Initially I wanted to record somehow the time a job has been in PD state
until it actually starts. To do so, I derived an eligible_time field which
was calculated as start_time - begin_time. This was not correct and a user
of the plugin reported that if the job remained a lot of time in PD state
the field underflowed. A more accurate approach would be calculating the
field as start_time - eligible_time (the actual Slurm eligible_time).
Since this field is not accessible within the log_record function which
receives a job_record struct as parameter, I've decided to just remove this
field from being logged.

Prior logic was getting confused with NUMA containing no cores.
bug 2745

This adds support for burst buffer re-issue of pre-load operation
when slurmctld restarts with job in configuring state. Also copy
NEWS item to v16.05.

Prevent job stuck in configuring state if slurmctld daemon restarted while
    PrologSlurmctld is running.
bugs 2789 and 2810

can work on a BGAS node.

Bug 2806

of continuously printing the message over and over as the problem will
most likely not resolve itself.

Bug 2797

account or wckey.

Fix for tracking job resource allocation when slurmctld is reconfigured
    while Cray Node Health Check (NHC) is running. Previous logic would fail to
    record the job's allocation then perform release operation upon NHC
    completeion, resulting in underflow error messages.
bug 2353

While here, mark options const, and add leading underscore to denote this
as a static function (only called within hostlist.c). Also change strcmp
to xstrcmp.

Commit a6ffef22 changed this function and would alter the input hn,
which led to subsequent calls to the function having wrong prefix
lengths for that hostrange precluding it from matching correctly.

Bug 2558.

and error files.

The buffer to be used for reading the system /proc/*/stat files
is moved from the stack to the heap (i.e. malloc'ed memory) and
initialized to zero, and increased in size from 1k to 4k.
I don't see how this could make any difference, but this anomaly
was reported by valgrind.
bug 2234

The #define of SLURMSTEPD_MEMCHECK in src/slurmd/common/slurmstepd_init.h
must be changed to enable memcheck or valgrind. Also change a #if in
src/slurmd/slurmd/req.c near where you find the "valgrind" references.
Bug 2334 diagnostics

If the QOS includes a time limit, skip checking the partition limit.
The QOS limit is checked separately elsewhere.

'the the' is is a a mistake mistake.

Wrong order of operations results in the return code being 0/1.

Fix for "scontrol -dd show job" with respect to displaying the specific
    CPUs allocated to a job on each node. Prior logic would only display
    the CPU information for the first node in the job allocation.
Bug introduced in commit 0f826c0b
due to misplaced parenthesis

Wrong order of operations results in the return code being 0/1.

If the plugin ever returns an error the variables weren't initialized so
when they were freed they could corrupt memory.

Bug 2790

Rename "in" to "input" in slurm_step_io_fds data structure defined in
    slurm.h. This is needed to avoid breaking Python with by using one of its
    keywords in a Slurm data structure.
bug 2755

Hi, before I never tried to set pmix plugin as the default in the slurm.conf.
Setting it in my new test installation highlighted the subject problem.
bug 2786

Bug 2787

C++ compilation against SPANK would fail otherwise.

>= 15.08 for some reports.

Needed to test ntasks_per_socket against both NO_VAL and INFINITE
for job allocaitons spanning multiple sockets.
bug 2766

Prevents correct error handling by rc being 0/1 instead of the original
return code.

Also fix slurm_send_only_controller_msg and slurm_send_only_node_msg
although these only result in bad printed values in the debug message.

Bug 2120

This bug was introduced by commit 21c52d2f
which fixed a different problem tracking resources associated with suspended
jobs. There are subtle differences between jobs that are suspended by a
user/administrator and jobs suspended by gang scheduling which resulted in
undercounting allocated CPUs when a job suspended by gang scheduling
was active at the same time of a slurmctld reconfiguration request.
See bugs 2353 (original bug related to commit 21c52d2f
and bug 2765

accounts) no default account is printed, previously NULL was printed.

This is just not printing it, but whole function should probably be
revisited as the rigmarole can probably be avoided as we always know what
the default is going to be if none is specified (first off the list).

The problem with that though is if the user has been added to a cluster
already and they have a default, but then added to a new cluster where
they don't have a default.  In this case you want to keep the first
clusters default, but set the default for the second cluster.

Bug 2725

Add missing unlock before return. Coverity 44888.

This reverts commit cc242de3
That patch fixed bug 2745, but breaks tests 1.89 and 1.91 on
typical Xeon processors

This bug was introduced by commit 21c52d2f
which fixed a different problem tracking resources associated with suspended
jobs. There are subtle differences between jobs that are suspended by a
user/administrator and jobs suspended by gang scheduling which resulted in
undercounting allocated CPUs when a job suspended by gang scheduling
was active at the same time of a slurmctld reconfiguration request.
See bugs 2353 (original bug related to commit 21c52d2f
and bug 2765