If a job requires specific nodes and can not run due to those nodes being
busy, the main scheduling loop will block those specific nodes rather than
the entire queue/partition.
bug 595

variable.

This reverts commit 1764a7e2.

Conflicts:

	NEWS

is not a corresponding association.

is not a corresponding association.

This corrects some of the logic in
commit e8b11948
There were several plugins storing the wrong plugin type code
(e.g. checkpoint/poe was recording state as checkpoint/blcr).
In addition, since there are new fields being stored, they
must be recorded as a different protocol_version, otherwise
all jobs would be lost when upgrading the system from an
earlier slurm release.

disabled.

Adds slurmctld/nonstop plugin, libsmdns.so library, slurm/smd_ns.h
and smd command. Also add regression tests for smd command.

Added sbatch '--parsable' option to output only the job id number and the
cluster name separated by a semicolon rather than "Submitted batch job....".
Errors will still be displayed.

Conflicts:
	doc/man/man5/slurm.conf.5

needed to forward a message the slurmd would core dump.

This might happen if someone runs autogen.sh with some old version
of libtool

When an error occurs when trying to suspend, resume, or requeue a
job array report the error with the array element in addition to
the native slurm job id. Previously it only reported the native
slurm job id.

reporting "array job_id" error for non-array jobs

describing that jobs must be drained from cluster before deploying
any checkpoint plugin.

Re-order existing code so that per-cpu memory limits will be
enforce with both cpus_per_task and ntasks_per_node limits.