general protection fault and howto capture core dump

[Artyom Nosov]

Adjust your sysctl variable fs.suid_dumpable to 1:
/sbin/sysctl fs.suid_dumpable = 1

and start memcached with -r flag

ps: we solved the same problem by using memcached-1.1.x

Miguel DeAvila wrote:
> We're having fairly regular outages in our memcache cluster ...
> the memcached process is exiting abnormally.
> 
> The only log entry we can see is via dmesg,
> 
>         memcached[3020] general protection rip:407ed9 rsp:7fff7cdeb238 error:0
> 
> Anyone seen this?
> 
> We also have not had any luck getting a core dump. We're running
> CentOS release 5 (Final).
> 
> I think we've taken the necessary steps to encourage a core dump,
> 
> [root at memcache04 ~]# cat /proc/sys/kernel/core_pattern
> core
> 
> [root at memcache04 ~]# cat /proc/sys/kernel/core_uses_pid
> 1
> 
> My one suspicion is that the current working dir for the memcached process
> is /root, but the process is running as the www user, so it does not have write
> permission.
> 
> Any tips?
> 
> thanks,
> 
> Miguel


-- 
Sincerely,

Artyom Nosov
chip at unixstyle.ru | JID chip at xmpp.ru | ICQ 128417264