This will be especially valuable when generating reports for a
  federation of clusters that no longer reflects the current
  federation.

Earlier tests would generate errors in a federation as the reports
  would include information from non-local clusters

In 6eec8022, the cluster's recv connection is now being destroyed when
the cluster is being destroyed. The problem that showed itself was that
when a remote cluster is removed from the federation, the controller
calls slurmdb_destroy_federation_rec() which destroys the cluster's in
the list. Both the persistent recv thread and the cluster's recv are
pointing to the same thing so when the controller removed the recv
persistent connection the recv thread was pointing to bad memory.

end at the same time.

Bug 3604.

Modify sreport to report all jobs in federation by default. Also add --local
    option.

Modify sacct to accept "--cluster all" option (in addition to the old
    "--cluster -1", which is still accepted). This makes sacct behave
    more like the other commands.

Modify sacct to report all jobs in federation by default. Also add --local
    option.

This RPC is still needed for version 17.02 commands executed
  against version 17.11 slurmctld daemon

perl api.

This changed in 17.11 where if the size was 0 we would return 0 which messes
up the perl api.

Bug 3644

jobs have a priority that is lower than the selected job. Previous logic
would permit other jobs with equal priority (no jobs with higher priority).

Bug 3650

circumstances.

array element per line.

Bug 3573

If the depend_list is NULL or has zero elements, the string should
be cleared as well.

Bug 3651.

The field needs to have its own copy, otherwise the pointer will
become invalid when xfree()'d by a separate array task.

Bug 3665.

So that it is the same max length as in src/common/env.c.

Used for explicitly laying out tasks on large CPU count nodes (e.g., KNL).

Bug 3675.

For use in federated environments.

in any case.

As specified in MPIR debug interface
(https://www.mpi-forum.org/docs/mpir-specification-10-11-2010.pdf),
the presence of the MPIR_partial_attach_ok symbol
should inform the debugger that the initial startup synchronization
is implemented in such a way that the tool need not attach
nor continue MPI processes that the user is not interested in controlling.

To implement this, SLURM chose to send SIGCONT to those processes that are
not attached by the debugger.

However, the old code does not reliably detect the condition
in which a process is traced by the debugger, and this
has lead to various side effects.

On some systems (e.g., TOSS2), the old code sends SIGCONT to
all of the target processes including those attached by the debugger.
On newer systems (e.g., TOSS3), it does not send SIGCONT
to the target processes at all.

It seems that one of the reasons for such undefined behavior
is the use of CLONE_PTRACE.
@grondo found no documentation that indicates
CLONE_PTRACE is for the case where the process is being attached
by a debugger.
More importantly, this code is matching clone(2) flags
to proc(5) process flags, which are not the same, as task->flags
defined as PF_* flags from kernel source include/linux/sched.h.

This patch fixes these problems by replacing
the old detection logic with ones based on the TracerPid field
in /proc/<pid>/status.

From proc(5), TracerPid: PID of process tracing this process (0 if not
being traced).

Improve job scheduling sort after sorting by priority we now sort by
submit time and then by job id.  We used to not consider submit time.  This
handles the case where the job_ids have rolled or we are doing federation
scheduling.

Bug 3524