Intel MPI, on MPI jobs initialisation through PMI, uses to call PMI_KVS_Put()
many many times from task at rank 0, and each on these call is followed by
PMI_KVS_Commit(). Slurm implementation of PMI_KVS_Commit() imposes a delay
to avoid DDOS on original srun. This delay is proportional to the total number.
It could be up to 3 secs for large jobs for ex. with 7168 tasks. Therefore,
when Intel MPI calls PMI_KVS_Commit() 475 times (mesured on a test case) from
task at rank 0, 28 minutes are spent in delay function.
All other tasks in the job are waiting for a PMI_Barrier. Therefore, there is
no risk for a DDOS from this single task 0. The patch alters the delaying time
calculation to make sure task at rank 0 will does not be delayed. All other
tasks are globally spreaded in the same time range as before.

This moves a bzero() call checked in with commit 30e45f8a
I also noticed that test1.14 was generating errors like this
"srun: error: cpus_per_node array is not set"
This was due to previously uninitialized variables now being
cleared by bzero (i.e. the old data was garbage, but avoided
the error message). The properly cleared variables were introduced
in commit 0252a63e
bug 1306

This is a correction to commit 0252a63e
Previous logic failed to populate data structure as used in another RPC
bug 1306

    of Slurm daemons.

The slurmstepd still needs to be fixed, which most likely can't be fixed
until 15.08.

Purge temporary files for purged jobs
Improve fault-tolerance of pthread_create() errors
Add more states (running, teardown, complete)
Trigger job scheduler upon stage-in completion
Remove attempt to periodically load state info for individual jobs
set time stamp on job record state changes

of Slurm daemons.

instead of only in the DBD.  We also make it possible for Operators to
manage WCKeys instead of just admins.

instead of only in the DBD.

instead of only in the DBD.

instead of only in the DBD.

instead of only in the DBD.

and SLURM_JOB_RESERVATION in the batch job.

the QOS they have access to in the account they are coordinator over.

looking for limits.  It turns out '' is the same as 0,

so both of these below are the same thing...

set @qos = '';
set @qos = 0;

correct acct name since it could possibly change with some queries.

but then sets CPUs to only represent the number of cores on the node.

Add asynchronous stage-out logic
Add asynchronous teardown logic

Correct arguments to spawned programs
Add asynchronous stage-in logic
Set up bbs_pre_run call logic
Fix locking deadlock in _teardown logic
Capture stdout and stderr from spawned processes

Allignment was bad and there was redundant test logic (duplicate errors)

Bug 1327

the code accordingly.