2.6.11.11 Assertion failure in journal_commit_transaction()

2.6.11.11 Assertion failure in journal_commit_transaction()

[Holger Kiehl]

Hello

On a busy server (dual Xeon with 4GB ram) running plain 2.6.11.11 form
kernel.org (distribution is Fedora Core 2) I got the following error:

Jun  4 11:40:55 apollo kernel: Assertion failure in journal_commit_transaction() at fs/jbd/commit.c:768: "jh->b_next_transaction == ((void *)0)"
Jun  4 11:40:55 apollo kernel: ------------[ cut here ]------------
Jun  4 11:40:55 apollo kernel: kernel BUG at fs/jbd/commit.c:768!
Jun  4 11:40:55 apollo kernel: invalid operand: 0000 [#1]
Jun  4 11:45:07 apollo syslogd 1.4.1: restart.

The system bootet automatically, from the ipmi motherboard logs I can see
that the hardware watchdog has reset the system at 11:42:55.
It's using an ext3 on top of a software raid10 (three raid1 combined with
raid0 over 6 SCSI disks). None of the software raids needed to rebuild after
the crash.

Any idea why the kernel stopped? The system has been running stable for a
year now.

If I forgot any information please ask so I can provide them.

Thanks,
Holger
-- 

Re: 2.6.11.11 Assertion failure in journal_commit_transaction()

[Jan Kara]

  Hello,

> On a busy server (dual Xeon with 4GB ram) running plain 2.6.11.11 form
> kernel.org (distribution is Fedora Core 2) I got the following error:
> 
> Jun  4 11:40:55 apollo kernel: Assertion failure in 
> journal_commit_transaction() at fs/jbd/commit.c:768: 
> "jh->b_next_transaction == ((void *)0)"
> Jun  4 11:40:55 apollo kernel: ------------[ cut here ]------------
> Jun  4 11:40:55 apollo kernel: kernel BUG at fs/jbd/commit.c:768!
> Jun  4 11:40:55 apollo kernel: invalid operand: 0000 [#1]
> Jun  4 11:45:07 apollo syslogd 1.4.1: restart.
> 
> The system bootet automatically, from the ipmi motherboard logs I can see
> that the hardware watchdog has reset the system at 11:42:55.
> It's using an ext3 on top of a software raid10 (three raid1 combined with
> raid0 over 6 SCSI disks). None of the software raids needed to rebuild after
> the crash.
> 
> Any idea why the kernel stopped? The system has been running stable for a
> year now.
  The kernel stopped because it detected a disk buffer in an unexpected
state. 2.6.12-rc5 kernel should contain some more fixes than 2.6.11.11
for similar problems so you can try that kernel. If you are able to see
the same problem with 2.6.12-rc5 then let us know please.

						Thanks for report
								Honza

-- 
Jan Kara <jack@suse.cz>
SuSE CR Labs

Re: 2.6.11.11 Assertion failure in journal_commit_transaction()

[Chris Wright]

* Jan Kara (jack@suse.cz) wrote:
>   The kernel stopped because it detected a disk buffer in an unexpected
> state. 2.6.12-rc5 kernel should contain some more fixes than 2.6.11.11
> for similar problems so you can try that kernel. If you are able to see
> the same problem with 2.6.12-rc5 then let us know please.

Do you feel any of those fixes are stable enough for -stable?
Especially this one looks ok:

 Subject: [PATCH] Fix log_do_checkpoint() assertion failure
 Message-ID: <20050601074059.GD5933@atrey.karlin.mff.cuni.cz>

thanks,
-chris

Re: 2.6.11.11 Assertion failure in journal_commit_transaction()

[Jan Kara]

> * Jan Kara (jack@suse.cz) wrote:
> >   The kernel stopped because it detected a disk buffer in an unexpected
> > state. 2.6.12-rc5 kernel should contain some more fixes than 2.6.11.11
> > for similar problems so you can try that kernel. If you are able to see
> > the same problem with 2.6.12-rc5 then let us know please.
> 
> Do you feel any of those fixes are stable enough for -stable?
> Especially this one looks ok:
> 
>  Subject: [PATCH] Fix log_do_checkpoint() assertion failure
>  Message-ID: <20050601074059.GD5933@atrey.karlin.mff.cuni.cz>
   Yes, that patch should be safe and fixes a really observed bug so I
guess it qualifies into -stable :)

								Honza

-- 
Jan Kara <jack@suse.cz>
SuSE CR Labs

Re: 2.6.11.11 Assertion failure in journal_commit_transaction()

[Chris Wright]

* Jan Kara (jack@suse.cz) wrote:
> > * Jan Kara (jack@suse.cz) wrote:
> > >   The kernel stopped because it detected a disk buffer in an unexpected
> > > state. 2.6.12-rc5 kernel should contain some more fixes than 2.6.11.11
> > > for similar problems so you can try that kernel. If you are able to see
> > > the same problem with 2.6.12-rc5 then let us know please.
> > 
> > Do you feel any of those fixes are stable enough for -stable?
> > Especially this one looks ok:
> > 
> >  Subject: [PATCH] Fix log_do_checkpoint() assertion failure
> >  Message-ID: <20050601074059.GD5933@atrey.karlin.mff.cuni.cz>
>    Yes, that patch should be safe and fixes a really observed bug so I
> guess it qualifies into -stable :)
> 

Thanks, I'll add to the -stable queue.
-chris