Previous logic never reopened the file, preventing proper functioning
of logrotate.

Use hash function to locate job records for improved performance.

Change partition write lock to a read lock as we use a different
mechanism for hidden partitions in getting individual jobs.

Correct logic returning remaining job dependencies in job information
reported by scontrol and squeue. Eliminates vestigial descriptors with
no job ID values (e.g. "afterany"). As depdencies are removed, the
job ID values were removed from the strings, but not the descriptors.
This eliminates both. It also checks the full job ID to make sure we do
not remove "afterany:1234" when job "123" completes.

Fix race condition on batch job termination that could result in a job exit
code of 0xfffffffe if the slurmd on node zero registers its active jobs at
the same time that slurmstepd is recording the job's exit code.
but 535

proctrack/cgroup - Add locking to prevent race condition where one job step
is ending for a user or job at the same time another job stepsis starting
and the user or job container is deleted from under the starting job step.
bug 447

Substantial performance improvement for systems with Shared=YES or FORCE
and large numbers of running jobs (replace bubble sort with quick sort).
Bug 525

Original code worked only for Cray systems. For other systems it
set gres_alloc to the total number of each GRES allocated on each
node to any job

bug 511

This might have worked fine for core reservations or when there
are sufficient idle nodes to use, the the select_g_resv_test()
function clears the node bitmap for nodes that it can not use
and the reservation create logic did not restore that bitmap
after a failed resource selection attempt. This logic restores
the node bitmap on a failed call to select_g_resv_test() so we
can add nodes to the bitmap of available nodes rather than having
it repeatedly cleared.
The logic also adds some performance enhancements that I will
add to in the next commit.

Correction to hostlist parsing bug introduced in v2.6.4 for hostlists with
more than one numeric range in brackets (e.g. rack[0-3]_blade[0-63]").
bug505

Add argument to priority plugin's priority_p_reconfig function to note
when the association and QOS used_cpu_run_secs field has been reset.
Without this flag, we remove time on "scontrol setdebug" or
"scontrol setdebugflag" that can result in used_cpu_run_secs
going negative or otherwise get bad values.
Correction to logic added in commit 6d793189
bug 423

Fix to work with change logic introduced in Slurm version 2.6.3
scheduling logic which prevented Maui/Moab from starting jobs.

Add support for -W block=true	(wait for job completion)
Clear PBS_NODEFILE environment variable
Credit to NCSC

Add support for dependencies of job array elements (e.g.
"sbatch --depend=afterok:123_4 ..."). This does not support
depenendencies of ALL job array elements, only individual job
array elements.

preforms the allocation, regression in 2.6.3.

Previously the SlurmdDebug value in slurm.conf was ignored if the
previous value was not 3/init

This avoids building hostlist information with NodeHostName and
NodeAddr information unless explisitly requested and can improve
performance for the default mode of operation by about 65%.

Correct sbatch documentation and job_submit/pbs plugin "%j" is job ID,
not "%J" (which is job_id.step_id).

Without this change a job with a reason of WAIT_PART_DOWN,
WAIT_PART_INACTIVE, WAIT_PART_NODE_LIMIT, WAIT_PART_TIME_LIMIT, or
WAIT_QOS_THRES would not be cleared when that reason no longer
applied.

Add cgroup create retry logic in case one step is starting at the
same time as another step is ending and the logic to create
and delete cgroups overlaps.
bug 447

If a node has GRES and multiple threads per core the select/cons_res
plugin can get stuck in an infinite loop.
See bug 475
Contributed by:
PREVOST Ludovic
NEC HPC Europe

Previously a node failure would always requeue the job

Restore default behavior of allocating cores to jobs on a cyclic basis
across the sockets unless SelectTypeParameters=CR_CORE_DEFAULT_DIST_BLOCK
or user specifies other distribution options.
Reverts commit 7fcdc7e5
bug 466

If the backfill scheduler relinquishes locks and the normal job
scheduler starts a job that the backfill scheduler was actively
working, the backfill scheduler will try to re-schedule that
same job, possibly resulting in an invalid memory reference
or other badness.

Fix for --cpu_bind=map_cpu/mask_cpu/map_ldom/mask_ldom plus
--mem_bind=map_mem/mask_mem options, broken in 2.6.2.
See commit 718382da