OGS 62u5p2 is a bug fix only release, and I don't recall Ron Chen or myself changing the default limits. Can you run a job that prints the resource limits (bash/sh - ulimit -a, tcsh/csh - limit) on the old & new cluster??
Also, if you still have the SGE 6.2u5 cluster running, then can you temporarily replace the execd/shepherd on an execution host and see if the limit is still set to 3GB?? If it is, then it really is a new bug; otherwise, it is caused by your settings in the new cluster... Rayson On Fri, Dec 2, 2011 at 12:21 PM, CB <[email protected]> wrote: > I found that RLIMIT_VMEM was set to 3GB from a trace file. This happens > with OGS 62u5p2 and UGE 800p1 source builds. > > 12/02/2011 03:11:11 [400:3254]: setting limits > 12/02/2011 03:11:11 [400:3254]: RLIMIT_CPU setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_FSIZE setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_DATA setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_STACK setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_CORE setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_VMEM/RLIMIT_AS setting: (soft > 3221225472 hard 3221225472) resulting: (soft 3221225472 hard 3221225472) > 12/02/2011 03:11:11 [400:3254]: RLIMIT_RSS setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > > BTW, GE 6.2u5 shows that it didn't change RLIMIT_VMEM as shown below: > 12/02/2011 11:23:18 [400:9290]: setting limits > 12/02/2011 11:23:18 [400:9290]: RLIMIT_CPU setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_FSIZE setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_DATA setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_STACK setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_CORE setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_VMEM/RLIMIT_AS setting: (soft > INFINITY hard INFINITY) resulting: (soft INFINITY hard INFINITY) > 12/02/2011 11:23:18 [400:9290]: RLIMIT_RSS setting: (soft INFINITY hard > INFINITY) resulting: (soft INFINITY hard INFINITY) > > Is this a known bug or a feature? > > Thanks, > - Chansup > > > _______________________________________________ > users mailing list > [email protected] > https://gridengine.org/mailman/listinfo/users > _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
