stop the current poll asap.

having the realtime server go down over and over again while waiting
for the poll to finish.

to the functions instead of the global rt_running flag

host names. Like done elsewhere in SLURM

users/accounts/flags

correctly for current job.

The uptime is extra information not used in the slurm.conf file

realtime server.

correct prefix

start in the future but is able to run now.

Previously the SIGCONT would be generated easily as part of gang
scheduling and cause the job step initiaition to abort.
This is being back-ported from SLURM v2.5/master.

One of our testers discovered an inconsistency between trigger names that
are displayed with 'strigger --get' and the trigger names passed to trigger
programs.

accounting as well.

We've discovered that our implementation of proctrack/cgroups includes
internal threads as well as the process pids so that "scontrol list pids"
shows them as well. The attached patch (courtesy of Matthieu Hautreux)
to 2.4.2 fixes this problem.

running on it are preemptable by scheduling job.

reported some suspended time.

static blocks.