The number of file descriptors is not part of the queue limit, see the message I sent to the list 2 months ago:
http://gridengine.org/pipermail/users/2012-May/003705.html If you really have a real use-case for setting the # of descriptors in the queue config, then let us know and we can implement that in OGS/GE (... when time permits). Rayson On Tue, Jul 10, 2012 at 3:32 PM, John Young <[email protected]> wrote: > I have a short test job that I can submit to different > queues on my cluster that appear to be configured the > same, but I get different results. Here is the job: > > ------------------------------------------- > #!/bin/tcsh > # > #$ -N show-limits > #$ -S /bin/tcsh > #$ -o show-limits.out > #$ -e show-limits.err > #$ -m ae > #$ -pe mpich 32 > > /bin/echo I am running on host `hostname`. > /bin/echo Original limits: > limit > /bin/echo > /bin/echo limit descriptors 65535 > limit descriptors 65535 > /bin/echo > /bin/echo New limits: > limit > ------------------------------------------- > > If I submit the job to all.q, it works and I get output > like this: > > I am running on host compute-1-51.local. > Original limits: > cputime unlimited > filesize unlimited > datasize unlimited > stacksize unlimited > coredumpsize 0 kbytes > memoryuse unlimited > vmemoryuse unlimited > descriptors 40960 > memorylocked 32 kbytes > maxproc 81920 > > limit descriptors 65535 > > New limits: > cputime unlimited > filesize unlimited > datasize unlimited > stacksize unlimited > coredumpsize 0 kbytes > memoryuse unlimited > vmemoryuse unlimited > descriptors 65535 > memorylocked 32 kbytes > maxproc 81920 > > If I submit the same job to the queue xyz-nodes, I get output > like this: > > I am running on host compute-0-19.local. > Original limits: > cputime unlimited > filesize unlimited > datasize unlimited > stacksize unlimited > coredumpsize 0 kbytes > memoryuse unlimited > vmemoryuse unlimited > descriptors 1024 > memorylocked 32 kbytes > maxproc 524288 > > limit descriptors 65535 > limit: descriptors: Can't set limit > > Notice that the default descriptors limit is different, and > under all.q, I can change it, but under xyz-nodes, I can't. > > I have carefully compared the queue configurations with > 'qconf -sc all.q' and 'qconf -sc xyz-nodes' and I do not see any > difference, so I don't understand why I get different behavior > from the two queues. > > Any ideas? > > JY > > -- > > JY > ------------------------------------------------------------ > John E. Young NASA LaRC B1148/R226 > Analytical Services and Materials, Inc. (757) 864-8659 > > _______________________________________________ > users mailing list > [email protected] > https://gridengine.org/mailman/listinfo/users _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
