sched/backfill: If max_rpc_cnt is configured and the backlog of RPCs has
    not cleared after yielding locks, then continue to sleep.

requesting any specific association.

calls.

snowflake10 after snowflake8

Disable srun --bcast option test if SlurmdUser != root as the
  operation/test  will not work otherwise.

This would have helped to identify the performance issue associated
with bug 2174

sacctmgr output used a suffix of "PU" for per user, but only
  accepted input with a suffix of "PerUser". The output all
  used a suffix of "PU". The code has been modified to accept
  both sufficies (e.g. MaxJobsPerUser=# or MaxJobsPU=#).

bug 2147

BurstBuffer/cray: Fix job record purging if cancelled from pending state.
The problem can occur when the a burst buffer record was created for
the job in the plugin data structure, but no burst buffers were actually
allocated for it.
bug 2165

BurstBuffer/cray: Enable clearing of burst buffer string on completed job
    as a means of recovering from a failure mode.
Format is "scontrol update jobid=### burstbuffer=".
partial resolution of bug 2165

This is a backup in the event that slurmstepd fails in such a
  fashion that nothing gets logged. Added to help diagnose
  bug 2028

bug 2028

BurstBuffer/cray: Add logic to terminate dw_wlm_cli child processes at
    shutdown.
bug 2166

Previous logic required the buffer name to work
bug 2167

Distinguish between job and step IDs in sbcast (or srun --bcast)
  logs.

bug introduced in commit 06b54823

Added srun option of --bcast to move executable file to compute nodes

burst_buffer/cray: Support file staging when job lacks job-specific buffer
    (i.e. only persistent burst buffers).
bug 2113

No change to document contents, just reorded the fields documented.

bug 2143

Always log some operations (previously logging was only triggered
  by long run time of the operations)
Log some operations with info() rather than debug() for DebugFlags=
  burstbuffer