- 18 May, 2016 7 commits
-
-
Morris Jette authored
-
Morris Jette authored
Remove some SQUEUE output format environment variables set by default on Cray systems, which break some tests
-
Morris Jette authored
AND should have been an OR. Modify test to log more details on failure.
-
Brian Christiansen authored
Group id was being overwritten by user id.
-
Brian Christiansen authored
Writing and reading to a file on a slow shared filesytem could cause this test to fail.
-
Brian Christiansen authored
-
Brian Christiansen authored
The filesystem may have problems removing and creating the same filename quickly.
-
- 17 May, 2016 15 commits
-
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
Correct description of the SLURMD_NODENAME environment variable in the sbatch and srun man pages.
-
Morris Jette authored
Also add some job time limits
-
Morris Jette authored
Increase a sleep for slower step start times on a Cray Make error message more detailed (add expected/actual count to message)
-
Morris Jette authored
One memory leak fixed Some other code moved around for better clarity
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
Run autogen.sh to pickup changes and resolve conflicts.
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Morris Jette authored
Previous logic could truncate CPU masks at 32-bits
-
Morris Jette authored
(1 << x) where "x" is a uint64_t would treat "1" as an int and roll over at 32-bits. Typecasting "1" to uint64_t eliminates that and supports a full 64-bit value.
-
- 16 May, 2016 5 commits
-
-
Josko Plazonic authored
Update slurm.spec file to have seff depend on slurm-perlapi.
-
Tim Wickberg authored
Conflicts: src/plugins/switch/generic/switch_generic.c src/slurmd/slurmstepd/task.c
-
Tim Wickberg authored
-
Jason Bacon authored
-
Morris Jette authored
-
- 13 May, 2016 13 commits
-
-
Giovanni Torres authored
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
Make test more robust for compute nodes with large CPU counts.
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
This change fixes Slurm's ability to optimize selection of resources for a job requesting feature counts where some of those node features are currently inactive (require node reboot to claim).
-
Danny Auble authored
-
Danny Auble authored
-
Danny Auble authored
# Conflicts: # src/common/slurm_acct_gather_energy.c # src/common/slurm_acct_gather_filesystem.c # src/common/slurm_acct_gather_infiniband.c # src/common/slurm_acct_gather_profile.c # src/common/slurm_jobacct_gather.c
-
Danny Auble authored
when in use. The problem here is the polling threads in the various acct_gather codes were detached and could possibly still be polling after the plugin had been unloaded making a seg fault with a backtrace like this... #0 0x00007fe7af008c00 in ?? () #1 0x00007fe7b1138479 in __nptl_deallocate_tsd () at pthread_create.c:175 #2 0x00007fe7b11398b0 in __nptl_deallocate_tsd () at pthread_create.c:326 #3 start_thread (arg=0x7fe7b1f12700) at pthread_create.c:346 #4 0x00007fe7b0e6fb5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:109 The fix was to make the threads non-detached and join them before calling a dlclose.
-