Slurm will not terminate jobs if they exceed requested memory.

Conflicts:
	META
	NEWS

Previously used --export command line option, which is not
recognized by salloc command

Conflicts:
	contribs/torque/qsub.pl

If a job is submitted that is dependent upon a job array in which
the first task of that array is already purged from slurmctld, then
find another element of that array and properly handle the dependency.

Conflicts:
	src/plugins/switch/cray/switch_cray.c
	src/slurmctld/proc_req.c

plugin and provide more information in the message.

together with the remote IP address and port.

On switch resource allocation failure, free partial allocation.
Failure mode was CAU could be allocated on some nodes, but not
others. The CAU allocated on nodes and switches up to the failure
point were never released.

added yet.

Don't block scheduling of entire job array if it could run in multiple
partitions. Backfill fix to match related to commit
c075ac75
bug 726

Don't block scheduling of entire job array if it could run in multiple
partitions.
bug 726

ephemeral port.

Conflicts:
	src/slurmctld/job_scheduler.c

Requeue batch job if Munge is down and credential can not be created.
bug 727

Do not overwrite existing reason for node being down or drained.
Previously draining a node would always overwrite any previous
"reason".
bug 724

the current settings.

This reverts commit 2a72aa51.

Conflicts:
	NEWS

Decided to take this out and wait until 14.11 to fix this correctly instead
of this bandaid.

Previously if a job was dependent upon a job array and the first
element of that job array was no longer in slurmctld (completed
and the job record purged) and the slurmctld restarted or was
reconfigured, the dependent job would start rather than waiting
for all elements of the job array to complete.
bug 714

Shorten the test31.1 description for better readability of errors.