Issue the "node_off" command on all nodes identified in a single
  call to capmc. Only if that fails will individual nodes be
  powered down using multiple pthreads. This improves efficiency
  while retaining the ability to operate on individual nodes
  when some failure occurs.
bug 2659

# Conflicts:
#	META

E.g. info, debug, etc.

Had:
if (specname && otherstuff) {
} if (specname) {
}

requesting the RUNNING state.

Batch steps for jobs requeued due to node failure don't get sent to datbase. See comment in _slurm_rpc_complete_batch_script.

# Conflicts:
#	auxdir/x_ac_databases.m4
#	configure

Backport of commit cca1616b from 16.05

MCS plugin should not have been retroactively added to the 15.08 RPCs, and
caused 'scontrol show config' from a 15.08 scontrol to a 16.05 slurmctld to
fail.

# Conflicts:
#	src/common/node_conf.c

pmix >= 1.1.4.