- 11 Mar, 2015 3 commits
-
-
Morris Jette authored
-
Morris Jette authored
Add new configuration parameter "set_watts" to explicitly set every node's power cap to the same value. Improve accuracy of power cap data logging. Fix logic for changing configuration values with SIGHUP or "scontrol reconfig" command. Correct power distribution calculation.
-
Brian Christiansen authored
Restores same behavior prior to ce32018a. Bug 1513
-
- 10 Mar, 2015 3 commits
-
-
Morris Jette authored
A fairly major update to use JSON data strurctures directly to set power caps for better scalability and performance. Added logic to send STDIN to the capmc command (with the JSON data).
-
Morris Jette authored
Relocate where power plugin is told about new job allocation. Previous logic failed to work for job allocated resources at submission time (top priority job does not need to go through the same job scheduling logic as jobs that are already in the queue, and thus failed to notify the power plugin at start time).
-
Morris Jette authored
-
- 06 Mar, 2015 1 commit
-
-
Morris Jette authored
-
- 05 Mar, 2015 5 commits
-
-
Danny Auble authored
-
Danny Auble authored
fails addition to commit 2c83bf4e
-
Danny Auble authored
message comes in
-
David Bigagli authored
-
David Bigagli authored
-
- 04 Mar, 2015 7 commits
-
-
Brian Christiansen authored
Bug 1501
-
Brian Christiansen authored
-
Brian Christiansen authored
Bug 1501
-
David Bigagli authored
-
Danny Auble authored
Conflicts: src/slurmctld/acct_policy.c
-
Danny Auble authored
-
Danny Auble authored
-
- 03 Mar, 2015 8 commits
-
-
Danny Auble authored
cluster(s) requested.
-
David Bigagli authored
-
Morris Jette authored
-
Brian Christiansen authored
Bug 1492
-
Morris Jette authored
-
Morris Jette authored
-
Morris Jette authored
For job running under a debugger, if the exec of the task fails, then cancel its I/O and abort immediately rather than waiting 60 seconds for I/O timeout.
-
Morris Jette authored
The option has not been functional or documented since Slurm version 2.0.
-
- 02 Mar, 2015 4 commits
-
-
David Bigagli authored
-
David Bigagli authored
-
Danny Auble authored
-
Danny Auble authored
-
- 27 Feb, 2015 9 commits
-
-
Morris Jette authored
This controls how long a requeued job must wait before it can restart, and 20 minutes is too long in most cases. Administrators can alter this configuration parameter if needed in case of slow Prolog or the like.
-
Morris Jette authored
Conflicts: src/slurmctld/job_mgr.c
-
Nicolas Joly authored
Add missing arguments to slurm_sched_p_newalloc/slurm_sched_p_freealloc documentation.
-
Nicolas Joly authored
-
Nicolas Joly authored
-
Morris Jette authored
-
Morris Jette authored
Use this to specify the lifetime of a job step credential.
-
Brian Christiansen authored
Bug 1476
-
Morris Jette authored
Set the delay time for job requeue to the job credential lifetime (1200 second by default). This insures that prolog runs on every node when a job is requeued. (This change will slow down launch of re-queued jobs). Without this change, if a job is restated within 1200 seconds, the nodes previously used would not run the prolog again, since the job ID is still seen as active (from the previous execution). It is also advisable to set the value of DEFAULT_EXPIRATION_WINDOW in src/common/slurm_cred.c to the lowest value reasonable. We need to add a new configuration parameter so this is easly changed in the future.
-