List:General Discussion« Previous MessageNext Message »
From:Van Date:April 29 1999 11:07pm
Subject:Re: As I wait to go home to fix my zombie server...
View as plain text  
Vincent Stoessel wrote:
> 
> Ok, I'm stuck at my day job until I can go home and fix
> my apache/mysql/realaudio server om redHat 5.0 , 2.0.36 linux.
> 
> While using a php script to do some updates to the local
> mysql serer, My server began dying: I am still logged into the
> machine via telnet but any command just gives me a segmentation fault.
> Fortunately this box syslogs to a remote server. Below is a block
> of info I'm getting from the machine.
> Did I just run out of memory or is it something more sinister.
> Thisngs I've done within the last 3 weeks
> 1.Hacked the kernal to set NR_OPEN to 1024 recompiled.
> 2. Hacked apache to accept up to 300 coonections in httpd.h
> 3. set mysql max connections to 600 and table cache to 200.
> I have 192 M of ram and I also run a RealAudio basic server off
> this box, it is limited to 40 streams.
> 
> Apr 29 15:10:09  kernel: Unable to load interpreter
> Apr 29 15:10:25  kernel: Unable to handle kernel paging request at virtual address
> 499fc000
> Apr 29 15:10:25  kernel: current->tss.cr3 = 0b4ab000,
> Apr 29 15:10:25  kernel: *pde = 06441067
> Apr 29 15:10:26  kernel: *pte = 00000000
> Apr 29 15:10:26  kernel: Oops: 0002
> Apr 29 15:10:26  kernel: CPU:    0
> Apr 29 15:10:26  kernel: EIP:    0010:[generic_file_read+995/1532]
> Apr 29 15:10:26  kernel: EFLAGS: 00010216
> Apr 29 15:10:26  kernel: eax: 05dd3000   ebx: 00001000   ecx: 00000002   edx:
> 00000000
> Apr 29 15:10:26  kernel: esi: 05dd3ff8   edi: 499fc000   ebp: 00352edc   esp:
> 00ee1f54
> Apr 29 15:10:26  kernel: ds: 0018   es: 002b   fs: 002b   gs: 002b   ss: 0018
> Apr 29 15:10:26  kernel: Process mysqld (pid: 23462, process nr: 354,
> stackpage=00ee1000)
> Apr 29 15:10:26  kernel: Stack: 00000018 03a2d2ec 00000000 00400000 0b68b600 004f6000
> 0050e000
> 00001000
> Apr 29 15:10:26  kernel:        00000000 047df000 00000001 00000001 047df000 004f6000
> 000f6000
> 00000000
> Apr 29 15:10:26  kernel:        00123560 0b68b600 03a2d2ec 499fb008 0030a000 00ee2018
> 49905008
> 000000c5
> Apr 29 15:10:26  kernel: Call Trace: [sys_read+196/236] [system_call+85/124]
> Apr 29 15:10:26  kernel: Code: f3 a5 83 e3 03 89 d9 f3 a4 07 55 e8 6d 34 00 00 8b 5c
> 24 1c
> Apr 29 15:10:26  kernel: Unable to handle kernel paging request at virtual address
> 453ff000
> Apr 29 15:10:26  kernel: current->tss.cr3 = 0b4ab000,
> Apr 29 15:10:26  kernel: *pde = 07fe2067
> Apr 29 15:10:26  kernel: *pte = 00000000
> Apr 29 15:10:26  kernel: Oops: 0002
> 
> Any insights, snide remarks, humor will be
> apreciated as I wait to leave.
> Thanks
> vinny@stripped
Vinnie, you have a hardware problem, most likely.  Seems to me to
indicate a swap file (perhaps the hard-drive used thereby) problem.  You
WILL get segfaults on an out of memory condition with a select that
requires more memory than you have physically available (meaning
phys+swap), but, once the memory is depleted, a happy kernel (i.e.
2.0.36 your current and once I've run into this with, or 2.2.6, in my
case now) will slowly kill memory hungry processes, and, ultimately free
memory to a normal state.  Can't speak for what happened to your data,
however, if it was an insert.  My experience with this indicator is on a
select that resulted in a 200Meg temp record-set (then only 64 phys; 64
swap).
But, the kernel problem is ominous.  It's pointing at a hard-drive
read.  The process was mysqld, since that was the straw that broke the
camel's back.  But, the mysqld was not the culprit, just the process
that caused the memory overload.
Hope that helps.
Regards,
Van
-- 
=========================================================================
Linux rocks!!!   www.dedserius.com
=========================================================================
Thread
As I wait to go home to fix my zombie server...Vincent Stoessel30 Apr
  • Re: As I wait to go home to fix my zombie server...Sasha Pachev30 Apr
  • Re: As I wait to go home to fix my zombie server...Van30 Apr
  • Re: As I wait to go home to fix my zombie server...Vinny Stoessel30 Apr