and additional info

    maximun node counts of pending jobs (the operation was previously 
    restricted to user root)

    drained).

    job's time limit).

    notify programs before reaching the end of their time limit.
    (So far, just document the option and get it from the command down
    to slurmctld. More to come later.)

Expand explanation of recent change in NEWS.

    the queue SLURM searched in an attempt to backfill jobs, default value 
    is 50 jobs.

Correction in computing a job's TotalProcs value when ThreadsPerCore>1 and
    allocating by cores or sockets.

 -- Move processing of node configuration information in slurm.conf and
    topology information in topology.conf from slurmctld into common and load 
    that information into slurmd. Use it to set environment variables for jobs
    SLURM_TOPOLOGY_ADDR and SLURM_TOPOLOGY_ADDR_PATTERN describing the network 
    topology for each task. Based upon patch from Mattheu Hautreux (CEA).

If a slurmd does not have a node listed in it's slurm.conf (slurm.conf's should be kept the same on all nodes) an error message is printed in the slurmctld log along with the message already being printed in the slurmd log for easier debugging.

    stderr to "Submitted batch job #" written to stdout.

 -- Recycled the "-A" argument indicate "account" for all the commands that
    accept the --account argument (srun -A to allocate is no longer supported.)

BLUEGENE - env vars such as SLURM_NNODES, SLURM_JOB_NUM_NODES, and SLURM_JOB_CPUS_PER_NODE now reference cnode counts instead of midplane counts.  SLURM_NODELIST still references midplane names.

    nodes already in another reservation.

  from the database

  This adds job preemption support to select/linear for preempt/qos and 
  optimizes job placement on nodes considering workload from other partitions.