Unbootable system after fsck failure

Bug #570134 reported by Jeroen T. Vermeulen on 2010-04-26
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Undecided
Unassigned

Bug Description

This may still be the same as bug 553374, though I thought that had been fixed and apparently it hasn't. The circumstances are not the same as in bug 563618; my hwclock seems to be fine.

On the laptop I upgraded from karmic to lucid, a failure during boot-time fsck again gave me a perpetual boot screen without any feedback and without any way out. (I know about the fsck error because pressing <esc> brought me to a text console that showed the output). I had to boot from a live CD again and run fsck manually.

So far I suspect the proprietary nVidia driver of causing the filesystem corruption; I don't recall the corruption happening while I was running without it. Another user with the same corruption problems in Karmic also had an nVidia card. If that is indeed the cause, a large class of users may find that Lucid spontaneously stops working after a while and from then on fails to boot without help from an expert and a live CD, no error message given. :(

summary: - Unbootable system after sck failure
+ Unbootable system after fsck failure
Fabio Marconi (fabiomarconi) wrote :

Hello
Is this problem present with the latest updated Lucid's or Maverick's packages?
Thanks in advance
Fabio

Changed in ubuntu:
status: New → Incomplete
Jeroen T. Vermeulen (jtv) wrote :

The symptoms have changed. It's now breaking as follows:

After the filesystem has corrupted and gone read-only, the kernel will not boot at all. There's no sign of the purple boot screen. I tried with Ubuntu kernels 2.6.35.22, 2.6.34.24, and a "2.6.36.999" upstream kernel that I was asked to install in the course of diagnosing another bug. Each of them dropped me into a busybox, though I believe the 2.6.34 kernel gave a lot of output and then hung, making it harder to see that I was in a busybox shell.

I booted off an old Lucid beta CD and ran e2fsck -v -f on my root partition. Dozens or perhaps a few hundred messages, mostly fsck counting more free blocks than expected (usually about a dozen more).

Now the system will boot from the 2.6.34 kernel; the 2.6.35 and 2.6.36 ones give a message (still in the white-on-black text console, so without ever getting to the purple boot screen) along the lines of udevadm callbacks not being allowed before udev is configured. The system won't budge after that. This is right after the "Starting up ..." message on the text console.

Fabio Marconi (fabiomarconi) wrote :

Hello
is possible from console launch
nano /var/log/dmesg or syslog or ker.log
of the affected kernels?
Thanks
Fabio

Jeroen T. Vermeulen (jtv) wrote :

I don't understand the question. The affected kernels do not give me a console. They only gave me a busybox console before I fsck'ed. Now that I've done that, all I can do is wait until the next time the filesystem corrupts.

Jeroen T. Vermeulen (jtv) wrote :

I'm sorry, I meant to say: the affected kernels do not give me a console _I can type in_.

Jeroen T. Vermeulen (jtv) wrote :

Some small corrections on what I said earlier:

Actually the kernel I can still boot is 2.6.32-24, not 2.6.34.24.

The failure message from kernels 2.6.35 (Ubuntu) and 2.6.26 (upstream) is:
udevadm trigger is not permitted while udev is unconfigured

I guess this is no longer related to the fsck though and should probably go into a separate bug.

Launchpad Janitor (janitor) wrote :

[Expired for Ubuntu because there has been no activity for 60 days.]

Changed in ubuntu:
status: Incomplete → Expired
Jeroen T. Vermeulen (jtv) wrote :

This bug may possibly be obsolete: now, in Maverick, a corrupted filesystem doesn't give me a boot screen at all. Instead I'm thrown into busybox (which is probably worse).

Changed in ubuntu:
status: Expired → New
shankao (shankao) on 2011-01-07
tags: added: karmic2lucid
shankao (shankao) wrote :

Once you're in that busybox, could you make your way to dmesg logs somehow? It would be nice to have them in order to do something with this bug.
I'm also changing this bug to the linux kernel, as leaving it in the general "ubuntu" is not usually a good idea. For future reports, please take a look to https://wiki.ubuntu.com/Bugs/FindRightPackage
Thanks!

affects: ubuntu → linux (Ubuntu)
Jeroen T. Vermeulen (jtv) wrote :

I think the last few times the busybox was not responsive to input. Even if it was, I have no idea how to get to dmesg from there!

Anyway, at the moment I'm booting again so it'll have to wait until the next time my FS corrupts.

shankao (shankao) wrote :

In the case that you get a no network machine, you can also try with a USB stick (if you have one). If busybox works you should have the mount and dmesg commands avaliable :)

Changed in linux (Ubuntu):
status: New → Incomplete
Jeroen T. Vermeulen (jtv) wrote :

Ah, I didn't know that, thanks.

By the way, USB sticks are probably not an option—I don't think most USB images support EFI.

Launchpad Janitor (janitor) wrote :

[Expired for linux (Ubuntu) because there has been no activity for 60 days.]

Changed in linux (Ubuntu):
status: Incomplete → Expired
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers