- 29 Jan, 2019 4 commits
-
-
Tim Wickberg authored
Private, does not need to be exposed externally.
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
Introduced in 360fb080. I don't believe this would solve the underlying issue on a system that is OOM, and there is another xmalloc call coming up immediately anyways that would then fatal().
-
- 28 Jan, 2019 4 commits
-
-
Danny Auble authored
# Conflicts: # contribs/lua/proctrack.lua # src/common/gres.c # testsuite/expect/README # testsuite/expect/globals
-
Nate Rini authored
bug6301
-
Nate Rini authored
Bug 6279
-
Nate Rini authored
Bug 6214
-
- 27 Jan, 2019 1 commit
-
-
Morris Jette authored
explicity typecast some size_t variables to long int to avoid compiler warnings
-
- 26 Jan, 2019 4 commits
-
-
Brian Christiansen authored
-
Marshall Garey authored
Bug 5742
-
Marshall Garey authored
Bug 5742
-
Marshall Garey authored
Bug 5742
-
- 25 Jan, 2019 11 commits
-
-
Morris Jette authored
The information will be logged for all tasks of a job array with a task count limit. bug 5093
-
Moe Jette authored
Bug 6398
-
Michael Hinton authored
On a heterogeneous system the gres.conf coming from various nodes might not be complete. The slurm.conf has a complete list of possible gres on the other hand. If the GRES doesn't exist we will print an error and use the id instead of the name, but this should never happen. Bug 6366
-
Felip Moll authored
When trying to update the partition list of a job it didn't get updated. Bug 6329
-
Morris Jette authored
bug 5693, 6397
-
Morris Jette authored
bugs 5693, 6397
-
Morris Jette authored
put entire description in one line of README file (for logging add missing exp_continue remove unused variable
-
Morris Jette authored
-
Morris Jette authored
Fix for tests that will fail with SelectTypeParmaters=CR_Socket (assuming each socket has more than one core).
-
Morris Jette authored
We want to avoid printing "FAILURE" in the function wait_for_job as a new use case may result in a job not starting in a timely fashion and NOT be an error. So change "FAILURE" in wait_for_job to "WARNING" and add checks for function errors in the tests as needed (most places already check and log errors). There were also many cases where "FAILURE would be printed by wait_for_job, but the job would not have a non-zero exit code and those are now fixed.
-
Morris Jette authored
Make sure that the count of CPUs allocated to a job is appropriate for the task count. bug 6274
-
- 24 Jan, 2019 1 commit
-
-
Marshall Garey authored
pam_slurm_adopt so special users can ssh to a node. This is an alternative to pam_access.so. Bug 6243
-
- 23 Jan, 2019 13 commits
-
-
Danny Auble authored
-
Dominik Bartkiewicz authored
Remove unneeded scontrol call.
-
Danny Auble authored
Bug 6357
-
Danny Auble authored
-
Dominik Bartkiewicz authored
Bug 6357
-
Dominik Bartkiewicz authored
Bug 6357
-
Jakub Yaghob authored
Bug 6320
-
Danny Auble authored
instead of first as previously done.
-
Danny Auble authored
Bug 6004
-
Dominik Bartkiewicz authored
specific tasks. Bug 6357
-
Josko Plazonic authored
Bug 6004
-
Paddy Doyle authored
As noted by Josko Plazonic in the bug report, the tres_usage_in_max value is bytes, whereas the script expects kb. Bug 6004
-
Paddy Doyle authored
Update seff to reflect API change from rss_max to tres_usage_in_max. Use Slurmdb::find_tres_count_in_string to parse out the TRES_MEM value. Bug 6004
-
- 22 Jan, 2019 1 commit
-
-
Morris Jette authored
This fix is needed to address the GRES specification in gres.conf having a Type option, while the GRES specification in slurm.conf does not.
-
- 21 Jan, 2019 1 commit
-
-
Morris Jette authored
If scontrol is used to change a node's GRES and the input string contains socket binding information (e.g. "gres=gpu:4(S:0),...") then ignore the socket binding information input. Use the binding as reported by slurmd in the node registration.
-