Shared memory table

[Yong Huang]

Rohit khaladkar wrote:
> Hi!The Shared memory table shmmni is reaching 100% in the middle of
> the day
> - at random .
>
> I do not know what process takes it to that level .
>
> What would be my way of finding that out ?
>
> Thanks for all the help!
>
> Rohit

As Kevin Krieser suggested, you can run "ipcs -m -i <ID>" or check /proc/sysvipc/shm. There's cpid (creator's pid) and nattch (how many processes attaching to each segment). If the creator still exists, you can find it in ps output. If not, there's really no way I know of to find it. Perhaps you can schedule a cron job to frequently check the cpid in ps output in the middle of the day.

If nattch is 0, a shared memory segment is not cleaned up properly and most likely you can safely remove it by ipcrm.

Yong Huang