- 09 Jun, 2018 3 commits
-
-
Morris Jette authored
Also includes some changes to gres.c and job_mgr.c needed to support all gres types
-
Morris Jette authored
Bug introduced in commit 5fb544f2
-
Morris Jette authored
This should have been part of commit 07921f8a Detected by test4.11
-
- 08 Jun, 2018 3 commits
-
-
Marshall Garey authored
test3.1 - verify that the user running the test is an Admin so that reconfigure will work. If not, throw a warning and quit to prevent the reconfigure macro from erroring out instead. Bug 4926.
-
Tim Wickberg authored
-
Morris Jette authored
This is in anticipation of an upcoming change to the cgroup hierarchy on a future CLE release. Bug 5145.
-
- 07 Jun, 2018 34 commits
-
-
Brian Christiansen authored
-
Brian Christiansen authored
If defined, is called when a node failes to resume by ResumeTimeout.
-
Marcin Stolarek authored
Bug 5275.
-
Brian Christiansen authored
Missed from 7796c940
-
Alejandro Sanchez authored
Bug 3792.
-
Alejandro Sanchez authored
Bug 3792.
-
Tim Wickberg authored
-
Alejandro Sanchez authored
Add new page, add to Makefile.am and run autogen.sh. Update additional documentation to point at this new page. Bug 3792.
-
Tim Wickberg authored
-
Tim Wickberg authored
Always zero on non-BG systems, so remove entirely.
-
Tim Wickberg authored
Also start removing BG code, and take out err_cpus which is always zero on non-BG.
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
Unused, but missed during prior clean up since it is split across files.
-
Tim Wickberg authored
Both are always one. (On non-BG.)
-
Tim Wickberg authored
-
Tim Wickberg authored
Always one (non-BG), so just change these to the increment operator.
-
Tim Wickberg authored
-
Tim Wickberg authored
SELECT_NODEDATA_SUBGRP_SIZE is always zero on non-BG, so all calling paths into this can be removed.
-
Tim Wickberg authored
-
Tim Wickberg authored
Removing the CLUSTER_FLAG_BG block forces some code up a tabspot, making this look a bit uglier than indended. Also, remove the error_cpus values, as those only applied to BG.
-
Tim Wickberg authored
NODE_STATE_ERROR is only valid on a BlueGene, so make this a no-op for now.
-
Tim Wickberg authored
As done previously, this can be removed as it always returns zero: select_g_select_nodeinfo_get(node_ptr->select_nodeinfo, SELECT_NODEDATA_SUBCNT, NODE_STATE_ERROR, &err_cpus); Simplify the resulting code, and remove CLUSTER_FLAG_BG code as well.
-
Tim Wickberg authored
No functional changes, but do cleanup some style in these updated blocks while copying.
-
Tim Wickberg authored
No changes yet.
-
Tim Wickberg authored
Helps avoid pointer soup like &((*msg)->node_scaling). No functional change.
-
Tim Wickberg authored
Mention in RELEASE_NOTES in the appropriate sections. Tidy up function declarations and formatting while here.
-
Tim Wickberg authored
This is always 0 on non-select/bluegene. Remove other BlueGene-specific output from this block while here. This is all leading up to removal of the node_scaling concept from Slurm's internals.
-
Tim Wickberg authored
-
Tim Wickberg authored
Make it easier to use with designated initializers in the future.
-
Brian Christiansen authored
If an array task exited with a RequeueExit* code, the array job wasn't being marked as requeued. Bug 5105
-
Isaac Hartung authored
if any task in the array was requeued. This is a hint to use "sacct --duplicates" to see the whole picture of the array job. Bug 5105
-