Re: amazingly, debian server keeps crashing
On Monday 18 April 2005 21:57, will trillich wrote:
> we're having trouble finding evidence anywhere
> that shows what's going south--
>
> from "kern.log":
> Apr 12 19:53:11 server kernel: Unable to handle kernel NULL pointer
dereference at virtual address 0000008c
> Apr 12 19:53:11 server kernel: PREEMPT
> Apr 12 19:53:11 server kernel: EIP is at journal_blocks_per_page+0x9/0x20
> Apr 12 19:53:11 server kernel: Process exim3 (pid: 18709,
> [__crc_xfrm_state_alloc+4281735/4557196] ext3_prepare_write+0x1a/0x140
> [do_anonymous_page+312/432] do_anonymous_page+0x138/0x1b0 Apr 12 19:53:11
> [__crc_xfrm_state_alloc+4270833/4557196] ext3_file_write+0x44/0xd0 [ext3]
> Apr 12 19:53:11 server kernel: [do_sync_write+128/176]
> should we be using kernel 2.4? other pointers welcome...
I'm no expert (at all) on kernel debugging, but from this backtrace, it looks
like something in your ext3 filesystem's journalling. You can easily disable
ext3 and use ext2, so that might be a good start to narrowing down the
problem. Bear in mind that ext2 will be more vulnerable to corruption
though, so don't leave it enabled for too long if it's not fixing the
problem, and do weigh the pros and cons before trying it :)
Do you preemption enabled? You might want to try without that, given the
PREEMPT line above. But maybe it always says that on kernels with preempt.
Anyway, I'm sure someone more experienced will have some help for you. Good
luck with it -- unstable machines can be a real pain.
--
Lee.
Reply to: