- 06 Dec, 2012 6 commits
-
-
Morris Jette authored
Previously the job would be accepted at job submit time if nodes required for its use (e.g. in a required node list or the node count) could not be satisfied at submit time due to an advanced reservation. Bug #187
-
Morris Jette authored
-
Danny Auble authored
-
Danny Auble authored
-
Danny Auble authored
-
Morris Jette authored
-
- 05 Dec, 2012 12 commits
-
-
Danny Auble authored
-
Nathan Yee authored
and is also enforced correctly
-
Yiannis Georgiou authored
1) updated man pages to better explain how the acct_gather_energy plugins give accurate results only in case of exclusive job allocations 2) remove unnecessary defines
-
Morris Jette authored
Split long lines, add brackets, indentation
-
Yiannis Georgiou authored
1)updated scontrol show node to report "n/s" (not supported) for energy/power values in case acct_gather_energy/rapl is configured and nodes do not support RAPL Previously, these values were reported as zero. The scontrol man page has been also updated to explain this. 2)updated sstat / sacct to report no value for ConsumedEnergy in case acct_gather_energy/rapl is configured, a job/step is allocated and at least one node does not support RAPL. Previously for this case, sstat and sacct reported a partial value for ConsumedEnergy. 3)updated the logging mechanism so that If acct_gather_energy/rapl is configured and a job/step is allocated any nodes that do not support RAPL, a single debug entry is made in the log file. Previously, multiple repeating error messages were sent to the console and the log file.
-
Danny Auble authored
-
Danny Auble authored
job on future step creation attempts.
-
Danny Auble authored
-
Danny Auble authored
bgsched::runjob::Terminated::Nodes::software_error_nodes() It shouldn't really affect the code. It only printed out cnodes, which most likely never really did anything.
-
Danny Auble authored
also cause it to run if the realtime server ever goes away.
-
Morris Jette authored
-
Morris Jette authored
Especially for newly started jobs, the PrologSlurmctld can change a job's QOS based upon resource allocation.
-
- 04 Dec, 2012 9 commits
-
-
Danny Auble authored
-
Danny Auble authored
DB2 so hard.
-
Danny Auble authored
-
Danny Auble authored
-
jette authored
Conflicts: src/slurmctld/job_scheduler.c
-
jette authored
-
jette authored
-
Morris Jette authored
Problems with jobs starting on reserved nodes
-
Carles Fenoy authored
Added check for preeption mode before changing time limit to 1 if flag no reserve is enabled, as this can cause a problem with jobs starting on nodes with future reservations.
-
- 03 Dec, 2012 6 commits
-
-
https://github.com/SchedMD/slurmjette authored
-
jette authored
-
Morris Jette authored
-
Morris Jette authored
-
jette authored
-
jette authored
This bug was introduced on Friday and fails to address an environment variable not being set
-
- 01 Dec, 2012 4 commits
-
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
-
- 30 Nov, 2012 3 commits
-
-
https://github.com/SchedMD/slurmjette authored
-
jette authored
-
Danny Auble authored
-