There was an error reported by slurmstepd trying to read the window
size from srun when srun was exiting. This eliminates that spurious
error message.

Avoid running some regression tests on select/linear systems

Avoid abort if trying to create a core reservation on a select/linear
configuration

salloc.1 sbatch.1 srun.1 nonstop.conf.5 slurm.conf.5

Don't put jobs that are not runnable on submit into Bad Constraint state,
reject them.  Only do that if the configuration changes after the job
was accepted.

failure as well as interactive user.

epilog complete message comes from that node do not process the batch step
information since the job has already been requeued because the epilog
script running isn't guaranteed in this situation.

start of requeued jobs.

It is only used for Hadoop and that code is still not fully fleshed out

Expand from 8 to 16 bits for more than 256 sizes

sched/backfill - Fix anomaly that could result in jobs being scheduled out
of order.
bug 911

Without this a batch script could be bound to a single CPU rather
than the full job allocation. This is a revision to commit
399cb897
without which batch scripts has no binding to CPUs via hwloc

squeue and scontrol to report a job's "shared" value based upon partition
options rather than reporting "unknown" if job submission does not use
--exclusive or --shared option.
bug 939

nodes instead of whatever was first in the system.

exit code.

a fix to handle sacct under this new method.