Ondrej Valousek <[email protected]> writes: > Hi Reuti, > > Ok it looks you have been right - the core limits defined in complex > configuration (qconf -sc) were indeed zero, but core limits defined on > queues were unlimited. > After defining these on queue config, problem went away. > Thanks,
For what it's worth, I recommend checking such things with procenv <http://arc.liv.ac.uk/SGE/howto/troubleshooting.html>. I actually default the core size to a small value so that there's an indication of core dumps without filling the filesystem, whatever the return code reported by the job (not that people take notice): $ qrsh procenv | grep -A6 RLIMIT_CORE RLIMIT_CORE: soft: current: 256 max: 18446744073709551615 hard: current: 18446744073709551615 max: 18446744073709551615 See <https://copr.fedorainfracloud.org/coprs/loveshack/livhpc/> for RHEL-ish procenv packages until it gets into EPEL/Fedora. _______________________________________________ SGE-discuss mailing list [email protected] https://arc.liv.ac.uk/mailman/listinfo/sge-discuss
