This is a kernel problem, *not* a gtar problem. Gtar is trying to do an
lstat() system call, and the kernel has apparently run out of resources
-- unless the disk hasn't been fsck'ed, in which case, there may well be
a block that is corrupted (and it probably contains inodes). Be
assured, if /var/log/messages gives you a "kernel:" message, then the
problem is either in the kernel itself, or possibly in the hardware
(i.e., you could have an uncorrected memory error (do not *ever* use
non-ECC memory on a machine you want to be even remotely reliable!)).
Recommendation: fsck the partition that gtar was working on. Run a
serious memory checker (wish I had one, but then, I only use ECC
memory).
Good luck,
Marty
[EMAIL PROTECTED] wrote:
>
> Hello everyone. On one machine, and one machine only, the backup process
> is crashing the server. In particular, it appears to be gtar that is
> making it hang. This machine is configured similarly to every other
> machine that I'm backing up, but this is the only one having problems.
>
> Here is the error that shows up in /var/log/messages:
>
> Oct 24 00:50:08 banjo kernel: Unable to handle kernel paging request at
> virtual address 0004a2bd
> Oct 24 00:50:08 banjo kernel: current->tss.cr3 = 133c0000, %%cr3 =
> 133c0000
> Oct 24 00:50:08 banjo kernel: *pde = 00000000
> Oct 24 00:50:08 banjo kernel: Oops: 0000
> Oct 24 00:50:08 banjo kernel: CPU: 0
> Oct 24 00:50:08 banjo kernel: EIP: 0010:[dput+295/328]
> Oct 24 00:50:08 banjo kernel: EFLAGS: 00010296
> Oct 24 00:50:08 banjo kernel: eax: 0004a27d ebx: dc027120
> ecx: c5616fa0 edx: d6794f80
> Oct 24 00:50:08 banjo kernel: esi: ffffffff edi: cf5a8330
> ebp: 00000568 esp: dcce3e7c
> Oct 24 00:50:08 banjo kernel: ds: 0018 es: 0018 ss: 0018
> Oct 24 00:50:08 banjo kernel: Process gtar (pid: 7926, process nr: 53,
> stackpage=dcce3000)
> Oct 24 00:50:08 banjo kernel: Stack: c56165a0 c012e758 dc027120 dcce3ed0
> dcce3ed0 c023715c 00001006 dcce3ed0
> Oct 24 00:50:08 banjo kernel: 00000001 00001006 c012f6d3 fffff562
> 00001006 00000000 c0268f60 c023715c
> Oct 24 00:50:08 banjo kernel: c0268f60 cf6598c0 dd26d320 dd26d36c
> 00000000 dcce3ed0 dcce3ed0 c012f73a
> Oct 24 00:50:08 banjo kernel: Call Trace: [prune_dcache+248/300]
> [try_to_free_inodes+199/264] [grow_inodes+30/384] [get_new_inode+18
> 5/280] [iget+88/96] [ext2_lookup+84/124] [real_lookup+79/160]
> Oct 24 00:50:08 banjo kernel: [lookup_dentry+296/488]
> [__namei+40/88] [sys_newlstat+14/96] [system_call+52/56] [startup_32+43
> /286]
> Oct 24 00:50:08 banjo kernel: Code: 8b 40 40 50 56 68 60 41 1f c0 e8 ee 49
> fe ff c7 05 00 00 00
>
> After this, the machine is fully hung, and I have to reboot it. I'm
> thinking maybe gtar is hosing the stack somehow.
>
> Any ideas? Or do I need to send this to the tar bug list? Here's the
> stats on the server itself:
>
> 400MHz PII, 512MB RAM, RAID-5 SCSI with a DPT Decade card. Kernel
> 2.2.16-3, Amanda 2.4.1p1, gtar 1.12.
>
> Thanks for any help!
>
> -D
--
Marty Shannon, RHCE, Sr. Systems Developer
mailto:[EMAIL PROTECTED]
