kernel update hoses system
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me. Alas, this has comprehensively hosed my system and I can no longer boot up, messages complaining about being unable to find the correct reiserfs and sata modules. Any way out or is it a reinstall? In which case I think it will be with Debian as I can't see them issuing an update like this. Hello there, SuSE techs: if you want to issue kernel updates that hose systems, be my guest. But in that case SuSE will join the couple of hundred other distros out there and few or no people use because they are so buggy and unreliable. There was a time when SuSE was a cut above all this. :) Fish
On Friday 24 December 2004 10:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
Alas, this has comprehensively hosed my system and I can no longer boot up, messages complaining about being unable to find the correct reiserfs and sata modules.
Any way out or is it a reinstall? In which case I think it will be with Debian as I can't see them issuing an update like this.
Hello there, SuSE techs: if you want to issue kernel updates that hose systems, be my guest. But in that case SuSE will join the couple of hundred other distros out there and few or no people use because they are so buggy and unreliable. There was a time when SuSE was a cut above all this.
apt is hardly a supported method of upgrading packages. If you want to complain about suse methods, you should use suse methods. You can boot from the CD and select the rescue system. Then mount your root partition and chroot to it. Something like mkdir tmp mount /dev/hda2 tmp chroot tmp Then mount /boot in case it is on a separate partition, and reinstall the kernel rpm from /var/lib/YaST2/you/mnt/i386/update/9.2/rpm/i586/ You might also want to check what apt did, how it upgraded the kernel rpm, did it use -i or -U?
On Friday 24 December 2004 10:34, Anders Johansson wrote:
Then mount /boot in case it is on a separate partition, and reinstall the kernel rpm from /var/lib/YaST2/you/mnt/i386/update/9.2/rpm/i586/
Correction: reinstall the kernel from wherever it is that apt downloads it to. The above is where YOU stores downloaded packages, I have no idea where apt puts them
On Friday 24 December 2004 10:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
Alas, this has comprehensively hosed my system and I can no longer boot up, messages complaining about being unable to find the correct reiserfs and sata modules.
Any way out or is it a reinstall? In which case I think it will be with Debian as I can't see them issuing an update like this.
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones From the rescue system mkdir tmp mount /dev/hda2 tmp (or whatever your root partition is) chroot tmp mount /boot (if it's on a separate partition) mk_initrd reboot, and you should be fine
On Friday 24 December 2004 10:04, Anders Johansson wrote:
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones
From the rescue system
mkdir tmp mount /dev/hda2 tmp (or whatever your root partition is) chroot tmp mount /boot (if it's on a separate partition) mk_initrd
reboot, and you should be fine
Thanks for that Anders you saved my Bacon. -- David Bottrill david@bottrill.org www.bottrill.org Registered Linux user number 330730 Internet SIP Phone: 1-747-244-2699
Anders Johansson wrote:
On Friday 24 December 2004 10:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
OK, I got the same problem on one machine. The problem is with how the initrd
Soon I expect to have more email in "security announcements" just on this problem than I have in the folder for this list. I've been waiting until the 9.2 ftp is available (because right now I'm too poor to afford even CA$112.14 plus shipping and taxes). Now I am beginning to wonder if I should just go ahead and install 9.1, or stay with 9.0 until 9.3 (or 10.0 or whatever is next) comes out and install that. Either way, 9.2 is beginning to like it is not a safe bet.
Darryl Gregorash wrote:
Anders Johansson wrote:
On Friday 24 December 2004 10:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
OK, I got the same problem on one machine. The problem is with how the initrd
Soon I expect to have more email in "security announcements" just on this problem than I have in the folder for this list.
I've been waiting until the 9.2 ftp is available (because right now I'm too poor to afford even CA$112.14 plus shipping and taxes). Now I am beginning to wonder if I should just go ahead and install 9.1, or stay with 9.0 until 9.3 (or 10.0 or whatever is next) comes out and install that. Either way, 9.2 is beginning to like it is not a safe bet.
If you would like to hear a voice from the wilderness: stay with v9.0 and save your pennies for a later release than 9.2. Maybe by then SuSE will buy back SUSE from Novell and we will all live happily everafter because SuSe will be once again the reliable SuSE that we know (er, knew). Cheers. -- A closed mouth gathers no foot.
Anders Johansson wrote: [snip]
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones
From the rescue system
mkdir tmp mount /dev/hda2 tmp (or whatever your root partition is) chroot tmp mount /boot (if it's on a separate partition) mk_initrd
reboot, and you should be fine
OK, this sorted me out and thank you very much for posting the information. FWIW, apt stores its files in /var/cache/apt/archives. This isn't the appropriate forum but I wish SuSE would support apt in some form or other because it is so fast and convenient. Apologies if I was a bit short earlier on. This must be a very complicated business and having followed SuSE for a long time now I know that the staff work extremely hard. Best wishes to you for a good break over the next couple of weeks. :) Fish
On Friday 24 December 2004 02:04, Anders Johansson wrote:
On Friday 24 December 2004 10:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
Alas, this has comprehensively hosed my system and I can no longer boot up, messages complaining about being unable to find the correct reiserfs and sata modules.
Any way out or is it a reinstall? In which case I think it will be with Debian as I can't see them issuing an update like this.
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones
So the new kernel is buggy? Jerome
From the rescue system
mkdir tmp mount /dev/hda2 tmp (or whatever your root partition is) chroot tmp mount /boot (if it's on a separate partition) mk_initrd
reboot, and you should be fine
On Fri, Dec 24, 2004 at 11:37:17AM -0800, Susemail wrote:
On Friday 24 December 2004 02:04, Anders Johansson wrote:
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones
So the new kernel is buggy?
It depends on your definition of "buggy". Regarding the problem with initrd please read http://lists.suse.com/archive/suse-security/2004-Dec/0118.html Regards, -Kastus
Alle 20:37, venerdì 24 dicembre 2004, Susemail ha scritto:
So the new kernel is buggy? Jerome
I think so. I've had the same problem. Thanks to a friend who was so patient to listen to me (i was a little angry) now i reverted to a good kernel. But really, pay attention to 2.6.8.24-10 default... I hope that a patch will be released soon. Cause i my anger time i was really on the road to erase all linux partitions on my hd
On Friday 24 December 2004 4:02 pm, Davide Dolcini wrote:
Alle 20:37, venerdì 24 dicembre 2004, Susemail ha scritto:
So the new kernel is buggy? Jerome
I think so.
I've had the same problem. Thanks to a friend who was so patient to listen to me (i was a little angry) now i reverted to a good kernel.
But really, pay attention to 2.6.8.24-10 default...
I hope that a patch will be released soon. Cause i my anger time i was really on the road to erase all linux partitions on my hd
The kernel is good, the install script is flawed, but reverting to the older kernel exposes your sys to a number of security flaws. Follow the URL that Kastus posted a few minutes ago and you get the bug fixes and the security patches in 2.6.8-24.10 PeterB -- -- Proud SUSE user since 5.2 Loving SUSE 9.2 My BLOG == http://vancampen.org/blog --
On Friday 24 December 2004 14:37, Susemail wrote:
On Friday 24 December 2004 02:04, Anders Johansson wrote:
OK, I got the same problem on one machine. The problem is with how the initrd gets created. For some reason it seems to pull in the wrong modules, the UML modules instead of the proper ones
So the new kernel is buggy? Jerome
The patches are buggy, yes. I simply rolled back to the default kernel by booting in manual mode, then running YaST. I haven't tried the initrd fix yet. I was getting reiserfs errors suggesting the modules were UML and not default. -- Regards, Steven
Mark, You don't say, but I'm going to guess that you're running version 9.2? A little while after the kernel upgrade came out, this was posted to the SuSE-Announce list:
Subject: "Re: SUSE Security Announcement: various kernel problems (SUSE-SA:2004:044)" Like some of you noted the 9.2 kernels RPMs are not there anymore, even though they are listed in the advisory.
This is due us receiving reports where the Yast Online Update updating this kernel made the machine fail to boot.
The reason is that a script is no longer creating the vmlinuz and initrd symlinks as is required for successful booting.
We are preparing updated fixed packages for this problem and will be releasing them hopefully within the next day(s).
Sorry for the inconvience, Marcus Meissner, Team Lead SUSE Security
The third paragraph is the key. Furthermore, it was discussed in this group in a thread, ",boot grub problemFW: [suse-security-announce] Re: SUSE Security Announcement: various kernel problems (SUSE-SA:2004:044)" initiated on the 22nd. If you can boot somehow (install CDs, e.g.) and mount whatever partition contains you system's "/boot" directory, you can make the symbolic links missing from the update. I believe they are: initrd -> initrd-2.6.8-24.5-default vmlinuz -> vmlinuz-2.6.8-24.5-default Of course, I'm merely passing on what Doug B. reported in the thread on SuSE-Linux-E. If you're, for example, running an SMP kernel, then the "-default" suffix will be "-smp" instead. Good luck. Randall Schulz On Friday 24 December 2004 01:24, Mark Crean wrote:
A new kernel came out this morning so I updated to it using apt, having seen a security announcement a couple of days ago. How very silly of me.
Alas, this has comprehensively hosed my system and I can no longer boot up, messages complaining about being unable to find the correct reiserfs and sata modules.
Any way out or is it a reinstall? In which case I think it will be with Debian as I can't see them issuing an update like this.
Hello there, SuSE techs: if you want to issue kernel updates that hose systems, be my guest. But in that case SuSE will join the couple of hundred other distros out there and few or no people use because they are so buggy and unreliable. There was a time when SuSE was a cut above all this.
:)
Fish
Well not fully: Randall R Schulz wrote:
The third paragraph is the key.
Furthermore, it was discussed in this group in a thread, ",boot grub problemFW: [suse-security-announce] Re: SUSE Security Announcement: various kernel problems (SUSE-SA:2004:044)" initiated on the 22nd.
If you can boot somehow (install CDs, e.g.) and mount whatever partition contains you system's "/boot" directory, you can make the symbolic links missing from the update. I believe they are:
initrd -> initrd-2.6.8-24.5-default vmlinuz -> vmlinuz-2.6.8-24.5-default
In fact it is ".10-default". So that it should look as follows: initrd -> initrd-2.6.8-24.10-default vmlinuz -> vmlinuz-2.6.8-24.10-default Yours Martin
Randall R Schulz wrote:
Mark,
You don't say, but I'm going to guess that you're running version 9.2? A little while after the kernel upgrade came out, this was posted to the SuSE-Announce list:
[snip] Thanks, yes, 9.2. On checking back I received this but didn't read it. I did know vaguely from quick scanning of subject headings in posts that there was a problem with the kernel update so I didn't update when I first heard about it, a couple of days ago. Then the stuff appeared on apt early this morning and I thought, aha, here are the promised revises to the update so it's now safe to install them. Only it wasn't. It's my fault for not checking things out more thoroughly. First cup of coffee + 8 a.m. + Christmas Eve = brain not yet in gear. Anyway, after some helfpul advice I am back up and running again. :) Fish
participants (12)
-
Anders Johansson -
Basil Chupin -
Darryl Gregorash -
David Bottrill -
Davide Dolcini -
Kastus -
Mark Crean -
Martin Deppe -
Peter B Van Campen -
Randall R Schulz -
Steven T. Hatton -
Susemail