Work is incomplete, but for now we get all jobs in federation and highlight
  the local nodes associated with each job.

Add "cluster_name" field to node_info_t and partition_info_t data structure.
    It is filled in only when the cluster is part of a federation and
    SHOW_GLOBAL flag used.
Functions slurm_load_node() slurm_load_partitions() modified to show all
    nodes/partitions in a federation when the SHOW_GLOBAL flag is used.

In order to test with larger systems, modify capmc script to generate
  output with arbitrary start and end NID values

Bug 3725

in general.

at shutdown. More likely to happen in the fed_mgr than the dbd since the
fed_mgr closes the persistent connections before they are signaled by
the persistent connection server.

pointer.

selected for a job.

Bug 3690

Document the node_features_p_node_xlate2() function and update the
  description of node_features_p_node_xlate(). These changes were
  made in commit 6690685a
bugs 3614, 3679

Allows for annotations such as
xassert(verify_lock(CONFIG_LOCK, READ_LOCK));
to ensure that the calling thread has the correct locks in
place to head off potential race conditions / corruption.

Remove outdated disclaimers, proctrack/cgroup is awesome.

(jobacct_gather/cgroup is still not recommended though.)

Change it to the default with the configurators as well.

It seems to be a popular item. It would be nice to eventually re-write
  the tool to use Slurm's perl API at some time...

Do not create a backfill resource reservation for jobs prevented
  from starting due to QOS limit.
bug 3680

As found by CLANG

No functional change, all are in comment blocks.

Restructure code slightly to make code path cleaner.

No functional change, all are in comment blocks.

As found by CLANG

Bug 3764.

d31965f3 triggered it

When running sacct from a federated client, the db returns jobs for each
cluster with duplicate jobs removed on each cluster. A federated job
could have ran on a different cluster when the before the jobid's
rolled. This patch filters out past old federated jobs and leaves the
newest ones.

With sacct, if multiple clusters were selected then the jobs would be
grouped together by cluster with each cluster's job sorted by submit
time. This patch sorts all of the job by submit time if multiple
clusters are requested.

KNL features: Always keep active and available features in the same order:
    first site-specific features, next MCDRAM modes, last NUMA modes.
bugs 3614, 3679

This change will avoid possible problems in the event that "capmc get_mcdram_cfg"
  fails to return a mcdram_pct value. Never observed, but this will harden the
  code just in case...
bug 3679