If the job to be suspended can not be found, then proceed with the
suspend RPC after 3 seconds, Previous logic would hang indefinitely.
The job would not be found if it was launched, then the slurmd restarted.
The record keeping for launched jobs after slurmd restarts is fixed
in a separate patch.

This reverts commit bf83bcf7.

This reverts commit 4294c561.

requeue, requeuehold and release operations.

added is a user process or not.

cont_id contained in it.

need to do.

Do not purge the script and environment files for completed jobs on
slurmctld reconfiguration or restart (they might be later requeued).
Purge the files only when the job record is purged.
bug 834

If a job --mem-per-cpu limit exceeds the partition or system limit, then
scale the job's memory limit and CPUs per task to satisfy the limit.
bug 848

not finished yet otherwise if requeued the job may enter an invalid
COMPLETING state.

messages, moving some messages to higher debug levels, and
consolidating some needlessly verbose messages.

applications from the minimum (1%) to the more appropriate 25%.

the Aries network.

network systems

- Resolves a bug involving nested batch steps and aeld.

(mostly useful for debugging, it isn't used for much)

Sets task_is_app, which tells the kernel that this is an application
process (for core specialization)

limits to 4 concurrent jobs per node for our network resources.

generic resources and removing unused warmswap code.

@bindir@ would resolve to ${prefix}/bin.

This patch fixes it, based on http://www.gnu.org/software/autoconf/manual/
autoconf-2.69/html_node/Installation-Directory-Variables.html

It also changes opt_modulefiles_slurm to opt_modulefiles_slurm.in but
I couldn't figure out how to get git diff to show that.

If a job allocates whole nodes (with --core-spec or --exclusive
option) and launches a step within that allocation all in a single
command (the srun creates the allocation and step at the same time)
and requests a specific CPU count (e.g. "-ntasks=# --cpus-per-task=#")
then allocate the job step only the requested CPU count, which may
be less than the job's allocation.
Bug 843

If shutdown of the slurmctld daemon is in progress, then stop
trying to schedule jobs or process reconfigure requests. These
are the only operations that take a significant amount of time
and only service to slow down the shutdown process. We want the
daemon to stop processing incoming RPCs and save state as soon
as possible.

Previous limit was 4 secs, raised to 10.

Increase delays if memory leak test
Disable one test if tracking lustre file system use (leaves open file)

Eliminate one-time memory leak by freeing RPC statistics table at
slurmctld shutdown

Eliminate a one-time memory leack of the assocation hash table at shutdown.