[CentOS] stack overflow

Thu Jan 15 10:17:08 UTC 2009
John Doe <jdmls at yahoo.com>

----- Original Message ----

> From: Peter Doherty <doherty at crystal.harvard.edu>
> To: centos at centos.org
> Sent: Wednesday, January 14, 2009 9:49:40 PM
> Subject: [CentOS] stack overflow
> 
> Hi,
> 
> I've got a fileserver that runs Centos 5.2.  It's been stable  
> otherwise stable for maybe a year or more, and now it's crashed three  
> times since Saturday.  The first two times the computer was completely  
> unresponsive, and there was nothing on the console, and nothing in the  
> logs.  I was beginning to suspect hardware, (esp. RAM, PSU, or maybe a  
> failed fan)
> Last night it locked up again, but this time there was something in / 
> var/log/messages. (see below)
> So far my searches indicate that increasing to 8K kernel stacks would  
> fix this.  The server has a couple 3ware SATA RAID cards, and I'm  
> running xfs.  The server does nightly disk to disk backups of a few  
> dozen workstations.  The server does always seem to crash overnight.  
> If the problem really is in the kernel stack size, it's odd that it  
> just started all of a sudden.
> I thought I'd post here before I looked into compiling a new kernel  
> with 8K stacks.  Thanks for any advice!

Did you check in your server hardware logs (if it has some; like IPMI?)?

JD