SUMMARY: Any Way To Avoid Too Many Processes Hanging A Server?

Many of you suggested giving a size paramater in /tmp mount line in
/etc/vfstab. While this may reduce a DOS attach by a user/process gobbling
up space in VM this does not address the problem of VM (swap and memory)
being overrun by wayward zillions of processes. Many of you also suggested
enabling a system parameter limiting user processes - maxuprc which I will
try to implement after a careful study of what value could be optimal. But I
guess the best way would be to have a sentinel job that watches periodically
the number of running processes. This could be done using a variety of
techniques - custom C program, shell script or PERL code to monitor, alert
and possibly kill offending source/parent processes...

Thanks all.

NELSON


>  -----Original Message-----
> On many occasions already, we've had servers hanging because of "/tmp:
> File system full, swap space limit exceeded" conditions which is mostly
> due to wayward processes spawning so many processes. Sometimes, an
> ill-written script that spawns itself infinitely. The only way to get out
> of these situations is a power-off and a hard boot of the server.
> 
> My Question:  Are there any best practices out there that would prevent
> the server from hanging or "bullet-proof" the system from such scenarios
> so that a SysAdmin can still get in? Would Solaris Resource Manager help?
> 
> Thanks.
> 
> 
> NELSON
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers