.403752@localhost.localdomain> References: <16936.46982.639620.403752@localhost.localdomain> Message-ID: <1110018669.31388.3.camel@christelle> Hi, May modify firstboot script and package. If you can't or don't want to modify rpm, you could modify firstboot script using %post. Friendly Athos10 Michael Aubertin michael.aubertin at free.fr athos10 at altern.org Le vendredi 04 mars 2005 ? 11:31 -0800, niko a ?crit : >Hello, > >I am using kickstart (EL3) to setup machines over the network. I would like to >be able to run a script only once when the machine reboots after a successful >installation (script must run with the right kernel and so on...) > >The %post sequence is executed before the reboot so that is not doing it. >I was thinking of adding a 'runonce' file in /etc/init.d to do the work and >remove itself. Has anyone a better solution? > >Niko > From Teemu.M.Jalonen at nokia.com Mon Mar 7 10:25:03 2005 From: Teemu.M.Jalonen at nokia.com (Teemu.M.Jalonen at nokia.com) Date: Mon, 7 Mar 2005 12:25:03 +0200 Subject: Static route setup Message-ID: > cat /etc/sysconfig/static-routes > eth1 host 10.170.127.10/32 gw 10.170.67.121 you can create file /etc/sysconfig/network-scripts/eth1.route with content ADDRESS0=10.170.127.10 NETMASK0=255.255.255.255 GATEWAY0=10.170.67.121 And if you have other routes for eth1, then: ADDRESS1=.... NETMASK1=.... GATEWAY1=.... From iain at uk.colt.net Mon Mar 7 14:34:06 2005 From: iain at uk.colt.net (Iain Conochie) Date: Mon, 07 Mar 2005 14:34:06 +0000 Subject: Static route setup In-Reply-To: References: Message-ID: <422C665E.8020505@uk.colt.net> Thank you very much sir! This is exactly what I needed Cheers Iain Teemu.M.Jalonen at nokia.com wrote: >>cat /etc/sysconfig/static-routes >>eth1 host 10.170.127.10/32 gw 10.170.67.121 >> >> > >you can create file /etc/sysconfig/network-scripts/eth1.route with content >ADDRESS0=10.170.127.10 >NETMASK0=255.255.255.255 >GATEWAY0=10.170.67.121 > >And if you have other routes for eth1, then: >ADDRESS1=.... >NETMASK1=.... >GATEWAY1=.... > >_______________________________________________ >Kickstart-list mailing list >Kickstart-list at redhat.com >https://www.redhat.com/mailman/listinfo/kickstart-list > > From bedouglas at earthlink.net Mon Mar 7 20:11:55 2005 From: bedouglas at earthlink.net (bruce) Date: Mon, 7 Mar 2005 12:11:55 -0800 Subject: remotely installing FC using vnc ... Message-ID: <032101c52351$e8e78fa0$0301a8c0@Mesa.com> hi... i'm trying to remotely install/upgrade a remote server. i am not a linux admin.. i have no idea as to what a 'boot option' is, nor where it should go, or what file it should be in!!! that said, if i can find the right information, or perhaps someone to talk to, i can probably get this process to work, and get rid of my frustration!!!!!!! thanks for any pointers/help/assistance with this... and please don't say 'use yum/apt-get', unless you can give me pointers as to how to clear up issues that i've had with yum!!!! thanks.... bruce From phr at doc.ic.ac.uk Mon Mar 7 20:42:22 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Mon, 7 Mar 2005 20:42:22 +0000 (GMT) Subject: remotely installing FC using vnc ... In-Reply-To: <032101c52351$e8e78fa0$0301a8c0@Mesa.com> References: <032101c52351$e8e78fa0$0301a8c0@Mesa.com> Message-ID: On Mon, 7 Mar 2005, bruce wrote: >i'm trying to remotely install/upgrade a remote server. [snip] With no physical access, either: - Use PXE to boot from the network, or - Configure the existing bootloader (LILO/GRUB) to launch the installer Perhaps a good starting point would be the installation guide: http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/x8664-multi-install-guide/ (This covers RHEL4, rather than FC3; they're close enough and I can't seem to find a link for the FC3 installation guide.) Cheers, Phil From rsweet at aoes.com Tue Mar 8 08:24:43 2005 From: rsweet at aoes.com (Ryan Sweet) Date: Tue, 8 Mar 2005 09:24:43 +0100 (CET) Subject: remotely installing FC using vnc ... In-Reply-To: <032101c52351$e8e78fa0$0301a8c0@Mesa.com> References: <032101c52351$e8e78fa0$0301a8c0@Mesa.com> Message-ID: On Mon, 7 Mar 2005, bruce wrote: > hi... > > i'm trying to remotely install/upgrade a remote server. i am not a linux > admin.. i have no idea as to what a 'boot option' is, nor where it should > go, or what file it should be in!!! Here you go: - find the "isolinux" dir from FC3. "cd" to it (e.g. "cd /home/bruce/fc3/i386/isolinux"). - as the root user, copy "vmlinuz" and "initrd.img" to your /boot directory. IMPORTANT: use different file names in the target dir: cp vmlinuz /boot/vmlinuz-fc3install cp initrd.img /boot/initrd-fc3install.img - if you are upgrading a server that has lilo (older redhat, you probably don't have this as it hasn't been the default since redhat 7.x) then edit /etc/lilo.conf, and copy one of the existing kernel and initrd entries, modifying it to use the files that you just copied to /boot/. Then run "/sbin/lilo". -if you are upgrading a server that boots with grub (you probably are), edit /etc/grub.conf as follows: -- title Fedora Core 3 installation root (hd0,0) kernel /vmlinuz-fc3install console=ttyS0 vnc ip=192.168.1.120 vnc-connect=192.168.2.54 initrd /initrd-fc3install.img -- Make sure that you replace the "root (hd0,0)" line with one that matches your existing grub.conf. Also replace the ip parameters with ones appropriate for your networks (you could also try ip=dhcp is you are using that). You may want to do a kickstart install, in which case you should search redhat.com for "kickstart installation", which has very good documentation. To learn more about the boot options that you can put on the "kernel" line see the linux kernel bootprompt howto: http://www.ibiblio.org/mdw/HOWTO/BootPrompt-HOWTO.html Once you create the new entry in /etc/grub.conf, make sure that you set the "default" parameter in /etc/grub.conf to default to your new entry (this is easiest if you make your new entry the first one and then set "default" to 0). The "vnc-connect" parameter (see the kickstart documentation) tells the system to start the GUI installer in a vncserver and to connect to the vncviewer listening on the specified address. To do this you must be able to open vnc connections from the remote site to your workstation. Once you have setup the installer, reboot the system and on your workstation run "vncviewer -listen". If everything goes correctly then the system should start the gui installer and connect it to your vncviewer. The argument "console=ttyS0" tells the installer to start on serial console 0. (usually com1 for dos/win32 folks). If you have serial console access and the hardware supports it, then you can do a text based install that way (remove the vnc arguments). good luck - -ryan > > that said, if i can find the right information, or perhaps someone to talk > to, i can probably get this process to work, and get rid of my > frustration!!!!!!! > > thanks for any pointers/help/assistance with this... > > and please don't say 'use yum/apt-get', unless you can give me pointers as > to how to clear up issues that i've had with yum!!!! > > thanks.... > > bruce > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -- Ryan Sweet Advanced Operations and Engineering Services AOES Group BV http://www.aoes.com Phone +31(0)71 5795521 Fax +31(0)71572 1277 From kurt at netdirect.ca Thu Mar 10 20:53:20 2005 From: kurt at netdirect.ca (Kurt Bishop) Date: Thu, 10 Mar 2005 15:53:20 -0500 Subject: Problems with Network-Based Install Message-ID: <1110488000.11417.37.camel@excelsior> Hi, We have been having some problems getting a Kickstart install to be 100% automated. We are PXE booting the system which then fetches the Kickstart file from a remote NFS share. However, the system we're installing to has 2 network devices and the installer always pops up and asks us which device it should use before fetching the Kickstart file. Manually selecting a device moves the install along just fine but we need this to be automated. I have tried adding the following: network ksdevice=eth0 ip=dhcp to the kernel parameters but the installer simply hangs at the "running /sbin/loader" line. The other virtual consoles don't show anything interesting either. Console 3 shows "* getting kickstart file" and console 4 just shows messages about the interfaces coming online. Here's some details on our environment: - RedHat Enterprise AS 3 Update 4 - using the pxeboot initrd & kernel from RHEL Disc 1 to get the system booted Any help here would be greatly appreciated. Thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From pbdlists at pinboard.com Thu Mar 10 15:25:16 2005 From: pbdlists at pinboard.com (pbdlists at pinboard.com) Date: Thu, 10 Mar 2005 16:25:16 +0100 Subject: RHEL-4 fails to asign IP Message-ID: <20050310162516.A53873@pinboard.com> I wonder whether I'm the only one having this problem (and thus probably doing something wrong) of if there really is a problem. We have a Satellite Server and we've been kickstarting servers from it for months. No problem. For various tests I probably kickstarted machines from within vmware well over a hundred times; boot off the RHEL-iso disk1, type 'linux ks=http://ourserver/path/to/label', wait for the box to appear to input IP, netmask, gateway and DNS (we have no DHCP in that network segment and there is no chance there ever will be one). The same procedure does not work any more for RHEL-4-WS; the IP settings entered in the box are not applied. After a timeout I can only continue with the CD installation. Switching to tty2, there is an eth0, but without IP. Assigning the IP and routes in that shell works and the interface is usable. But this is not a workaround, as switching to a shell can only be done after selecting to continue with the CD installation. - checksum of the ISO image used is fine - same problem with RHEL-4-WS and RHEL-4-DT (desktop) - same procedure on the same machine works file for RHEL-3-WS Unfortunately I don't have a physical machine to see whether the same thing happens there. What the heck am I doing wrong? Kurt From phr at doc.ic.ac.uk Thu Mar 10 23:21:47 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Thu, 10 Mar 2005 23:21:47 +0000 (GMT) Subject: Problems with Network-Based Install In-Reply-To: <1110488000.11417.37.camel@excelsior> References: <1110488000.11417.37.camel@excelsior> Message-ID: On Thu, 10 Mar 2005, Kurt Bishop wrote: >We have been having some problems getting a Kickstart install to be >100% automated. We are PXE booting the system which then fetches the >Kickstart file from a remote NFS share. However, the system we're >installing to has 2 network devices and the installer always pops up >and asks us which device it should use before fetching the Kickstart >file. Manually selecting a device moves the install along just fine but >we need this to be automated. I have tried adding the following: > >network ksdevice=eth0 ip=dhcp > >to the kernel parameters but the installer simply hangs at the >"running /sbin/loader" line. What about ksdevice=link ? Other things to consider: - Do you have consistent parts of anaconda, i.e. do the initrd and vmlinuz files come from the same minor distro (RHEL3U4) as the NFS-shared second stage? - Does syslog (/var/log/messages) on the NFS server say anything useful? Cheers, Phil From kurt at netdirect.ca Fri Mar 11 02:22:58 2005 From: kurt at netdirect.ca (Kurt Bishop) Date: Fri, 11 Mar 2005 02:22:58 +0000 GMT Subject: Problems with Network-Based Install Message-ID: <1915797695-1110507771-cardhu_blackberry.rim.net-6732-@engine37> Hi Philip, Thanks for the hints. The initrd & vmlinuz were pulled from the same distro CDs as the install being done. The NFS server is a Windows box (I know I know) but I can check the logs for anything interesting. I will also try ksdevice=link in the morning and report back. Thanks again. -----Original Message----- From: Philip Rowlands Date: Thu, 10 Mar 2005 23:21:47 To:Discussion list about Kickstart Subject: Re: Problems with Network-Based Install On Thu, 10 Mar 2005, Kurt Bishop wrote: >We have been having some problems getting a Kickstart install to be >100% automated. We are PXE booting the system which then fetches the >Kickstart file from a remote NFS share. However, the system we're >installing to has 2 network devices and the installer always pops up >and asks us which device it should use before fetching the Kickstart >file. Manually selecting a device moves the install along just fine but >we need this to be automated. I have tried adding the following: > >network ksdevice=eth0 ip=dhcp > >to the kernel parameters but the installer simply hangs at the >"running /sbin/loader" line. What about ksdevice=link ? Other things to consider: - Do you have consistent parts of anaconda, i.e. do the initrd and vmlinuz files come from the same minor distro (RHEL3U4) as the NFS-shared second stage? - Does syslog (/var/log/messages) on the NFS server say anything useful? Cheers, Phil _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list ------------------------------- Kurt Bishop Systems Analyst/Consultant Net Direct Inc. 12-564 Weber Street North Waterloo, ON N2L 5C6 Ph: 519-883-1172 x5104 Fx: 519-883-8533 http://www.netdirect.ca From andrew.w.robinson at mms.gov Fri Mar 11 15:42:41 2005 From: andrew.w.robinson at mms.gov (Robinson, Andrew W.) Date: Fri, 11 Mar 2005 08:42:41 -0700 Subject: When and where is /dev/cdrom created? Message-ID: <379313C94B76D2119AB60008C7A402E409E46375@IMSNOLAA> Could someone tell me when and where the link /dev/cdrom is created by the system? As part of my kickstart installation of a RHEL3 system, I want to create the link /dev/dvd. Experimenting on my existing system, I found I could point it to the same dev file that the link /dev/cdrom points to: Ln -s $( ls -l /dev/cdrom | sed 's/^.*-> //' ) /dev/dvd I put this command in the %post section of my kickstart file. If failed however, apparently because the link /dev/cdrom did not yet exist. When I log in the first time after the installation has completed, the link /dev/cdrom does exist. So somewhere in the boot process the system creates the link. I've poked around the system, but haven't found where. One approach to my problem might be to create an initrc script with a high chkconfig number that runs the above command late in the boot sequence. But that seems like a sledgehammer approach. If I knew when in the boot sequence /dev/cdrom was created and by what software, I am hoping to craft a more elegant solution. So, anyone know where /dev/cdrom is created? Thanks! Andrew Robinson From gregcope at gmail.com Fri Mar 11 16:40:54 2005 From: gregcope at gmail.com (Greg Cope) Date: Fri, 11 Mar 2005 16:40:54 +0000 Subject: kickstart stall - dell 1650 Message-ID: Hi All, I am trying to kickstart the install of a few dell sc1650's. I can get it to PXEboot ok, but it stalls just after booting with: 54 keymaps are avaliable Any ideas on how to debug this? Greg From davidbartlett2003 at yahoo.co.uk Fri Mar 11 16:45:22 2005 From: davidbartlett2003 at yahoo.co.uk (David Bartlett) Date: Fri, 11 Mar 2005 16:45:22 +0000 (GMT) Subject: When and where is /dev/cdrom created? In-Reply-To: 6667 Message-ID: <20050311164522.96146.qmail@web26303.mail.ukl.yahoo.com> > So, anyone know where /dev/cdrom is created? If you are running Fedora Core 3, this may be due to a bug in a version of the udev subsystem (device manager which creates device nodes as needed). See http://www.mjmwired.net/resources/mjm-fedora-fc3.shtml#cdrom for a quick note on this. If you are not running Fedora Core 3, then delete this post! Rgds, David Bartlett. Send instant messages to your online friends http://uk.messenger.yahoo.com From phr at doc.ic.ac.uk Fri Mar 11 19:30:51 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Fri, 11 Mar 2005 19:30:51 +0000 (GMT) Subject: When and where is /dev/cdrom created? In-Reply-To: <379313C94B76D2119AB60008C7A402E409E46375@IMSNOLAA> References: <379313C94B76D2119AB60008C7A402E409E46375@IMSNOLAA> Message-ID: On Fri, 11 Mar 2005, Robinson, Andrew W. wrote: >Could someone tell me when and where the link /dev/cdrom is created by >the system? updfstab, run by kudzu. My kickstart %post disables kudzu, so I have to run updfstab in lieu. Cheers, Phil From James.Oden at tekelec.com Fri Mar 11 19:40:26 2005 From: James.Oden at tekelec.com (Oden, James) Date: Fri, 11 Mar 2005 13:40:26 -0600 Subject: When and where is /dev/cdrom created? Message-ID: <58292FA6B3EEFD49AEDAF6597E21E717CE870B@DCEVS2.tekelec.com> > -----Original Message----- > From: kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com] On Behalf Of > Philip Rowlands > Sent: Friday, March 11, 2005 2:31 PM > To: Discussion list about Kickstart > Subject: Re: When and where is /dev/cdrom created? > > On Fri, 11 Mar 2005, Robinson, Andrew W. wrote: > > >Could someone tell me when and where the link /dev/cdrom is > created by > >the system? > > updfstab, run by kudzu. My kickstart %post disables kudzu, so > I have to > run updfstab in lieu. > Thanks Phil. I have wondered for the longest why my kickstarts never had the /dev/cdrom links. It never was a real big deal to me, but this would definitely fix it. Cheers...james > > Cheers, > Phil > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > From strange at nsk.no-ip.org Fri Mar 11 19:46:22 2005 From: strange at nsk.no-ip.org (Luciano Miguel Ferreira Rocha) Date: Fri, 11 Mar 2005 19:46:22 +0000 Subject: When and where is /dev/cdrom created? In-Reply-To: <58292FA6B3EEFD49AEDAF6597E21E717CE870B@DCEVS2.tekelec.com> References: <58292FA6B3EEFD49AEDAF6597E21E717CE870B@DCEVS2.tekelec.com> Message-ID: <20050311194622.GE5475@nsk.no-ip.org> On Fri, Mar 11, 2005 at 01:40:26PM -0600, Oden, James wrote: > > -----Original Message----- > > From: kickstart-list-bounces at redhat.com > > [mailto:kickstart-list-bounces at redhat.com] On Behalf Of > > Philip Rowlands > > Sent: Friday, March 11, 2005 2:31 PM > > To: Discussion list about Kickstart > > Subject: Re: When and where is /dev/cdrom created? > > > > On Fri, 11 Mar 2005, Robinson, Andrew W. wrote: > > > > >Could someone tell me when and where the link /dev/cdrom is > > created by > > >the system? > > > > updfstab, run by kudzu. My kickstart %post disables kudzu, so > > I have to > > run updfstab in lieu. > > > Thanks Phil. I have wondered for the longest why my kickstarts never > had the > /dev/cdrom links. It never was a real big deal to me, but this would > definitely fix it. Why don't you just make the dvd link point to the cdrom link? Regards, Luciano Rocha From jerome.tournier at idealx.com Tue Mar 15 13:55:04 2005 From: jerome.tournier at idealx.com (Jerome Tournier) Date: Tue, 15 Mar 2005 14:55:04 +0100 Subject: usb disk detected before scsi controller Message-ID: <20050315135504.GD3100@idealx.com> Hi all, i have a problem with my kickstart installation: i am installing my linux with an external usb disk. The problem is that the usb disk is detected as /dev/sda and the scsi disk as /dev/sdb and /dev/sdc. Is there a way to specify anaconda (via ks) to first load the ips module for scsi disk and then the usb one ? Thanks for any help ! -- Jerome From john.j.poole at usa-spaceops.com Tue Mar 15 16:25:15 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Tue, 15 Mar 2005 11:25:15 -0500 Subject: e1000 kickstart, still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB987@usaflcms03.usa-spaceops.ksc.nasa.gov> Greetings, I load/reload our PC's remotely(RHEL 3WS update 4 currently). I copy my kickstart file, vmlinuz and initrd.img to /boot. I then modify grub to boot off the /boot/install-vmlinuz and get kickstart file from /boot locally. In the KS file, I can install via NFS or HTTP. I have managed to hack(with help from some kind folks on the web) the initrd.img and install a "loader" file with patches to fix e1000 kickstart problems. NFS works but HTTP still the first attempt to download the install image. Clicking OK on the fails screen causes it to resume and work ok. I just read the new anaconda-9.1.4.1-1 fixes the e1000 problem. I installed anaconda-9.1.4.1-1 on my PC and did a "make". I then copied the newly created "loader" into the initrd.img file from the RHEL update 4 /isolinux directory. HTTP still fails on the first attempt and then picks up after selecting OK. I have really struggled to get these machines loaded without human intervention. I am an Admin type and not a programmer and was elated when I managed to get an NFS install to work. I ultimately would like to use HTTP. Could someone advise what I am missing or doing wrong? I guess I need a step by step guide. This problem seems to have lingered a long time with no easy fix/workaround provided for us dummies. Any assistance would be appreciated. Thanks, John Poole John J. Poole Mail Stop: USK-615 Phone: 321-861-0561 Email: john.j.poole at usa-spaceops.com Opinions expressed are my own and are not endorsed by United Space Alliance. From ebrown at lanl.gov Tue Mar 15 18:06:44 2005 From: ebrown at lanl.gov (Edward F. Brown) Date: Tue, 15 Mar 2005 11:06:44 -0700 (MST) Subject: e1000 kickstart, still In-Reply-To: <2CE66A1ABBED8C4B85DFF5E0860987A32CB987@usaflcms03.usa-spaceops.ksc.na sa.gov> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB987@usaflcms03.usa-spaceops.ksc.nasa.gov> Message-ID: <32842.128.165.7.76.1110910004.squirrel@webmail.lanl.gov> > This problem seems to have lingered a long time with no easy > fix/workaround provided for us dummies. Any assistance would be > appreciated. It has indeed been around a long time now. I thought Update 4 fixed it, but we have also seen it since then. Have you investigated the settings on the gigabit switch that the box is plugged into? Disabling "spanning tree" seems to help. Some users have resorted to putting a small 10/100 hub inline between the box being built and the switch, not a very good solution, especially in your remote build situation... -Ed From Robert.McKenzie at csr.com Wed Mar 16 13:00:03 2005 From: Robert.McKenzie at csr.com (Robert McKenzie) Date: Wed, 16 Mar 2005 13:00:03 -0000 Subject: Kickstart problems... Message-ID: I have been spending way too much time trying to get a functional pxe kickstart server going. I've been through at least 4 or 5 howtos on this and still come up to the same end each time. The problem is... The new system loads it's kernel from the server using PXE with no problem, but then it insists on trying to get a DHCP address (dynamic). I have in my dhcpd.conf a static IP for the host, based on MAC but this seems only to work for the first step of the PXE boot. What am I doing wrong? (I know, it's a loaded question). My entire PXE config is at http://thehostedbox.com/~rmckenzi/pxe-kickstart/server.tgz What I've included in the file (for those that have time to have a look for me) is /etc/dhcpd.conf and the entire /tftpboot/ directory (minus the actual initrd.img and vmlinuz images from the FC3 directory. I can only think it's something simple that I'm missing or have done wrong. It would be greatly appreciated for anyone that can have a look and offer some advice. Cheers!!! Robert McKenzie UNIX Support CSR Churchill House Cowley Road Business Park Cambridge CB4 0WZ T. 01223 692219 F. 01223 692001 Robert.McKenzie at csr.com www.csr.com ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From Gordon.Keegan at FMR.COM Wed Mar 16 17:36:02 2005 From: Gordon.Keegan at FMR.COM (Keegan, Gordon) Date: Wed, 16 Mar 2005 12:36:02 -0500 Subject: small custom CD issue/question Message-ID: <386604AA76CB7F4CB15FC4B4839EDC3F154223@MSGMMKDRC2WIN.DMN1.FMR.COM> I have built a custom CD and ks.cfg file to do an automated install and setup. It's a fairly stripped-down package set with the Core and Base groups and a few additional individual rpm's. The only odd thing is that when it's done and returns to the "Complete, press to reboot" screen, pressing does nothing. I have to ++ to reboot. I'm hoping that someone has seen this behavior before (and figured out what caused it!) Below are the isolinux.cfg file and ks.cfg files. Thanks in advance! Gordon Keegan isolinux.cfg: ------------ serial 0 9600 default local prompt 1 timeout 600 display boot.msg F1 boot.msg F2 examples.msg F3 help.msg label local localboot 0x80 label install kernel vmlinuz append initrd=initrd.img ks=cdrom:/ks.cfg local text label install-serial kernel vmlinuz append initrd=initrd.img ks=cdrom:/ks.cfg local text console=tty0 console=ttyS0,9600n8 label rescue kernel vmlinuz append initrd=initrd.img text rescue label noks kernel vmlinuz append initrd=initrd.img ks.cfg: ------ install lang en_US langsupport en_US keyboard us mouse none rootpw --iscrypted $1encryptedrootpasswordhashstring. text cdrom firewall --disabled authconfig --enableshadow --enablemd5 timezone --utc America/New_York bootloader --linear --location=mbr ## ## Don't want X and don't probe for a monitor ## skipx #xconfig --monitor=vga ## ## parts.ks is created below in the %pre section ## %include /tmp/parts.ks ## ## Here is our package list ## '-' at start of the package name indicates exclusion. ## This ought to remove packages that are part of Base ## We include kernel-* because RHN cannot provision those ## due to the up2date PkgSkipList ## %packages @Base compat-libstdc++ XFree86-libs glibc-headers libcap libgcj libpcap libtool-libs lockdev m4 ntp perl-CGI perl-CPAN perl-DB_File perl-suidperl sendmail-cf strace usbutils zsh ##---------------------------------------------------------------------- -----## %post --nochroot ## re-create /mnt/source to get at CDROM if [ -e /tmp/cdrom ]; then mkdir /mnt/source mount /tmp/cdrom /mnt/source fi ## for later reference cat /proc/cmdline > /mnt/sysimage/var/tmp/ks-cmdline cat /mnt/source/ks.cfg > /mnt/sysimage/var/tmp/ks.cfg ## setup post-install scripts/dirs for chroot'ed postinstall stuff if [ -e /mnt/source/post/postinstall.tar ]; then mkdir /mnt/sysimage/var/tmp/post cd /mnt/source/post tar -cf - . | (cd /mnt/sysimage/var/tmp/post; tar -xvf -) cd / fi if [ -e /tmp/cdrom ]; then umount /mnt/source fi ##---------------------------------------------------------------------- -----## %post chvt 3 if grep nopost /var/tmp/ks-cmdline; then echo "" echo " * found 'nopost' install option ... postinstall skipped" echo "" else echo "" echo "Configuring tmpfs" echo "--------------------" if ! grep '/tmp' /etc/fstab ; then echo " * added a tmpfs entry to /etc/fstab" echo "tmpfs /tmp tmpfs size=1024m 1 2" >> /etc/fstab fi echo "" echo "Postinstall Stuff" echo "--------------------" if [ -s /var/tmp/post/postinstall.tar ]; then echo " * found postinstall.tar" echo " * extracting postinstall.tar" cd /var/tmp/post tar -xvf postinstall.tar fi ## ## execute post-install scripts ## if [ -f /var/tmp/post/postinstall/postinstall.sh ]; then echo " * executing postinstall" echo "" cd /var/tmp/post/postinstall bash ./postinstall.sh else echo " * ERROR - unable to acquire & execute rh-postinstall" fi sleep 2 echo "##############################" echo "## CONFIGURATION COMPLETE " `date` echo "##############################" chvt 1 ##---------------------------------------------------------------------- -----## %pre ## ## Dynamic Partitioning ## The output of this script is %include(d) above ## NEWDEV=`sed 's/.*rootdev=\([^ $]*\).*/\1/' < /proc/cmdline` echo "##" echo "## newdev = $NEWDEV" if test "$NEWDEV" = "cciss" ; then DEVICE1="--ondisk cciss/c0d0" elif test "$NEWDEV" = "ida" ; then DEVICE1="--ondisk ida/c0d0" elif test "$NEWDEV" = "hd" ; then DEVICE1="--ondisk hda" elif test "$NEWDEV" = "sd" ; then DEVICE1="--ondisk sda" else DEVICE1="--ondisk sda" fi echo "## kickstart configured with $DEVICE1" cat > /tmp/parts.ks < Message-ID: <20050316183133.79CE03A58@bigbox.ps-ax.com> You need to add the keyword 'reboot' to your ks.cfg. > I have built a custom CD and ks.cfg file to do an automated install and > setup. It's a fairly stripped-down package set with the Core and Base > groups and a few additional individual rpm's. The only odd thing is > that when it's done and returns to the "Complete, press to > reboot" screen, pressing does nothing. I have to > ++ to reboot. I'm hoping that someone has seen this > behavior before (and figured out what caused it!) > > Below are the isolinux.cfg file and ks.cfg files. > > Thanks in advance! > Gordon Keegan > > > isolinux.cfg: > ------------ > serial 0 9600 > default local > prompt 1 > timeout 600 > display boot.msg > F1 boot.msg > F2 examples.msg > F3 help.msg > > label local > localboot 0x80 > > label install > kernel vmlinuz > append initrd=initrd.img ks=cdrom:/ks.cfg local text > > label install-serial > kernel vmlinuz > append initrd=initrd.img ks=cdrom:/ks.cfg local text console=tty0 > console=ttyS0,9600n8 > > label rescue > kernel vmlinuz > append initrd=initrd.img text rescue > > label noks > kernel vmlinuz > append initrd=initrd.img > > > > ks.cfg: > ------ > install > lang en_US > langsupport en_US > keyboard us > mouse none > rootpw --iscrypted $1encryptedrootpasswordhashstring. > text > cdrom > firewall --disabled > authconfig --enableshadow --enablemd5 > timezone --utc America/New_York > bootloader --linear --location=mbr > > ## > ## Don't want X and don't probe for a monitor > ## > > skipx > #xconfig --monitor=vga > > ## > ## parts.ks is created below in the %pre section > ## > > %include /tmp/parts.ks > > ## > ## Here is our package list > ## '-' at start of the package name indicates exclusion. > ## This ought to remove packages that are part of Base > ## We include kernel-* because RHN cannot provision those > ## due to the up2date PkgSkipList > ## > > %packages > @Base > compat-libstdc++ > XFree86-libs > glibc-headers > libcap > libgcj > libpcap > libtool-libs > lockdev > m4 > ntp > perl-CGI > perl-CPAN > perl-DB_File > perl-suidperl > sendmail-cf > strace > usbutils > zsh > > > > ##---------------------------------------------------------------------- > -----## > > %post --nochroot > > ## re-create /mnt/source to get at CDROM > > if [ -e /tmp/cdrom ]; then > mkdir /mnt/source > mount /tmp/cdrom /mnt/source > fi > > ## for later reference > cat /proc/cmdline > /mnt/sysimage/var/tmp/ks-cmdline > cat /mnt/source/ks.cfg > /mnt/sysimage/var/tmp/ks.cfg > > ## setup post-install scripts/dirs for chroot'ed postinstall stuff > if [ -e /mnt/source/post/postinstall.tar ]; then > mkdir /mnt/sysimage/var/tmp/post > cd /mnt/source/post > tar -cf - . | (cd /mnt/sysimage/var/tmp/post; tar -xvf -) > cd / > fi > > if [ -e /tmp/cdrom ]; then > umount /mnt/source > fi > > > ##---------------------------------------------------------------------- > -----## > > %post > > chvt 3 > > if grep nopost /var/tmp/ks-cmdline; then > > echo "" > echo " * found 'nopost' install option ... postinstall skipped" > echo "" > > else > > > echo "" > echo "Configuring tmpfs" > echo "--------------------" > > if ! grep '/tmp' /etc/fstab ; then > echo " * added a tmpfs entry to /etc/fstab" > echo "tmpfs /tmp tmpfs > size=1024m 1 2" >> /etc/fstab > fi > > echo "" > echo "Postinstall Stuff" > echo "--------------------" > > if [ -s /var/tmp/post/postinstall.tar ]; then > echo " * found postinstall.tar" > echo " * extracting postinstall.tar" > > cd /var/tmp/post > tar -xvf postinstall.tar > > fi > > > ## > ## execute post-install scripts > ## > if [ -f /var/tmp/post/postinstall/postinstall.sh ]; then > echo " * executing postinstall" > echo "" > cd /var/tmp/post/postinstall > bash ./postinstall.sh > else > echo " * ERROR - unable to acquire & execute rh-postinstall" > fi > sleep 2 > > > echo "##############################" > echo "## CONFIGURATION COMPLETE " `date` > echo "##############################" > > chvt 1 > > > ##---------------------------------------------------------------------- > -----## > > %pre > > ## > ## Dynamic Partitioning > ## The output of this script is %include(d) above > ## > > NEWDEV=`sed 's/.*rootdev=\([^ $]*\).*/\1/' < /proc/cmdline` > > echo "##" > echo "## newdev = $NEWDEV" > > if test "$NEWDEV" = "cciss" ; then > DEVICE1="--ondisk cciss/c0d0" > > elif test "$NEWDEV" = "ida" ; then > DEVICE1="--ondisk ida/c0d0" > > elif test "$NEWDEV" = "hd" ; then > DEVICE1="--ondisk hda" > > elif test "$NEWDEV" = "sd" ; then > DEVICE1="--ondisk sda" > > else > DEVICE1="--ondisk sda" > fi > > echo "## kickstart configured with $DEVICE1" > > cat > /tmp/parts.ks < clearpart --all --initlabel > part swap --size 2047 $DEVICE1 --asprimary > part /boot --fstype ext3 --size 128 $DEVICE1 > part / --fstype ext3 --size 2047 $DEVICE1 > part /var --fstype ext3 --size 2047 $DEVICE1 > part swap --size 2047 $DEVICE1 > part /export/home --fstype ext3 --size 1 $DEVICE1 --grow --maxsize > 1024 > EOPARTS > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -- Brad Doctor, CISSP From phr at doc.ic.ac.uk Wed Mar 16 22:59:02 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Wed, 16 Mar 2005 22:59:02 +0000 (GMT) Subject: Kickstart problems... In-Reply-To: References: Message-ID: On Wed, 16 Mar 2005, Robert McKenzie wrote: >I have been spending way too much time trying to get a functional pxe >kickstart server going. I've been through at least 4 or 5 howtos on >this and still come up to the same end each time. The problem is... > >The new system loads it's kernel from the server using PXE with no >problem, but then it insists on trying to get a DHCP address (dynamic). >I have in my dhcpd.conf a static IP for the host, based on MAC but this >seems only to work for the first step of the PXE boot. Perhaps anaconda isn't getting enough info from the command line's "ip=10.102.144.98", thus resorting to DHCP. I'd expect at least the netmask would be needed. From the docs (/usr/share/doc/anaconda-*): ip= IP to use for a network installation, use 'dhcp' for DHCP. netmask= Netmask to use for a network installation. gateway= Gateway to use for a network installation. dns= Comma separated list of nameservers to use for a network installation. Cheers, Phil From crow69dude at hotmail.com Thu Mar 17 02:24:22 2005 From: crow69dude at hotmail.com (Crow69 aka thedude) Date: Wed, 16 Mar 2005 20:24:22 -0600 Subject: Kickstart problems... In-Reply-To: Message-ID: Robert, Here is an example of my /etc/dhcpd.conf file that works with pxeboot: ddns-update-style none; ignore client-updates; subnet 192.168.1.0 netmask 255.255.255.0 { default-lease-time 21600; max-lease-time 43200; group { allow booting; allow bootp; filename "pxelinux.0"; use-host-decl-names on; host n1 { hardware ethernet 00:d0:4f:38:bf:c0; fixed-address 192.168.1.11; } host n2 { hardware ethernet 00:d0:4f:38:bf:d0; fixed-address 192.168.1.12; } } } Here is my /tftpboot/pxelinux.cfg/default file: serial 0 9600 default Redhat prompt 1 timeout 1 label Redhat kernel isolinux/vmlinuz append ksdevice=eth0 load_ramdisk=1 nofb network ks=nfs:192.168.1.5:/path/to/kickstart/ks.cfg initrd=isolinux/initrd.img Maybe you can try my configs and see if they work for you? There are 2 dhcp lookups when doing pxe boot, I believe. The first is the pxe boot itself, and the 2nd is part of the kickstart when booting the cdrom Linux images to start the actual kickstart mechanism. (My dhcpd.conf above is only used for static IPed kickstart dhcp requests specificly specified in the dhcpd.conf file and not general system dhcp service requests.) Make sure you restart the dhcpd service after modifying your dhcpd.conf file! Example: service dhcpd restart Let me know if this helps or not. Jeff >From: "Robert McKenzie" >Reply-To: Discussion list about Kickstart >To: >Subject: Kickstart problems... >Date: Wed, 16 Mar 2005 13:00:03 -0000 > >I have been spending way too much time trying to get a functional pxe >kickstart server going. I've been through at least 4 or 5 howtos on >this and still come up to the same end each time. The problem is... > >The new system loads it's kernel from the server using PXE with no >problem, but then it insists on trying to get a DHCP address (dynamic). >I have in my dhcpd.conf a static IP for the host, based on MAC but this >seems only to work for the first step of the PXE boot. > >What am I doing wrong? (I know, it's a loaded question). > >My entire PXE config is at >http://thehostedbox.com/~rmckenzi/pxe-kickstart/server.tgz > >What I've included in the file (for those that have time to have a look >for me) is /etc/dhcpd.conf and the entire /tftpboot/ directory (minus >the actual initrd.img and vmlinuz images from the FC3 directory. > >I can only think it's something simple that I'm missing or have done >wrong. It would be greatly appreciated for anyone that can have a look >and offer some advice. > >Cheers!!! > > > > > >Robert McKenzie >UNIX Support >CSR >Churchill House >Cowley Road Business Park >Cambridge >CB4 0WZ > >T. 01223 692219 >F. 01223 692001 >Robert.McKenzie at csr.com >www.csr.com > > > >********************************************************************** >This email and any files transmitted with it are confidential and >intended solely for the use of the individual or entity to whom they >are addressed. If you have received this email in error please notify >the system manager. > >********************************************************************** > >_______________________________________________ >Kickstart-list mailing list >Kickstart-list at redhat.com >https://www.redhat.com/mailman/listinfo/kickstart-list From Paul.Miles at quadriga.com Thu Mar 17 09:39:35 2005 From: Paul.Miles at quadriga.com (Miles, Paul) Date: Thu, 17 Mar 2005 09:39:35 -0000 Subject: Software RAID and anaconda Message-ID: <5D339AE687B0C8438185AE62ED06697D0EA320@ukchmail01.mail.quadriga.com> Hello, We're using custom kickstarts to install to HP DL140 servers. Since these servers have 2 IDE hard drives, we're using RAID 1 software raid to configure them. The actual raid configuration works perfectly, the systems install, although very slowly. I tracked the slowness problem to the fact that anaconda appears to be automatically rebuilding the RAID array on each install. I've tried trashing the MBR (ie, dd if=/dev/zero of=/dev/hda bs=1k count=512; dd if=/dev/zero of=/dev/hdb bs=1k count=512) and this seems to perfectly remove the partition table information as shown by fdisk -l but yet still as soon as anaconda makes the partitions during the kickstart, the RAID rebuilding kicks off again. Vital statistics : Fedora core 3 HP Dl140 - xeon processor 1GB ram 2 x IDE hard drives Any thoughts or suggestions would be gratefully received, Paul This e-mail is the property of Quadriga Worldwide Ltd The message (and any associated files) is intended only for the use of the individual or entity to which it is addressed and may contain information that is confidential, subject to copyright or constitutes a trade secret. If you are not the intended recipient you are hereby notified that any dissemination, copying or distribution of this message, or files associated with this message, is strictly prohibited. If you have received this message in error, please notify us immediately by replying to the message and deleting it from your computer. Messages sent to and from us may be monitored. Internet communications cannot be guaranteed to be secure or error-free as information could be intercepted, corrupted, lost, destroyed, arrive late or incomplete, or contain viruses. Therefore, we do not accept responsibility for any errors or omissions that are present in this message, or any attachment, that have arisen as a result of e-mail transmission. If verification is required, please request a hard-copy version. Any views or opinions presented are solely those of the author and do not necessarily represent those of the company. From Robert.McKenzie at csr.com Thu Mar 17 10:26:51 2005 From: Robert.McKenzie at csr.com (Robert McKenzie) Date: Thu, 17 Mar 2005 10:26:51 -0000 Subject: Kickstart problems... Message-ID: Thanks Philip, Adding these corrected the problem of the second DHCP request. I had another message that suggests the problem for the failure of the second DHCP request actually be down to the fact that when then kernel loaded by bootp is loaded it resets the Ethernet ports and causes our switch to go into spanning-tree learning mode causing a long delay (45seconds or so), this appears to be the problem we have here. Even after forcing the options below there was a significant pause (40-60 seconds) while the system "Determines the host and domain names". After this, everything loads fine and the server is building. Thanks for the help. Wish I'd turned to the list before, too much wasted time on this was spent. Cheers!!! -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Philip Rowlands Sent: 16 March 2005 22:59 To: Discussion list about Kickstart Subject: Re: Kickstart problems... On Wed, 16 Mar 2005, Robert McKenzie wrote: >I have been spending way too much time trying to get a functional pxe >kickstart server going. I've been through at least 4 or 5 howtos on >this and still come up to the same end each time. The problem is... > >The new system loads it's kernel from the server using PXE with no >problem, but then it insists on trying to get a DHCP address (dynamic). >I have in my dhcpd.conf a static IP for the host, based on MAC but this >seems only to work for the first step of the PXE boot. Perhaps anaconda isn't getting enough info from the command line's "ip=10.102.144.98", thus resorting to DHCP. I'd expect at least the netmask would be needed. From the docs (/usr/share/doc/anaconda-*): ip= IP to use for a network installation, use 'dhcp' for DHCP. netmask= Netmask to use for a network installation. gateway= Gateway to use for a network installation. dns= Comma separated list of nameservers to use for a network installation. Cheers, Phil _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** From johnpaulmp at amiindia.co.in Thu Mar 17 11:30:37 2005 From: johnpaulmp at amiindia.co.in (johnpaul) Date: Thu, 17 Mar 2005 17:00:37 +0530 Subject: Software RAID and anaconda References: <5D339AE687B0C8438185AE62ED06697D0EA320@ukchmail01.mail.quadriga.com> Message-ID: <016401c52ae4$cc56d540$8300000a@johnpaul> Hi Can you post the kickstart file's contents. That may give some clue John ----- Original Message ----- From: "Miles, Paul" To: Sent: Thursday, March 17, 2005 3:09 PM Subject: Software RAID and anaconda > Hello, > > We're using custom kickstarts to install to HP DL140 servers. Since > these servers have 2 IDE hard drives, we're using RAID 1 software raid > to configure them. > > The actual raid configuration works perfectly, the systems install, > although very slowly. > > I tracked the slowness problem to the fact that anaconda appears to be > automatically rebuilding the RAID array on each install. I've tried > trashing the MBR (ie, dd if=/dev/zero of=/dev/hda bs=1k count=512; dd > if=/dev/zero of=/dev/hdb bs=1k count=512) and this seems to perfectly > remove the partition table information as shown by fdisk -l but yet > still as soon as anaconda makes the partitions during the kickstart, the > RAID rebuilding kicks off again. > > Vital statistics : > > Fedora core 3 > HP Dl140 - xeon processor > 1GB ram > 2 x IDE hard drives > > Any thoughts or suggestions would be gratefully received, > > Paul > This e-mail is the property of Quadriga Worldwide Ltd > The message (and any associated files) is intended only for the use of > the individual or entity to which it is addressed and may contain information > that is confidential, subject to copyright or constitutes a trade secret. If > you are not the intended recipient you are hereby notified that any > dissemination, copying or distribution of this message, or files associated > with this message, is strictly prohibited. If you have received this > message in error, please notify us immediately by replying to the message > and deleting it from your computer. Messages sent to and from us may be > monitored. Internet communications cannot be guaranteed to be secure or > error-free as information could be intercepted, corrupted, lost, destroyed, > arrive late or incomplete, or contain viruses. Therefore, we do not accept > responsibility for any errors or omissions that are present in this > message, or any attachment, that have arisen as a result of e-mail > transmission. If verification is required, please request a hard-copy > version. Any views or opinions presented are solely those of the author and > do not necessarily represent those of the company. > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > > -- > This message was scanned for spam and viruses by BitDefender. > For more information please visit http://linux.bitdefender.com/ > -- This message was scanned for spam and viruses by BitDefender. For more information please visit http://linux.bitdefender.com/ From jim at rossberry.com Thu Mar 17 11:56:33 2005 From: jim at rossberry.com (Jim Wildman) Date: Thu, 17 Mar 2005 06:56:33 -0500 (EST) Subject: Software RAID and anaconda In-Reply-To: <5D339AE687B0C8438185AE62ED06697D0EA320@ukchmail01.mail.quadriga.com> References: <5D339AE687B0C8438185AE62ED06697D0EA320@ukchmail01.mail.quadriga.com> Message-ID: On Thu, 17 Mar 2005, Miles, Paul wrote: > I tracked the slowness problem to the fact that anaconda appears to be > automatically rebuilding the RAID array on each install. I've tried > trashing the MBR (ie, dd if=/dev/zero of=/dev/hda bs=1k count=512; dd > if=/dev/zero of=/dev/hdb bs=1k count=512) and this seems to perfectly > remove the partition table information as shown by fdisk -l but yet > still as soon as anaconda makes the partitions during the kickstart, the > RAID rebuilding kicks off again. This is the nature of RAID. When you create the RAID, well, it gets created which means the images have to be synced, and with RAID 1, everything happens on both drives. With IDE drives, probably on the same channel, it's gonna be slow. Have you considered creating the RAID after install? ------------------------------------------------------------------------ Jim Wildman, CISSP, RHCE jim at rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine From Gordon.Keegan at FMR.COM Thu Mar 17 15:09:15 2005 From: Gordon.Keegan at FMR.COM (Keegan, Gordon) Date: Thu, 17 Mar 2005 10:09:15 -0500 Subject: small custom CD issue/question Message-ID: <386604AA76CB7F4CB15FC4B4839EDC3F1364C1@MSGMMKDRC2WIN.DMN1.FMR.COM> Except that I don't necessarily want it to automatically reboot when it's done, I _want_ it to wait until I press . It's just that pressing doesn't do anything. Any other suggestions? Gordon -----Original Message----- From: Brad Doctor [mailto:bdoctor at ps-ax.com] Sent: Wednesday, March 16, 2005 1:32 PM To: Discussion list about Kickstart Subject: Re: small custom CD issue/question You need to add the keyword 'reboot' to your ks.cfg. > I have built a custom CD and ks.cfg file to do an automated install and > setup. It's a fairly stripped-down package set with the Core and Base > groups and a few additional individual rpm's. The only odd thing is > that when it's done and returns to the "Complete, press to > reboot" screen, pressing does nothing. I have to > ++ to reboot. I'm hoping that someone has seen this > behavior before (and figured out what caused it!) > > Below are the isolinux.cfg file and ks.cfg files. > > Thanks in advance! > Gordon Keegan > > > isolinux.cfg: > ------------ > serial 0 9600 > default local > prompt 1 > timeout 600 > display boot.msg > F1 boot.msg > F2 examples.msg > F3 help.msg > > label local > localboot 0x80 > > label install > kernel vmlinuz > append initrd=initrd.img ks=cdrom:/ks.cfg local text > > label install-serial > kernel vmlinuz > append initrd=initrd.img ks=cdrom:/ks.cfg local text console=tty0 > console=ttyS0,9600n8 > > label rescue > kernel vmlinuz > append initrd=initrd.img text rescue > > label noks > kernel vmlinuz > append initrd=initrd.img > > > > ks.cfg: > ------ > install > lang en_US > langsupport en_US > keyboard us > mouse none > rootpw --iscrypted $1encryptedrootpasswordhashstring. > text > cdrom > firewall --disabled > authconfig --enableshadow --enablemd5 > timezone --utc America/New_York > bootloader --linear --location=mbr > > ## > ## Don't want X and don't probe for a monitor > ## > > skipx > #xconfig --monitor=vga > > ## > ## parts.ks is created below in the %pre section > ## > > %include /tmp/parts.ks > > ## > ## Here is our package list > ## '-' at start of the package name indicates exclusion. > ## This ought to remove packages that are part of Base > ## We include kernel-* because RHN cannot provision those > ## due to the up2date PkgSkipList > ## > > %packages > @Base > compat-libstdc++ > XFree86-libs > glibc-headers > libcap > libgcj > libpcap > libtool-libs > lockdev > m4 > ntp > perl-CGI > perl-CPAN > perl-DB_File > perl-suidperl > sendmail-cf > strace > usbutils > zsh > > > > ##---------------------------------------------------------------------- > -----## > > %post --nochroot > > ## re-create /mnt/source to get at CDROM > > if [ -e /tmp/cdrom ]; then > mkdir /mnt/source > mount /tmp/cdrom /mnt/source > fi > > ## for later reference > cat /proc/cmdline > /mnt/sysimage/var/tmp/ks-cmdline > cat /mnt/source/ks.cfg > /mnt/sysimage/var/tmp/ks.cfg > > ## setup post-install scripts/dirs for chroot'ed postinstall stuff > if [ -e /mnt/source/post/postinstall.tar ]; then > mkdir /mnt/sysimage/var/tmp/post > cd /mnt/source/post > tar -cf - . | (cd /mnt/sysimage/var/tmp/post; tar -xvf -) > cd / > fi > > if [ -e /tmp/cdrom ]; then > umount /mnt/source > fi > > > ##---------------------------------------------------------------------- > -----## > > %post > > chvt 3 > > if grep nopost /var/tmp/ks-cmdline; then > > echo "" > echo " * found 'nopost' install option ... postinstall skipped" > echo "" > > else > > > echo "" > echo "Configuring tmpfs" > echo "--------------------" > > if ! grep '/tmp' /etc/fstab ; then > echo " * added a tmpfs entry to /etc/fstab" > echo "tmpfs /tmp tmpfs > size=1024m 1 2" >> /etc/fstab > fi > > echo "" > echo "Postinstall Stuff" > echo "--------------------" > > if [ -s /var/tmp/post/postinstall.tar ]; then > echo " * found postinstall.tar" > echo " * extracting postinstall.tar" > > cd /var/tmp/post > tar -xvf postinstall.tar > > fi > > > ## > ## execute post-install scripts > ## > if [ -f /var/tmp/post/postinstall/postinstall.sh ]; then > echo " * executing postinstall" > echo "" > cd /var/tmp/post/postinstall > bash ./postinstall.sh > else > echo " * ERROR - unable to acquire & execute rh-postinstall" > fi > sleep 2 > > > echo "##############################" > echo "## CONFIGURATION COMPLETE " `date` > echo "##############################" > > chvt 1 > > > ##---------------------------------------------------------------------- > -----## > > %pre > > ## > ## Dynamic Partitioning > ## The output of this script is %include(d) above > ## > > NEWDEV=`sed 's/.*rootdev=\([^ $]*\).*/\1/' < /proc/cmdline` > > echo "##" > echo "## newdev = $NEWDEV" > > if test "$NEWDEV" = "cciss" ; then > DEVICE1="--ondisk cciss/c0d0" > > elif test "$NEWDEV" = "ida" ; then > DEVICE1="--ondisk ida/c0d0" > > elif test "$NEWDEV" = "hd" ; then > DEVICE1="--ondisk hda" > > elif test "$NEWDEV" = "sd" ; then > DEVICE1="--ondisk sda" > > else > DEVICE1="--ondisk sda" > fi > > echo "## kickstart configured with $DEVICE1" > > cat > /tmp/parts.ks < clearpart --all --initlabel > part swap --size 2047 $DEVICE1 --asprimary > part /boot --fstype ext3 --size 128 $DEVICE1 > part / --fstype ext3 --size 2047 $DEVICE1 > part /var --fstype ext3 --size 2047 $DEVICE1 > part swap --size 2047 $DEVICE1 > part /export/home --fstype ext3 --size 1 $DEVICE1 --grow --maxsize > 1024 > EOPARTS > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -- Brad Doctor, CISSP From Paul.Miles at quadriga.com Fri Mar 18 10:38:44 2005 From: Paul.Miles at quadriga.com (Miles, Paul) Date: Fri, 18 Mar 2005 10:38:44 -0000 Subject: Software RAID and anaconda Message-ID: <5D339AE687B0C8438185AE62ED06697D0EA321@ukchmail01.mail.quadriga.com> Hello, Just to update the list, as Jim suggested, I couldn't find away to stop the RAID array from initialising during kickstart. However, after a little hunting with google I found a way to throttle back the amount of disk IO to use for the rebuild. This is in : /proc/sys/dev/raid/speed_limit_max /proc/sys/dev/raid/speed_limit_min I reduced both of these down as part of my %pre section and now the kickstart runs at a reasonable speed. Thanks for all your help, and appologies for the stupidly long email disclamer, I have no control over it :-( Paul This e-mail is the property of Quadriga Worldwide Ltd The message (and any associated files) is intended only for the use of the individual or entity to which it is addressed and may contain information that is confidential, subject to copyright or constitutes a trade secret. If you are not the intended recipient you are hereby notified that any dissemination, copying or distribution of this message, or files associated with this message, is strictly prohibited. If you have received this message in error, please notify us immediately by replying to the message and deleting it from your computer. Messages sent to and from us may be monitored. Internet communications cannot be guaranteed to be secure or error-free as information could be intercepted, corrupted, lost, destroyed, arrive late or incomplete, or contain viruses. Therefore, we do not accept responsibility for any errors or omissions that are present in this message, or any attachment, that have arisen as a result of e-mail transmission. If verification is required, please request a hard-copy version. Any views or opinions presented are solely those of the author and do not necessarily represent those of the company. From john.j.poole at usa-spaceops.com Fri Mar 18 16:17:18 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Fri, 18 Mar 2005 11:17:18 -0500 Subject: e1000 still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB990@usaflcms03.usa-spaceops.ksc.nasa.gov> Ed, Thanks. I have read about the "spanning tree" issues. We have little input to the networking group and are a few Linux Boxes in a Windows world. I guess I was looking for a short "how-to" on building a new boot image to see if I have overlooked something. So close but no cigar. I am unsure if I have all the changes incorporated in my new initrd.img or if I must build a new vmlinuz also. Maybe by the time I figure it out a working release will be available. Thanks for taking the time to respond. Take care, John -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com]On Behalf Of kickstart-list-request at redhat.com Sent: Wednesday, March 16, 2005 12:01 PM To: kickstart-list at redhat.com Subject: Kickstart-list Digest, Vol 13, Issue 10 Send Kickstart-list mailing list submissions to kickstart-list at redhat.com To subscribe or unsubscribe via the World Wide Web, visit https://www.redhat.com/mailman/listinfo/kickstart-list or, via email, send a message with subject or body 'help' to kickstart-list-request at redhat.com You can reach the person managing the list at kickstart-list-owner at redhat.com When replying, please edit your Subject line so it is more specific than "Re: Contents of Kickstart-list digest..." From peter at boku.net Fri Mar 18 16:45:11 2005 From: peter at boku.net (Peter Eisch) Date: Fri, 18 Mar 2005 10:45:11 -0600 Subject: e1000 still In-Reply-To: <2CE66A1ABBED8C4B85DFF5E0860987A32CB990@usaflcms03.usa-spaceops.ksc.nasa.gov> Message-ID: Which RH version you working with? On 3/18/05 10:17 AM, "Poole, John J" wrote: > Ed, > Thanks. I have read about the "spanning tree" issues. We have little input to > the networking group and are a few Linux Boxes in a Windows world. I guess > I was looking for a short "how-to" on building a new boot image to see if I > have overlooked something. So close but no cigar. I am unsure if I have all > the > changes incorporated in my new initrd.img or if I must build a new vmlinuz > also. Maybe by the time I figure it out a working release will be available. > Thanks for taking the time to respond. > Take care, > John > From corey at mojohost.com Fri Mar 18 17:00:37 2005 From: corey at mojohost.com (Corey Baldwin) Date: Fri, 18 Mar 2005 12:00:37 -0500 Subject: automated response Message-ID: <10503181200.AA15604031@mojohost.com> I will be away from the office from Friday, March 18 through Sunday, March 20th. If this is a support question, please visit http://www.mojohost.com/support and enter a trouble ticket. In case of emergency, please contact Brad Mitchell at 248-842-6900. Thank you, Corey Baldwin From john.j.poole at usa-spaceops.com Fri Mar 18 17:26:05 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Fri, 18 Mar 2005 12:26:05 -0500 Subject: e1000 still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> Peter, Started with RHEL 3WS update 1 and are currently up to RHEL 3WS update 4. I got the source for Anaconda-9.1.4.1-1 and compiles "loader" and put into the /isolinux/initrd.img from the update 4 cd's. Looks like and NFS install works but when I put url --url=http://myserver/source it fails downloading the image initially and selecting "OK" at the prompts without changing any parameters, it takes off and loads. John -----Original Message----- From: Peter Eisch [mailto:peter at boku.net] Sent: Friday, March 18, 2005 11:45 AM To: Discussion list about Kickstart Subject: Re: e1000 still Which RH version you working with? On 3/18/05 10:17 AM, "Poole, John J" wrote: > Ed, > Thanks. I have read about the "spanning tree" issues. We have little input to > the networking group and are a few Linux Boxes in a Windows world. I guess > I was looking for a short "how-to" on building a new boot image to see if I > have overlooked something. So close but no cigar. I am unsure if I have all > the > changes incorporated in my new initrd.img or if I must build a new vmlinuz > also. Maybe by the time I figure it out a working release will be available. > Thanks for taking the time to respond. > Take care, > John > From s.davison at computer.org Sat Mar 19 15:07:41 2005 From: s.davison at computer.org (S. W. Davison) Date: Sat, 19 Mar 2005 10:07:41 -0500 Subject: e1000 still In-Reply-To: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> Message-ID: <423C403D.90400@computer.org> John and Others - I'm having the same problem. It shows up as a dialog box saying "That directory could not be mounted from the server". Then if you tell it to try again, without changing any parameters, it eventually manages to mount the NFS share and proceed with the installation. I have not gotten as far as John has toward a solution. I cannot get the NFS install to work without interacting with the install dialogs (sometimes several times) to tell it to try the NFS mount again. Can you tell me how you modified loader.c and installed it into initrd.img to get NFS installs working? Or maybe you can point me to the right documentation source? We are using Red Hat Enterprise Linux 3, Update 4. According to rpm --query, it has anaconda-9.1.4.1. That version of anaconda (according to http://rhn.redhat.com/errata/RHEA-2004-518.html) is supposed to have a fix for the problem. I assume that the initrd.img and vmlinuz from the isolinux directory of CD 1 of the rhel3u4 install set were built using the up-to-date anaconda, pump, etc. packages that seem to be included in the rhel3u4 distro, but I don't know how to verify that. We filed a support request with Red Hat. Their response was basically that anaconda and the rest of the installer were working as designed, and that we should alter the parameters on our HP Procurve switches (turning "spanning tree" off and turning "portfast" on were suggested). Since the NFS mount always succeeds eventually, it seems to me that the problem should be fixable by adjusting timing parameters or retry limits in the code. (Of course, I haven't looked at that code yet.) But I can't figure out how to get the rebuilt installer binaries into initrd.img, if that's where they have to to. So I would be grateful for any suggestions or documentation you might have. Stowe Davison Poole, John J wrote: > Peter, > Started with RHEL 3WS update 1 and are currently up to RHEL 3WS update 4. I got the source for Anaconda-9.1.4.1-1 and compiles "loader" and put into the > /isolinux/initrd.img from the update 4 cd's. Looks like and NFS install works but when I put url --url=http://myserver/source it fails downloading the image > initially and selecting "OK" at the prompts without changing any parameters, it takes off and loads. > John > > > -----Original Message----- > From: Peter Eisch [mailto:peter at boku.net] > Sent: Friday, March 18, 2005 11:45 AM > To: Discussion list about Kickstart > Subject: Re: e1000 still > > > > Which RH version you working with? > > On 3/18/05 10:17 AM, "Poole, John J" wrote: > > >>Ed, >> Thanks. I have read about the "spanning tree" issues. We have little input to >>the networking group and are a few Linux Boxes in a Windows world. I guess >>I was looking for a short "how-to" on building a new boot image to see if I >>have overlooked something. So close but no cigar. I am unsure if I have all >>the >>changes incorporated in my new initrd.img or if I must build a new vmlinuz >>also. Maybe by the time I figure it out a working release will be available. >>Thanks for taking the time to respond. >>Take care, >>John >> > > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > From peter at boku.net Sat Mar 19 17:38:50 2005 From: peter at boku.net (Peter Eisch) Date: Sat, 19 Mar 2005 11:38:50 -0600 Subject: e1000 still In-Reply-To: <423C403D.90400@computer.org> Message-ID: This sounds right. The spanningtree config will listen on the port for 30 seconds before forwarding packets from the port. If it were a driver issue or such, it wouldn't ever work. By disabiling the spanningtree (or setting portfast or whatever the option is on your LAN fabric) or adding a 30 second pause before trying the network should get you going on the first try. peter On 3/19/05 9:07 AM, "S. W. Davison" wrote: > We filed a support request with Red Hat. Their > response was basically that anaconda and the rest of > the installer were working as designed, and that we > should alter the parameters on our HP Procurve > switches (turning "spanning tree" off and turning > "portfast" on were suggested). From jeroen at science.uva.nl Sun Mar 20 23:06:38 2005 From: jeroen at science.uva.nl (Jeroen Roodhart) Date: Mon, 21 Mar 2005 00:06:38 +0100 Subject: e1000 still -- Possible solution Message-ID: <423E01FE.8070107@science.uva.nl> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I just browsed through my kickstart-list summary and saw that you seem to have the same problems with anaconda that we had back with FC2 and e1000 NICs. The first NFS-mount doesn't work with anaconda, but if you then do the mount by hand it works just fine, right? OK, someone back then (forgot your name, sorry ;)) suggested a simple patch to isys/nfsmount.c; Just try to remount the thing for a maximum number of times. This was the patch that I used to make things work: *** nfsmount.c 2004-05-06 23:40:21.000000000 +0200 - --- nfsmount.c 2004-06-10 12:55:56.600409365 +0200 *************** *** 198,209 **** ~ { ~ struct pmaplist *pmap; ~ static struct pmap p = {0, 0, 0, 0}; ~ server_addr->sin_port = PMAPPORT; ~ pmap = pmap_getmaps(server_addr); ! if (!pmap) ! return NULL; ~ if (version > MAX_NFSPROT) ~ version = MAX_NFSPROT; - --- 198,218 ---- ~ { ~ struct pmaplist *pmap; ~ static struct pmap p = {0, 0, 0, 0}; + int retries = 0; ~ server_addr->sin_port = PMAPPORT; ~ pmap = pmap_getmaps(server_addr); ! if (!pmap) { ! while (!pmap && (retries < 30)) { ! sleep(2); ! pmap = pmap_getmaps(server_addr); ! retries++; ! } ! if (!pmap) ! return NULL; ! } ! ~ if (version > MAX_NFSPROT) ~ version = MAX_NFSPROT; Simple enough and it worked for me, mayhaps it will work for you. Good luck!, Jeroen - -- Jeroen Roodhart University of Amsterdam jeroen at science.uva.nl Faculty of Science / ICT-Group Systeem- en netwerkbeheer Tel. 020 525 7203 / 06 51338165 - -- See http://www.science.uva.nl/~jeroen for openPGP public key -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCPgH+37AP1zFtDU0RApHQAJ9LiH22/s8fTlk5WNTdofsnBqrxGACgjtr2 5w0HCASCZxYnjuNp9BL83WU= =hMH7 -----END PGP SIGNATURE----- From john.j.poole at usa-spaceops.com Mon Mar 21 16:52:34 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Mon, 21 Mar 2005 11:52:34 -0500 Subject: e1000 still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB997@usaflcms03.usa-spaceops.ksc.nasa.gov> To all, https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=110036 I was able to get NFS kickstart to work by using the two patches at the bottom of the bug report above. At the time I was doing an install using DHCP and NFS. I modified dhcp.c and imount per instructions in this bug report. It seemed to make NFS wait the length of time required to work doing NFS installs. It did not work for HTTP installs however. I do not know what part of these patches were eventually applied in the latest anaconda. I do have a problem getting switch or router changes applied. Maybe it would be nice if there were a flag (-wait_forever) for those who are forced to live with 30 second delays :) Thanks, John -----Original Message----- From: S. W. Davison [mailto:s.davison at computer.org] Sent: Saturday, March 19, 2005 10:08 AM To: Discussion list about Kickstart Subject: Re: e1000 still John and Others - I'm having the same problem. It shows up as a dialog box saying "That directory could not be mounted from the server". Then if you tell it to try again, without changing any parameters, it eventually manages to mount the NFS share and proceed with the installation. I have not gotten as far as John has toward a solution. I cannot get the NFS install to work without interacting with the install dialogs (sometimes several times) to tell it to try the NFS mount again. Can you tell me how you modified loader.c and installed it into initrd.img to get NFS installs working? Or maybe you can point me to the right documentation source? We are using Red Hat Enterprise Linux 3, Update 4. According to rpm --query, it has anaconda-9.1.4.1. That version of anaconda (according to http://rhn.redhat.com/errata/RHEA-2004-518.html) is supposed to have a fix for the problem. I assume that the initrd.img and vmlinuz from the isolinux directory of CD 1 of the rhel3u4 install set were built using the up-to-date anaconda, pump, etc. packages that seem to be included in the rhel3u4 distro, but I don't know how to verify that. We filed a support request with Red Hat. Their response was basically that anaconda and the rest of the installer were working as designed, and that we should alter the parameters on our HP Procurve switches (turning "spanning tree" off and turning "portfast" on were suggested). Since the NFS mount always succeeds eventually, it seems to me that the problem should be fixable by adjusting timing parameters or retry limits in the code. (Of course, I haven't looked at that code yet.) But I can't figure out how to get the rebuilt installer binaries into initrd.img, if that's where they have to to. So I would be grateful for any suggestions or documentation you might have. Stowe Davison Poole, John J wrote: > Peter, > Started with RHEL 3WS update 1 and are currently up to RHEL 3WS update 4. I got the source for Anaconda-9.1.4.1-1 and compiles "loader" and put into the > /isolinux/initrd.img from the update 4 cd's. Looks like and NFS install works but when I put url --url=http://myserver/source it fails downloading the image > initially and selecting "OK" at the prompts without changing any parameters, it takes off and loads. > John > > > -----Original Message----- > From: Peter Eisch [mailto:peter at boku.net] > Sent: Friday, March 18, 2005 11:45 AM > To: Discussion list about Kickstart > Subject: Re: e1000 still > > > > Which RH version you working with? > > On 3/18/05 10:17 AM, "Poole, John J" wrote: > > >>Ed, >> Thanks. I have read about the "spanning tree" issues. We have little input to >>the networking group and are a few Linux Boxes in a Windows world. I guess >>I was looking for a short "how-to" on building a new boot image to see if I >>have overlooked something. So close but no cigar. I am unsure if I have all >>the >>changes incorporated in my new initrd.img or if I must build a new vmlinuz >>also. Maybe by the time I figure it out a working release will be available. >>Thanks for taking the time to respond. >>Take care, >>John >> > > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > From ebrown at lanl.gov Mon Mar 21 17:59:15 2005 From: ebrown at lanl.gov (Ed Brown) Date: Mon, 21 Mar 2005 10:59:15 -0700 Subject: e1000 still In-Reply-To: <423C403D.90400@computer.org> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> Message-ID: <1111427955.29152.140.camel@edbrown.lanl.gov> On Sat, 2005-03-19 at 08:07, S. W. Davison wrote: > We filed a support request with Red Hat. Their > response was basically that anaconda and the rest of > the installer were working as designed, and that we > should alter the parameters on our HP Procurve > switches (turning "spanning tree" off and turning > "portfast" on were suggested). It's hard to understand this position by RedHat. Anaconda is clearly NOT working as designed. This has been going on for more than a year, with problems reported for ftp, http, and NFS installs, using HP, 3com, and Cisco switches. Altering switch configurations just to be able to install is not always possible, much less a desirable solution. The issue does seem to involve some subtle timing problem in anaconda. Efforts to capture traffic to pinpoint the breakdown fail because just inserting a supposedly passive analyzer between the box being built and the switch make the problem unreproducible. As an aside, this problem has other consequences in an organization. Promoting wider use of RedHat among co-workers is a harder sell when something like this goes on unresolved, and apparently unacknowledged, for so long. Ed Brown, RHCE Los Alamos National Laboratory From phr at doc.ic.ac.uk Mon Mar 21 22:01:11 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Mon, 21 Mar 2005 22:01:11 +0000 (GMT) Subject: e1000 still In-Reply-To: <1111427955.29152.140.camel@edbrown.lanl.gov> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> <1111427955.29152.140.camel@edbrown.lanl.gov> Message-ID: On Mon, 21 Mar 2005, Ed Brown wrote: >It's hard to understand this position by RedHat. Anaconda is clearly >NOT working as designed. It's the error messages (or lack thereof) which annoy me. Hands up [virtually] whoever has been dropped to the "Language Select" screen after a misconfigured kickstart attempt? It's almost enough to want to patch it... Do spanning-tree links appear "up" to the host, before the switch relents and starts to forward? If not, perhaps it's possible to programmatically delay in a %pre script. Cheers, Phil From James_Martin at ao.uscourts.gov Mon Mar 21 22:29:04 2005 From: James_Martin at ao.uscourts.gov (James_Martin at ao.uscourts.gov) Date: Mon, 21 Mar 2005 17:29:04 -0500 Subject: e1000 still In-Reply-To: Message-ID: This is an age old problem going back to my RH 7.2 days. It does have to do with span & tree and port fast and all of that stuff. I had to summon my 1337 C programming skills to solve it. Get a copy of the anaconda source rpm, install it, and take a look at the init.c in the loader2/ directory. What I did was hard coded a 30 second sleep in there, and never had a problem since. Of course after you mod it and compile it, you'll have to put it inito a new initrd.img. #> diff init.c.old init.c 615a616,618 > > printf("sleeping 30 seconds.. thanks JSM!\n"); > sleep(30); in context it looks like this: if (!testing) doklog("/dev/tty4"); /* Go into normal init mode - keep going, and then do a orderly shutdown when: 1) /bin/install exits 2) we receive a SIGHUP */ printf("sleeping 30 seconds.. thanks JSM!\n"); sleep(30); printf("running install...\n"); setsid(); This basically forces it to sleep for whatever time you specify until loader is launched. By that time, the network should be up and ready. I know it's hacky, but it's worked for the past 4 years. James James S. Martin, RHCE Contractor Administrative Office of the United States Courts Washington, DC (202) 502-2394 kickstart-list-bounces at redhat.com wrote on 03/21/2005 05:01:11 PM: > On Mon, 21 Mar 2005, Ed Brown wrote: > > >It's hard to understand this position by RedHat. Anaconda is clearly > >NOT working as designed. > > It's the error messages (or lack thereof) which annoy me. Hands up > [virtually] whoever has been dropped to the "Language Select" screen > after a misconfigured kickstart attempt? > > It's almost enough to want to patch it... > > Do spanning-tree links appear "up" to the host, before the switch > relents and starts to forward? If not, perhaps it's possible to > programmatically delay in a %pre script. > > > > Cheers, > Phil > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From ebrown at lanl.gov Tue Mar 22 00:42:13 2005 From: ebrown at lanl.gov (Ed Brown) Date: Mon, 21 Mar 2005 17:42:13 -0700 Subject: e1000 still In-Reply-To: References: Message-ID: <1111452133.29152.553.camel@edbrown.lanl.gov> On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > This is an age old problem going back to my RH 7.2 days. Yes, this is old news, hashed over on several lists, in many threads. James solution may help the original post-er to this thread, who was doing remote installs, with no access to the switches. The point, or mine anyway, is that hacking the code or having to modify switch configurations to be able to install the OS, are poor alternatives to be faced with. How is this not a problem with anaconda, and why hasn't RedHat owned it / fixed it? -Ed From keanboon at gawab.com Tue Mar 22 09:03:59 2005 From: keanboon at gawab.com (keanboon) Date: Tue, 22 Mar 2005 09:03:59 GMT Subject: execute shell script in postinstall Message-ID: <20050322090359.13173.qmail@gawab.com> An HTML attachment was scrubbed... URL: From phr at doc.ic.ac.uk Tue Mar 22 09:57:11 2005 From: phr at doc.ic.ac.uk (Philip Rowlands) Date: Tue, 22 Mar 2005 09:57:11 +0000 (GMT) Subject: execute shell script in postinstall In-Reply-To: <20050322090359.13173.qmail@gawab.com> References: <20050322090359.13173.qmail@gawab.com> Message-ID: On Tue, 22 Mar 2005, keanboon wrote: >I?want to execute a shell script in postinstall. Where should i put my >shell script? How can i execute it? Hard disk, optical disk, network... e.g. : %post wget http://server.example.com/script.sh sh script.sh >Besides that, i also get an error message while?running the kickstart >file: >Error opening: kickstart file >/tmp/ks.cfg? No such file or directory Could be one of several problems; what was the kernel command line? Cheers, Phil From brilong at cisco.com Tue Mar 22 13:38:50 2005 From: brilong at cisco.com (Brian Long) Date: Tue, 22 Mar 2005 08:38:50 -0500 Subject: e1000 still In-Reply-To: References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> <1111427955.29152.140.camel@edbrown.lanl.gov> Message-ID: <1111498730.4162.13.camel@brilong-lnx.cisco.com> On Mon, 2005-03-21 at 22:01 +0000, Philip Rowlands wrote: > Do spanning-tree links appear "up" to the host, before the switch > relents and starts to forward? If not, perhaps it's possible to > programmatically delay in a %pre script. Yes, the link is up, but the switch will not forward packets until the spanning tree calculation is complete. /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s From bda20 at cam.ac.uk Tue Mar 22 13:40:44 2005 From: bda20 at cam.ac.uk (Ben) Date: Tue, 22 Mar 2005 13:40:44 +0000 (GMT) Subject: e1000 still In-Reply-To: <1111498730.4162.13.camel@brilong-lnx.cisco.com> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> <1111427955.29152.140.camel@edbrown.lanl.gov> <1111498730.4162.13.camel@brilong-lnx.cisco.com> Message-ID: On Tue, 22 Mar 2005, Brian Long wrote: > On Mon, 2005-03-21 at 22:01 +0000, Philip Rowlands wrote: > >> Do spanning-tree links appear "up" to the host, before the switch >> relents and starts to forward? If not, perhaps it's possible to >> programmatically delay in a %pre script. > > Yes, the link is up, but the switch will not forward packets until the > spanning tree calculation is complete. You could try turning on 'portfast' on the port. This should allow packet flow during the calculation period. Ben -- Unix Support, MISD, University of Cambridge, England Plugger of wire, typer of keyboard, imparter of Clue Life Is Short. It's All Good. From andrew.w.robinson at mms.gov Tue Mar 22 14:41:15 2005 From: andrew.w.robinson at mms.gov (Robinson, Andrew W.) Date: Tue, 22 Mar 2005 07:41:15 -0700 Subject: execute shell script in postinstall Message-ID: <379313C94B76D2119AB60008C7A402E409E46384@IMSNOLAA> > From: Philip Rowlands [mailto:phr at doc.ic.ac.uk] > > On Tue, 22 Mar 2005, keanboon wrote: > > >I?want to execute a shell script in postinstall. Where > should i put my > >shell script? How can i execute it? > > Hard disk, optical disk, network... e.g. : > > %post > wget http://server.example.com/script.sh > sh script.sh Just to add a couple of points from my own experience (yes, the hard way :). - If you execute the script directly, './script' instead of 'sh script.sh', be sure to make it executable. Phil's method would have saved me some grief. - Until you reboot, you have the limited installation environment and tool set. For example, sed is available, but awk is not. Andrew Robinson From James_Martin at ao.uscourts.gov Tue Mar 22 14:51:57 2005 From: James_Martin at ao.uscourts.gov (James_Martin at ao.uscourts.gov) Date: Tue, 22 Mar 2005 09:51:57 -0500 Subject: e1000 still In-Reply-To: <1111452133.29152.553.camel@edbrown.lanl.gov> Message-ID: I agree. The less hacking the better. Red Hat could easily put a sleep variable in the init.c that is set at the syslinux command line. Have you type something like: linux network_wait=30 That parameter would then be passed to init which would then wait the specified amount of time. Changing your network configuration to accommodate a operating system install is ridiculous. Jeremy/Paul, opinions? James James S. Martin, RHCE Contractor Administrative Office of the United States Courts Washington, DC (202) 502-2394 kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: > On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > > This is an age old problem going back to my RH 7.2 days. > > Yes, this is old news, hashed over on several lists, in many threads. > James solution may help the original post-er to this thread, who was > doing remote installs, with no access to the switches. > > The point, or mine anyway, is that hacking the code or having to modify > switch configurations to be able to install the OS, are poor > alternatives to be faced with. How is this not a problem with anaconda, > and why hasn't RedHat owned it / fixed it? > > -Ed > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From jhegge at permeo.com Tue Mar 22 15:34:58 2005 From: jhegge at permeo.com (Hegge, Johnathan) Date: Tue, 22 Mar 2005 09:34:58 -0600 Subject: e1000 still Message-ID: > -----Original Message----- > From: James_Martin at ao.uscourts.gov > [mailto:James_Martin at ao.uscourts.gov] > Sent: Tuesday, March 22, 2005 8:52 AM > To: Discussion list about Kickstart > Subject: Re: e1000 still > > I agree. The less hacking the better. Red Hat could easily > put a sleep > variable in the init.c that is set at the syslinux command > line. Have you > type something like: > > linux network_wait=30 I think spanning tree can take up to 45 seconds (Cisco gurus can correct) and I've often seen it take some time. It is a good idea. > > That parameter would then be passed to init which would then wait the > specified amount of time. Changing your network configuration to > accommodate a operating system install is ridiculous. I completely agree. Turning off spanning tree could cause further problems, even if not related directly to the RH server. Add a few aging hubs in the topology and turn off spanning tree, then just start counting until the network gets knocked down. Instead of turning off spanning tree, I'd turn on portfast and add storm control as well -- at least to protect the servers from ill-behaved clients. > > Jeremy/Paul, opinions? > > > James > > > James S. Martin, RHCE > Contractor > Administrative Office of the United States Courts > Washington, DC > (202) 502-2394 > > kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: > > > On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > > > This is an age old problem going back to my RH 7.2 days. > > > > Yes, this is old news, hashed over on several lists, in > many threads. > > James solution may help the original post-er to this thread, who was > > doing remote installs, with no access to the switches. > > > > The point, or mine anyway, is that hacking the code or > having to modify > > switch configurations to be able to install the OS, are poor > > alternatives to be faced with. How is this not a problem > with anaconda, > > and why hasn't RedHat owned it / fixed it? > > > > -Ed > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > > From john.j.poole at usa-spaceops.com Tue Mar 22 15:45:56 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Tue, 22 Mar 2005 10:45:56 -0500 Subject: e1000 still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB99F@usaflcms03.usa-spaceops.ksc.nasa.gov> James, I will try the 30 second sleep as soon as I have the time. I think it may a workaround for the port tree spanning issue. I will post results when I test it. I agree with everyone about this issue being ignored. I have spent many hours trying to make installs work. Sure adds credence to those who say Linux is too labor intensive to be practical for business applications. I have read that this problem also shows up with other Ethernet cards but cannot confirm this. Thanks to all who posted, it nice to know I am not the only one having install problems. John -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com]On Behalf Of kickstart-list-request at redhat.com Sent: Tuesday, March 22, 2005 9:52 AM To: kickstart-list at redhat.com Subject: Kickstart-list Digest, Vol 13, Issue 17 Send Kickstart-list mailing list submissions to kickstart-list at redhat.com To subscribe or unsubscribe via the World Wide Web, visit https://www.redhat.com/mailman/listinfo/kickstart-list or, via email, send a message with subject or body 'help' to kickstart-list-request at redhat.com You can reach the person managing the list at kickstart-list-owner at redhat.com When replying, please edit your Subject line so it is more specific than "Re: Contents of Kickstart-list digest..." From Matt.Fahrner at coat.com Tue Mar 22 15:57:08 2005 From: Matt.Fahrner at coat.com (Matt Fahrner) Date: Tue, 22 Mar 2005 10:57:08 -0500 Subject: e1000 still In-Reply-To: <2CE66A1ABBED8C4B85DFF5E0860987A32CB99F@usaflcms03.usa-spaceops.ksc.nasa.gov> References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB99F@usaflcms03.usa-spaceops.ksc.nasa.gov> Message-ID: <42404054.30509@coat.com> I haven't paid attention carefully, but I assume that you've set "spanning-tree portfast" if it's a 10 or 100mb. Or set "set spantree PortFast #/# enable". This usually solves these problems. http://www.cisco.com/warp/public/473/12.html Of course, having not paid attention to all of the earlier messages, this is probably useless and already dismissed... - Matt Poole, John J wrote: > James, > I will try the 30 second sleep as soon as I have the time. I think it may a workaround for the port tree spanning issue. I will post results when I test > it. I agree with everyone about this issue being ignored. I have spent many hours trying to make installs work. Sure adds credence to those who say Linux is > too labor intensive to be practical for business applications. I have read that this problem also shows up with other Ethernet cards but cannot confirm > this. Thanks to all who posted, it nice to know I am not the only one having install problems. > John > > > -----Original Message----- > From: kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com]On Behalf Of > kickstart-list-request at redhat.com > Sent: Tuesday, March 22, 2005 9:52 AM > To: kickstart-list at redhat.com > Subject: Kickstart-list Digest, Vol 13, Issue 17 > > > Send Kickstart-list mailing list submissions to > kickstart-list at redhat.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.redhat.com/mailman/listinfo/kickstart-list > or, via email, send a message with subject or body 'help' to > kickstart-list-request at redhat.com > > You can reach the person managing the list at > kickstart-list-owner at redhat.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kickstart-list digest..." > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list -- --------------------------------------------------------------------- Matt Fahrner 2 South Park St. Manager of Networking Willis House Burlington Coat Factory Warehouse Lebanon, N.H. 03766 TEL: (603) 448-4100 xt 5150 USA FAX: (603) 443-6190 Matt.Fahrner at COAT.COM --------------------------------------------------------------------- From jeroen at science.uva.nl Tue Mar 22 17:06:09 2005 From: jeroen at science.uva.nl (Jeroen Roodhart) Date: Tue, 22 Mar 2005 18:06:09 +0100 Subject: e1000 still Message-ID: <42405081.6090504@science.uva.nl> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 |-----Original Message----- |> From: James_Martin at ao.uscourts.gov |> [mailto:James_Martin at ao.uscourts.gov] |> Sent: Tuesday, March 22, 2005 8:52 AM |> To: Discussion list about Kickstart |> Subject: Re: e1000 still |> |> I agree. The less hacking the better. Red Hat could easily |> put a sleep |> variable in the init.c that is set at the syslinux command |> line. Have you |> type something like: |> |> linux network_wait=30 |I think spanning tree can take up to 45 seconds (Cisco gurus can |correct) and I've often seen it take some time. It is a good idea. Well, for what it is worth, we _did_ try numerous of combinations of portfast/spanning tree settings at the time and none worked with the e1000. At least for NFS mounts it seems reasonable to check if the mount worked and if not retry for a set number of times. Anaconda defaults to NFS over UDP (can one easily change this BTW?) so this to me just seems like prudent programming... Regards, Jeroen - -- Jeroen Roodhart University of Amsterdam jeroen at science.uva.nl Faculty of Science / ICT-Group Systeem- en netwerkbeheer Tel. 020 525 7203 / 06 51338165 - -- See http://www.science.uva.nl/~jeroen for openPGP public key -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) iD8DBQFCQFCB37AP1zFtDU0RAsNgAKCLyx98KXZTHzo+7bZBI2GmHmx6NACg1jsP P9wzG899b2QMeBI/XH3B+vw= =6FEt -----END PGP SIGNATURE----- From mikem.rtp at gmail.com Tue Mar 22 17:29:45 2005 From: mikem.rtp at gmail.com (Mike McLean) Date: Tue, 22 Mar 2005 12:29:45 -0500 Subject: execute shell script in postinstall In-Reply-To: <379313C94B76D2119AB60008C7A402E409E46384@IMSNOLAA> References: <379313C94B76D2119AB60008C7A402E409E46384@IMSNOLAA> Message-ID: <4f50e0680503220929a08703e@mail.gmail.com> wrote: > - Until you reboot, you have the limited installation environment and tool set. For example, sed is available, but awk is not. Actually, unless you add --nochroot to %post, you have access to everything that was just installed. From ebrown at lanl.gov Tue Mar 22 17:37:34 2005 From: ebrown at lanl.gov (Ed Brown) Date: Tue, 22 Mar 2005 10:37:34 -0700 Subject: e1000 still In-Reply-To: References: Message-ID: <1111513053.18219.35.camel@edbrown.lanl.gov> I don't believe there is a need to wait 30 or 45 seconds for spanning tree calculations on the switch. There is something much more subtle going on here. As I mentioned before, we've tried capturing traffic with a passive analyzer between the box being built and the switch, and just this change to the network connections allow the install to proceed normally (as does inserting a small 10/100 hub inline, or replacing the gig switch with a 100Mb switch.) It isn't about spanning tree calculations on the switch, I think there is some sub-second change in the characteristics of the negotiation or anaconda's network setup, when spanning tree is on or off, that make a difference. -Ed On Tue, 2005-03-22 at 07:51, James_Martin at ao.uscourts.gov wrote: > I agree. The less hacking the better. Red Hat could easily put a sleep > variable in the init.c that is set at the syslinux command line. Have you > type something like: > > linux network_wait=30 > > That parameter would then be passed to init which would then wait the > specified amount of time. Changing your network configuration to > accommodate a operating system install is ridiculous. > > Jeremy/Paul, opinions? > > > James > > > James S. Martin, RHCE > Contractor > Administrative Office of the United States Courts > Washington, DC > (202) 502-2394 > > kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: > > > On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > > > This is an age old problem going back to my RH 7.2 days. > > > > Yes, this is old news, hashed over on several lists, in many threads. > > James solution may help the original post-er to this thread, who was > > doing remote installs, with no access to the switches. > > > > The point, or mine anyway, is that hacking the code or having to modify > > switch configurations to be able to install the OS, are poor > > alternatives to be faced with. How is this not a problem with anaconda, > > and why hasn't RedHat owned it / fixed it? > > > > -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 jeroen at science.uva.nl Wed Mar 23 09:43:14 2005 From: jeroen at science.uva.nl (Jeroen Roodhart) Date: Wed, 23 Mar 2005 10:43:14 +0100 Subject: e1000 still Message-ID: <42413A32.8030005@science.uva.nl> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 |I don't believe there is a need to wait 30 or 45 seconds for spanning |tree calculations on the switch. There is something much more subtle |going on here. As I mentioned before, we've tried capturing traffic |with a passive analyzer between the box being built and the switch, and |just this change to the network connections allow the install to >proceed |normally (as does inserting a small 10/100 hub inline, or replacing the |gig switch with a 100Mb switch.) It isn't about spanning tree |calculations on the switch, I think there is some sub-second change in |the characteristics of the negotiation or anaconda's network setup, >when |spanning tree is on or off, that make a difference. | |-Ed I just want to support Ed here, this is the _exact_ behaviour that we experienced. I really don't see the issue here. I you retry try when it fails, you just make the proces more fault tolerant (normal programming I think, especially [as said before] when you use UDP). If it works in one go, you don't have any adverse effects (it's just as fast) and if it doesn't, it works with a short delay. What's the deal? In the mean time, when we experience this behaviour again (seems to have dissapeared when going to FC3) I'll patch the code myself. With kind regards, Jeroen - -- Jeroen Roodhart University of Amsterdam jeroen at science.uva.nl Faculty of Science / ICT-Group Systeem- en netwerkbeheer Tel. 020 525 7203 / 06 51338165 - -- See http://www.science.uva.nl/~jeroen for openPGP public key -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) iD8DBQFCQToy37AP1zFtDU0RAtnwAJ9RipKbZxzWWSRsbyVW/I39B7sH6wCeN2BS VAu/Rj9aha/xl0m37IcUGSE= =TY1P -----END PGP SIGNATURE----- From keanboon at gawab.com Wed Mar 23 09:48:01 2005 From: keanboon at gawab.com (keanboon) Date: Wed, 23 Mar 2005 09:48:01 GMT Subject: execute shell script in postinstall In-Reply-To: References: <20050322090359.13173.qmail@gawab.com> Message-ID: <20050323094801.20742.qmail@gawab.com> An HTML attachment was scrubbed... URL: From brilong at cisco.com Wed Mar 23 14:26:30 2005 From: brilong at cisco.com (Brian Long) Date: Wed, 23 Mar 2005 09:26:30 -0500 Subject: e1000 and NFS retry Message-ID: <1111587990.28418.0.camel@brilong-lnx2.cisco.com> I opened a Bug: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=151908 Feel free to add your 2 cents and also open support requests about this problem. If enough of us open support requests, they should address the issue in RHEL 3 and 4. /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s From ebrown at lanl.gov Wed Mar 23 15:44:15 2005 From: ebrown at lanl.gov (Ed Brown) Date: Wed, 23 Mar 2005 08:44:15 -0700 Subject: e1000 still In-Reply-To: <42413A32.8030005@science.uva.nl> References: <42413A32.8030005@science.uva.nl> Message-ID: <1111592655.7615.73.camel@edbrown.lanl.gov> On Wed, 2005-03-23 at 02:43, Jeroen Roodhart wrote: > If it works in one go, you don't have any adverse effects (it's just as > fast) and if it doesn't, it works with a short delay. What's the > deal? > In the mean time, when we experience this behaviour again (seems to > have dissapeared when going to FC3) I'll patch the code myself. One deal is that in some cases it doesn't work at all. When using ftp, frequently no amount of retrying at the 'Failed to log in to ftp server' screen succeeds (though I have seen occasional, eventual success at this point). Another deal, as in the case of the person who started this thread, is that remote installs are not possible. If you're happy patching broken code yourself, that's great, but please don't discount those who find this a much less than desirable solution. -Ed From jeroen at science.uva.nl Wed Mar 23 16:53:06 2005 From: jeroen at science.uva.nl (Jeroen Roodhart) Date: Wed, 23 Mar 2005 17:53:06 +0100 Subject: e1000 still Message-ID: <42419EF2.9010309@science.uva.nl> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 |If you're happy patching broken code yourself, that's great, but please |don't discount those who find this a much less than desirable solution. Heh, the woes of email conversation ;) What I (meant to have) implied is that they should do the failure recovery in the main stream code, as it doesn't affect normal performance and is better when there is a problem (UDP standard practise). But if for some (weird) reason the anaconda maintainers refuse to do that, I'll do it myself... Ofcourse I'd prefer not to. Cheers, Jeroen - -- Jeroen Roodhart University of Amsterdam jeroen at science.uva.nl Faculty of Science / ICT-Group Systeem- en netwerkbeheer Tel. 020 525 7203 / 06 51338165 - -- See http://www.science.uva.nl/~jeroen for openPGP public key -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) iD8DBQFCQZ7x37AP1zFtDU0RAgfCAKC57juEzEgt3dr6MwaN27HlJSFYdACdFtuz xFZzKhQ7T9Lw445XDriQrvk= =nuO1 -----END PGP SIGNATURE----- From asterr at pobox.com Wed Mar 23 17:46:56 2005 From: asterr at pobox.com (asterr) Date: Wed, 23 Mar 2005 09:46:56 -0800 (PST) Subject: e1000 still In-Reply-To: <1111513053.18219.35.camel@edbrown.lanl.gov> Message-ID: Ed, In my experience troubleshooting this issue, I have also seen that a passive analyzer (fluke) in between the host and the switch made the problem disappear. On closer analysis, this was found to be due to the fact that the analyzer we used established a circuit to the switch port, so the switch port was not shut down when the host was rebooted. Because the switch port never went down, the spanning tree did not need to be recalculated. -Aaron On Tue, 22 Mar 2005, Ed Brown wrote: > I don't believe there is a need to wait 30 or 45 seconds for spanning > tree calculations on the switch. There is something much more subtle > going on here. As I mentioned before, we've tried capturing traffic > with a passive analyzer between the box being built and the switch, and > just this change to the network connections allow the install to proceed > normally (as does inserting a small 10/100 hub inline, or replacing the > gig switch with a 100Mb switch.) It isn't about spanning tree > calculations on the switch, I think there is some sub-second change in > the characteristics of the negotiation or anaconda's network setup, when > spanning tree is on or off, that make a difference. > > -Ed > > > > On Tue, 2005-03-22 at 07:51, James_Martin at ao.uscourts.gov wrote: > > I agree. The less hacking the better. Red Hat could easily put a sleep > > variable in the init.c that is set at the syslinux command line. Have you > > type something like: > > > > linux network_wait=30 > > > > That parameter would then be passed to init which would then wait the > > specified amount of time. Changing your network configuration to > > accommodate a operating system install is ridiculous. > > > > Jeremy/Paul, opinions? > > > > > > James > > > > > > James S. Martin, RHCE > > Contractor > > Administrative Office of the United States Courts > > Washington, DC > > (202) 502-2394 > > > > kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: > > > > > On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > > > > This is an age old problem going back to my RH 7.2 days. > > > > > > Yes, this is old news, hashed over on several lists, in many threads. > > > James solution may help the original post-er to this thread, who was > > > doing remote installs, with no access to the switches. > > > > > > The point, or mine anyway, is that hacking the code or having to modify > > > switch configurations to be able to install the OS, are poor > > > alternatives to be faced with. How is this not a problem with anaconda, > > > and why hasn't RedHat owned it / fixed it? > > > > > > -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 ebrown at lanl.gov Wed Mar 23 18:34:56 2005 From: ebrown at lanl.gov (Ed Brown) Date: Wed, 23 Mar 2005 11:34:56 -0700 Subject: e1000 still In-Reply-To: References: Message-ID: <1111602895.7615.144.camel@edbrown.lanl.gov> Interesting. My experience with those tools is very limited, but that's not my idea of how a "passive" device would behave! But I wonder why this is only an issue with the e1000 driver and anaconda? And why aren't 30-45 second delays seen at the beginning of any install through a switch with spanning tree enabled? Also, I've moved network cables on switch ports and they are instantly working. It just doesn't add up to me that spanning tree calculations are really a factor here, if they indeed take that long. -Ed On Wed, 2005-03-23 at 10:46, asterr wrote: > Ed, > > In my experience troubleshooting this issue, I have also seen that a passive > analyzer (fluke) in between the host and the switch made the problem disappear. > > On closer analysis, this was found to be due to the fact that the analyzer > we used established a circuit to the switch port, so the switch port was > not shut down when the host was rebooted. Because the switch port never went > down, the spanning tree did not need to be recalculated. > > -Aaron > > On Tue, 22 Mar 2005, Ed Brown wrote: > > > I don't believe there is a need to wait 30 or 45 seconds for spanning > > tree calculations on the switch. There is something much more subtle > > going on here. As I mentioned before, we've tried capturing traffic > > with a passive analyzer between the box being built and the switch, and > > just this change to the network connections allow the install to proceed > > normally (as does inserting a small 10/100 hub inline, or replacing the > > gig switch with a 100Mb switch.) It isn't about spanning tree > > calculations on the switch, I think there is some sub-second change in > > the characteristics of the negotiation or anaconda's network setup, when > > spanning tree is on or off, that make a difference. > > > > -Ed > > > > > > > > On Tue, 2005-03-22 at 07:51, James_Martin at ao.uscourts.gov wrote: > > > I agree. The less hacking the better. Red Hat could easily put a sleep > > > variable in the init.c that is set at the syslinux command line. Have you > > > type something like: > > > > > > linux network_wait=30 > > > > > > That parameter would then be passed to init which would then wait the > > > specified amount of time. Changing your network configuration to > > > accommodate a operating system install is ridiculous. > > > > > > Jeremy/Paul, opinions? > > > > > > > > > James > > > > > > > > > James S. Martin, RHCE > > > Contractor > > > Administrative Office of the United States Courts > > > Washington, DC > > > (202) 502-2394 > > > > > > kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: > > > > > > > On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: > > > > > This is an age old problem going back to my RH 7.2 days. > > > > > > > > Yes, this is old news, hashed over on several lists, in many threads. > > > > James solution may help the original post-er to this thread, who was > > > > doing remote installs, with no access to the switches. > > > > > > > > The point, or mine anyway, is that hacking the code or having to modify > > > > switch configurations to be able to install the OS, are poor > > > > alternatives to be faced with. How is this not a problem with anaconda, > > > > and why hasn't RedHat owned it / fixed it? > > > > > > > > -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 > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From peter at boku.net Wed Mar 23 19:01:53 2005 From: peter at boku.net (Peter Eisch) Date: Wed, 23 Mar 2005 13:01:53 -0600 Subject: e1000 still In-Reply-To: <1111602895.7615.144.camel@edbrown.lanl.gov> Message-ID: The reason that it isn't as ubiquitous has to do with either hardware or driver implementation. With some drivers, like the e1000, it toggles the interface (physical carrier to the switch) after some late event like anaconda bringing the interface up. Other controllers will bring the interface up at power-on and when the driver (because of anaconda) brings the interface up, it's a no-op. The interface remains up. The time it takes from power-on to anaconda trying the interface is > 30 seconds, so things just continue properly. Let's assume that even the e1000 interface (as well as other controllers) bring the interface up at power-on, something takes the interface down when anaconda does the ifconfig up. It could be driver or hardware that fully oscillates the physical carrier, I don't know because I haven't looked. peter On 3/23/05 12:34 PM, "Ed Brown" wrote: > Interesting. My experience with those tools is very limited, but that's > not my idea of how a "passive" device would behave! > > But I wonder why this is only an issue with the e1000 driver and > anaconda? And why aren't 30-45 second delays seen at the beginning of > any install through a switch with spanning tree enabled? Also, I've > moved network cables on switch ports and they are instantly working. It > just doesn't add up to me that spanning tree calculations are really a > factor here, if they indeed take that long. > > -Ed > > > On Wed, 2005-03-23 at 10:46, asterr wrote: >> Ed, >> >> In my experience troubleshooting this issue, I have also seen that a passive >> analyzer (fluke) in between the host and the switch made the problem >> disappear. >> >> On closer analysis, this was found to be due to the fact that the analyzer >> we used established a circuit to the switch port, so the switch port was >> not shut down when the host was rebooted. Because the switch port never went >> down, the spanning tree did not need to be recalculated. >> >> -Aaron >> >> On Tue, 22 Mar 2005, Ed Brown wrote: >> >>> I don't believe there is a need to wait 30 or 45 seconds for spanning >>> tree calculations on the switch. There is something much more subtle >>> going on here. As I mentioned before, we've tried capturing traffic >>> with a passive analyzer between the box being built and the switch, and >>> just this change to the network connections allow the install to proceed >>> normally (as does inserting a small 10/100 hub inline, or replacing the >>> gig switch with a 100Mb switch.) It isn't about spanning tree >>> calculations on the switch, I think there is some sub-second change in >>> the characteristics of the negotiation or anaconda's network setup, when >>> spanning tree is on or off, that make a difference. >>> >>> -Ed >>> >>> >>> >>> On Tue, 2005-03-22 at 07:51, James_Martin at ao.uscourts.gov wrote: >>>> I agree. The less hacking the better. Red Hat could easily put a sleep >>>> variable in the init.c that is set at the syslinux command line. Have you >>>> type something like: >>>> >>>> linux network_wait=30 >>>> >>>> That parameter would then be passed to init which would then wait the >>>> specified amount of time. Changing your network configuration to >>>> accommodate a operating system install is ridiculous. >>>> >>>> Jeremy/Paul, opinions? >>>> >>>> >>>> James >>>> >>>> >>>> James S. Martin, RHCE >>>> Contractor >>>> Administrative Office of the United States Courts >>>> Washington, DC >>>> (202) 502-2394 >>>> >>>> kickstart-list-bounces at redhat.com wrote on 03/21/2005 07:42:13 PM: >>>> >>>>> On Mon, 2005-03-21 at 15:29, James_Martin at ao.uscourts.gov wrote: >>>>>> This is an age old problem going back to my RH 7.2 days. >>>>> >>>>> Yes, this is old news, hashed over on several lists, in many threads. >>>>> James solution may help the original post-er to this thread, who was >>>>> doing remote installs, with no access to the switches. >>>>> >>>>> The point, or mine anyway, is that hacking the code or having to modify >>>>> switch configurations to be able to install the OS, are poor >>>>> alternatives to be faced with. How is this not a problem with anaconda, >>>>> and why hasn't RedHat owned it / fixed it? >>>>> >>>>> -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 >>> >> >> _______________________________________________ >> 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 ebrown at lanl.gov Wed Mar 23 21:53:32 2005 From: ebrown at lanl.gov (Ed Brown) Date: Wed, 23 Mar 2005 14:53:32 -0700 Subject: e1000 and NFS retry In-Reply-To: <1111587990.28418.0.camel@brilong-lnx2.cisco.com> References: <1111587990.28418.0.camel@brilong-lnx2.cisco.com> Message-ID: <1111614812.7615.251.camel@edbrown.lanl.gov> I plan to add to your bug report, though my experience is with ftp installs. The recent discussion makes me think I need to do some more experimenting (though I dread wasting more time troubleshooting this problem). I haven't conciously waited for more than a minute at the "failed to log in to ftp server" screen before hitting enter to try again, to see if allowing that extra time makes a consistent difference. Thanks for opening this up... -Ed On Wed, 2005-03-23 at 07:26, Brian Long wrote: > I opened a Bug: > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=151908 > > Feel free to add your 2 cents and also open support requests about this > problem. If enough of us open support requests, they should address the > issue in RHEL 3 and 4. > > /Brian/ From info at hostinthebox.net Wed Mar 23 23:05:19 2005 From: info at hostinthebox.net (dan) Date: Wed, 23 Mar 2005 16:05:19 -0700 Subject: Proper location of ks.cfg file Message-ID: <4241F62F.4030702@hostinthebox.net> Hello, all - I've been toying around with Kickstart-based installations, and even gone as far as making a PHP page that dynamically generates a kickstart install file, that is available over HTTP kickstart, for a specific machine based on MAC address. However, I have never been able to completely automate the process. I've always had to specify ks= at the command-prompt. I've seen several documents that mention how a kickstart file is initialized, but very few that seem sure on the physical location on disk of the ks.cfg file. Some say inside isolinux/, some say on the root of the disk. Others I've read say that this "cannot be done". Even RedHat's own Administartor's Guide isn't particularly clear. So I guess what I'm asking here is, what should the physical location of the ks.cfg file be? I suppose I can manipulate this, relative to the disk layout, by editing syslinux.cfg, but I'd rather do it the right way, and not have to create some dirty hack for this issue. Thanks for your time -dant From jim at rossberry.com Thu Mar 24 01:29:42 2005 From: jim at rossberry.com (Jim Wildman) Date: Wed, 23 Mar 2005 20:29:42 -0500 (EST) Subject: Proper location of ks.cfg file In-Reply-To: <4241F62F.4030702@hostinthebox.net> References: <4241F62F.4030702@hostinthebox.net> Message-ID: On Wed, 23 Mar 2005, dan wrote: > I've been toying around with Kickstart-based installations, and even > gone as far as making a PHP page that dynamically generates a kickstart > install file, that is available over HTTP kickstart, for a specific > machine based on MAC address. I would be very interested in seeing your code. > > However, I have never been able to completely automate the process. > I've always had to specify ks= at the command-prompt. Can't get away, from that, but you can automate it several ways. 1) pxe You can specify your ks option as part of a pxe invocation. ie.. [root at hross pxelinux.cfg]# cat base.cfg prompt=0 label linux kernel vmlinuz-34 append rw load_ramdisk=1 prompt_ramdisk=0 ramdisk_size=16384 initrd=initrd-34.img ks=http://10.8.80.254/centos34/base-ks.cfg vnc (the last three lines have to be a single line) Go here http://www.owlriver.com/tips/pxe-install/ for a short cookbook on doing pxe. Also here http://www.colug.net/notes/0502mtg/ 2) You can modify isolinux.cfg on a boot cd so the default stanza has the correct invocation 3) If you are rebuilding boxes, pull the initrd.img and vmlinuz off of boot.iso and create an rpm that has a %post script that uses grubby to add the correct info to /etc/grub.conf Here is a %post section... %post /sbin/grubby --add-kernel=/boot/rh3boot.vmlinuz \ --initrd /boot/rh3boot.initrd.img \ --args "console=ttyS0,tty0 ks=http://10.129.7.209/ksbase/ksconf.php?type =rhas30u3as" \ --title "Red Hat 3 install ($version)" /sbin/grubby --add-kernel=/boot/rh3boot.vmlinuz \ --initrd /boot/rh3boot.initrd.img \ --args "ks=nfs:10.128.138.253:/nfsserver/lrh3/installdb/lrh3-u3-ks.cfg" \ --title "Red Hat 3 install ($version) test" ------------------------------------------------------------------------ Jim Wildman, CISSP, RHCE jim at rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine From john.j.poole at usa-spaceops.com Thu Mar 24 18:26:51 2005 From: john.j.poole at usa-spaceops.com (Poole, John J) Date: Thu, 24 Mar 2005 13:26:51 -0500 Subject: e1000 still Message-ID: <2CE66A1ABBED8C4B85DFF5E0860987A32CB9A7@usaflcms03.usa-spaceops.ksc.nasa.gov> James, I tried the sleep 30 in the init but my "HTTP" install still fails. I hope I am not fighting another problem, but it acts just like the nfs failure. I get a messages stating that netstg2.img could not be downloaded. When I select OK to the popup windows without changing the settings, the install continues without failing. Thanks for the help. I wish I had had your init fix earlier it was easier than modifying imount.c and dhcp.c. Thanks, John -----Original Message----- From: James_Martin at ao.uscourts.gov [mailto:James_Martin at ao.uscourts.gov] Sent: Monday, March 21, 2005 5:29 PM To: Discussion list about Kickstart Subject: Re: e1000 still This is an age old problem going back to my RH 7.2 days. It does have to do with span & tree and port fast and all of that stuff. I had to summon my 1337 C programming skills to solve it. Get a copy of the anaconda source rpm, install it, and take a look at the init.c in the loader2/ directory. What I did was hard coded a 30 second sleep in there, and never had a problem since. Of course after you mod it and compile it, you'll have to put it inito a new initrd.img. #> diff init.c.old init.c 615a616,618 > > printf("sleeping 30 seconds.. thanks JSM!\n"); > sleep(30); in context it looks like this: if (!testing) doklog("/dev/tty4"); /* Go into normal init mode - keep going, and then do a orderly shutdown when: 1) /bin/install exits 2) we receive a SIGHUP */ printf("sleeping 30 seconds.. thanks JSM!\n"); sleep(30); printf("running install...\n"); setsid(); This basically forces it to sleep for whatever time you specify until loader is launched. By that time, the network should be up and ready. I know it's hacky, but it's worked for the past 4 years. James James S. Martin, RHCE Contractor Administrative Office of the United States Courts Washington, DC (202) 502-2394 kickstart-list-bounces at redhat.com wrote on 03/21/2005 05:01:11 PM: > On Mon, 21 Mar 2005, Ed Brown wrote: > > >It's hard to understand this position by RedHat. Anaconda is clearly > >NOT working as designed. > > It's the error messages (or lack thereof) which annoy me. Hands up > [virtually] whoever has been dropped to the "Language Select" screen > after a misconfigured kickstart attempt? > > It's almost enough to want to patch it... > > Do spanning-tree links appear "up" to the host, before the switch > relents and starts to forward? If not, perhaps it's possible to > programmatically delay in a %pre script. > > > > Cheers, > Phil > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From azeem81 at msn.com Thu Mar 24 20:33:20 2005 From: azeem81 at msn.com (azeem ahmad) Date: Fri, 25 Mar 2005 01:33:20 +0500 Subject: FC installation Message-ID: hi list i want to install FC3 from a hard disk. i.e. like i have a four cd set and i boot my computer from first cd and anaconda starts and the installation processes automatically starts and i change the disks and installation gets complete. the same way i want all the four cds copied onto a hard disk say ide1. and i want to boot my computer from that ide1 in the same way like boots with first installation cd rom. and then automatically start anaconds to complete the installation. is there any ways Regards Azeem _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ From florin at andrei.myip.org Fri Mar 25 08:00:20 2005 From: florin at andrei.myip.org (Florin Andrei) Date: Fri, 25 Mar 2005 00:00:20 -0800 Subject: copy files from CD to hard-drive Message-ID: <1111737620.7568.28.camel@rivendell.home.local> I created a custom DVD installer, that also has a kickstart file. On the custom DVD i also put some files that i want to copy to the installed system after kickstart finishes up its job. The problem is, i can't seem to find a way to do that. I have a pretty large %post section, wich must NOT be run with -- nochroot (a lot of commands must be run on the target system). But from a chrooted %post i cannot access the install DVD. If i add after that another %post, this time with --nochroot, i get "Error code 256 encountered running a kickstart %pre/%post script" This seems like such a typical situation, it's impossible that someone else didn't solve it already. So, how do i copy those supplemental files from the install DVD? -- Florin Andrei http://florin.myip.org/ From brilong at cisco.com Fri Mar 25 13:54:10 2005 From: brilong at cisco.com (Brian Long) Date: Fri, 25 Mar 2005 08:54:10 -0500 Subject: copy files from CD to hard-drive In-Reply-To: <1111737620.7568.28.camel@rivendell.home.local> References: <1111737620.7568.28.camel@rivendell.home.local> Message-ID: <1111758850.17043.20.camel@brilong-lnx2.cisco.com> On Fri, 2005-03-25 at 03:00, Florin Andrei wrote: > I created a custom DVD installer, that also has a kickstart file. On the > custom DVD i also put some files that i want to copy to the installed > system after kickstart finishes up its job. > > The problem is, i can't seem to find a way to do that. > I have a pretty large %post section, wich must NOT be run with -- > nochroot (a lot of commands must be run on the target system). But from > a chrooted %post i cannot access the install DVD. > If i add after that another %post, this time with --nochroot, i get > "Error code 256 encountered running a kickstart %pre/%post script" > > This seems like such a typical situation, it's impossible that someone > else didn't solve it already. So, how do i copy those supplemental files > from the install DVD? %post --nochroot copy files from /mnt/cdrom/foo to /mnt/sysimage/foo %post run all your postinstall routines /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s From jkeating at j2solutions.net Fri Mar 25 23:47:13 2005 From: jkeating at j2solutions.net (Jesse Keating) Date: Fri, 25 Mar 2005 15:47:13 -0800 Subject: FC installation In-Reply-To: References: Message-ID: <1111794433.3471.249.camel@jkeating2.hq.pogolinux.com> On Fri, 2005-03-25 at 01:33 +0500, azeem ahmad wrote: > hi list > i want to install FC3 from a hard disk. i.e. like i have a four cd set > and i > boot my computer from first cd and anaconda starts and the > installation > processes automatically starts and i change the disks and installation > gets > complete. the same way i want all the four cds copied onto a hard disk > say > ide1. and i want to boot my computer from that ide1 in the same way > like > boots with first installation cd rom. and then automatically start > anaconds > to complete the installation. is there any ways You can insert into your current grub or lilo config a line that will boot the installer kernel (use maybe the pxe vmlinuz and initrd) with the 'askmethod' string. Then point to the directory that holds the rest of the install sets. -- Jesse Keating RHCE (geek.j2solutions.net) Fedora Legacy Team (www.fedoralegacy.org) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) Was I helpful? Let others know: http://svcs.affero.net/rm.php?r=jkeating From seph at directionless.org Sat Mar 26 03:28:07 2005 From: seph at directionless.org (seph) Date: Fri, 25 Mar 2005 22:28:07 -0500 Subject: e1000 still In-Reply-To: <423C403D.90400@computer.org> (S. W. Davison's message of "Sat, 19 Mar 2005 10:07:41 -0500") References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> Message-ID: "S. W. Davison" writes: > We filed a support request with Red Hat. Their > response was basically that anaconda and the rest of > the installer were working as designed, and that we > should alter the parameters on our HP Procurve > switches (turning "spanning tree" off and turning > "portfast" on were suggested). That's not in keeping with the errors I was seeing. My understanding is that if it was a STP problem, the machine wouldn't get an addresses or be pingable. I was also seeing problems on small gigabit switches that didn't do STP. At this point, I haven't thought about it in awhile. So I forget the details of what I was seeing. seph From ebrown at lanl.gov Sat Mar 26 15:49:49 2005 From: ebrown at lanl.gov (Edward F. Brown) Date: Sat, 26 Mar 2005 08:49:49 -0700 (MST) Subject: e1000 still In-Reply-To: References: <2CE66A1ABBED8C4B85DFF5E0860987A32CB994@usaflcms03.usa-spaceops.ksc.nasa.gov> <423C403D.90400@computer.org> Message-ID: <32826.128.165.7.144.1111852189.squirrel@webmail.lanl.gov> >> We filed a support request with Red Hat. Their >> response was basically that anaconda and the rest of >> the installer were working as designed, and that we >> should alter the parameters on our HP Procurve >> switches (turning "spanning tree" off and turning >> "portfast" on were suggested). > > That's not in keeping with the errors I was seeing. My understanding > is that if it was a STP problem, the machine wouldn't get an addresses > or be pingable. I was also seeing problems on small gigabit switches > that didn't do STP. > > At this point, I haven't thought about it in awhile. So I forget the > details of what I was seeing. It's been some weeks since I looked closely at the problem too, but this fits with what I remember, and why the association with STP seems incidental, not causal, to me. The reverse dns lookup would succeed, the 2nd stage image would be retrieved, and the last message on the alt-F3 console, when "Failed to log in to ftp server" appeared on F1, would be about trying to retrieve the kickstart file. -Ed From florin at andrei.myip.org Sat Mar 26 20:44:27 2005 From: florin at andrei.myip.org (Florin Andrei) Date: Sat, 26 Mar 2005 12:44:27 -0800 Subject: copy files from CD to hard-drive In-Reply-To: <1111758850.17043.20.camel@brilong-lnx2.cisco.com> References: <1111737620.7568.28.camel@rivendell.home.local> <1111758850.17043.20.camel@brilong-lnx2.cisco.com> Message-ID: <1111869868.7245.3.camel@rivendell.home.local> On Fri, 2005-03-25 at 08:54 -0500, Brian Long wrote: > On Fri, 2005-03-25 at 03:00, Florin Andrei wrote: > > I created a custom DVD installer, that also has a kickstart file. On the > > custom DVD i also put some files that i want to copy to the installed > > system after kickstart finishes up its job. > > > > The problem is, i can't seem to find a way to do that. > > %post --nochroot > > copy files from /mnt/cdrom/foo to /mnt/sysimage/foo > > %post > > run all your postinstall routines Hm, even when placing the non-chrooted %post before the chrooted one i still get error 256. The problem seems to be that, at %post time, the installer CD/DVD is already unmounted AND the directory /mnt/source has disappeared. The solution was to do this: %post --nochroot mkdir -p /mnt/source mount -o ro /tmp/cdrom /mnt/source cp -a /mnt/source/foo /mnt/sysimage/bar umount /tmp/cdrom %post blah All this (mount/umount, mkdir) should be mentioned in the kickstart HOWTO, it's not at all intuitive. -- Florin Andrei http://florin.myip.org/ From florin at andrei.myip.org Sat Mar 26 20:45:37 2005 From: florin at andrei.myip.org (Florin Andrei) Date: Sat, 26 Mar 2005 12:45:37 -0800 Subject: fstab - no labels Message-ID: <1111869937.7245.6.camel@rivendell.home.local> Is there a way to tell anaconda to create /etc/fstab with no labels (but using actual /dev nodes instead)? -- Florin Andrei http://florin.myip.org/ From mshuler at rackspace.com Sat Mar 26 23:21:23 2005 From: mshuler at rackspace.com (Michael Shuler) Date: Sat, 26 Mar 2005 17:21:23 -0600 Subject: fstab - no labels In-Reply-To: <1111869937.7245.6.camel@rivendell.home.local> References: <1111869937.7245.6.camel@rivendell.home.local> Message-ID: <4245EE73.80402@rackspace.com> Florin Andrei wrote: > Is there a way to tell anaconda to create /etc/fstab with no labels (but > using actual /dev nodes instead)? Yes, there is, and I set this up for all kickstarts - labels are bad news when you have more than one partition/device that contains LABEL=/, such as new and old OS drives... It is also a good idea to edit /boot/grub/grub.conf during %post, to remove 'LABEL=/' and replace it with the proper device name. Create an updates.img in the 'base' directory of your distribution, which contains the file 'fsset.py', patched from the anaconda source. Here's my patch from rhel-4: --- fsset.py-orig 2005-01-21 15:25:10.000000000 -0600 +++ fsset.py 2005-01-21 15:27:55.000000000 -0600 @@ -1048,10 +1048,10 @@ fstab = "" for entry in self.entries: if entry.mountpoint: - if entry.getLabel(): - device = "LABEL=%s" % (entry.getLabel(),) - else: - device = devify(entry.device.getDevice()) +# if entry.getLabel(): +# device = "LABEL=%s" % (entry.getLabel(),) +# else: + device = devify(entry.device.getDevice()) fstab = fstab + entry.device.getComment() fstab = fstab + format % (device, entry.mountpoint, entry.fsystem.getName(), Kind Regards, Michael Shuler Rackspace Managed Hosting