| 1:48:59 PM PDT - Tue, Oct 8th 2013 |
|
Quote:Tpirojsi Oct 8th 6:28 pmQuote:Edoapra Oct 8th 4:46 pmTee
qstat is not checking the running processes on the compute nodes.
In order to check the status of running processes (and for the ipcs output as well), you need to login to the compute nodes (using ssh, for example)
Oh! Thank you for shedding some light on this for me. You are absolutely correct. I logged in to computer nodes, and punched in the 'ipcs -a' command and did see some shared memory segments showed up. I used the tools provided in nwchem package to clean them up and saw the swap space came back to normal. It seems to solve qrsh_starter problem too!
I really appreciate your help.
Tee
Indeed, I still see some qrsh_starter errors but very few. Do you have any idea what is it related to?
|