Checking which kernel limit is breached

[Dave Hewetson]

How can I dynamically check global counts to determine which kernel tunable setting is being breached? I'm running a c program which does a tp-connect into a remote Tuxedo service, in order to simulate 25,000 Tuxedo connections. Each tp-connect runs in a Unix process, and we are able to start 16,348 before the next one fails. The failure is on the local system, where the only values in the region of 16348 are msgseg and semaem - both are 16384, but neither seems relevant to the problem.

[James R. Ferguson]

Hi Dave:

The system table report of 'glance' [toggled with a 't'] is useful for a few kernel parameters. 'sar -v' is valuable for several too.

You might find it useful to list the values and limits of your tuned set with:

# kmtune -l

Additionally, it is always useful to consult '/usr/include/sys/errno.h' for the meaning of 'errno'.

Regards!

...JRF...

[U.SivaKumar_2]

Hi,

#sar -v 1 10

This will monitor
6:53 text-sz ov proc-sz ov inod-sz ov file-sz ov


The related kernel paramters are
1) nproc
2) ninode
3) nfiles

regards,
U.SivaKumar