Rely on POWERING_DOWN bit. This allows POWERING_DOWN nodes to be cleared
after a restart -- since suspend_node_bitmap was local to power_save.c.

Bug 6333

Bug 6333

Instead of just guessing the time, let's use the original time.

Bug 6333

While "powering down" the nodes aren't eligible to be allocated.
Nodes will remain "powering down" for SuspendTimeout time.

Bug 6333

NODE_STATE_POWER_SAVE == node is actually off

Bug 6333

Run suspend and resume more often than ResumeTimeout after last suspend.
Don't allocate suspending nodes until after SuspendTimeout.

Bug 6333

If the MPS server is started with the environment variable
CUDA_MPS_ACTIVE_THREAD_PERCENTAGE, then the MPS server will
be limited to the percentage of the GPU total, which will not
work as desired if additional jobs are initiated.

User percentage logic was incorrect and unnecesarry.
Removed the logic and associated test.

Bug 5667

Bug 5667

Bug 5667

Make tests able to work in greater variety of configurations

This will start and stop the MPS server as needed

This should only happen if something bad happens in the test

The previous logic could let a job requiring a persistent burst
buffer start and fail without checking that the buffer already
exists. Was causing regression tests 35.1 and 35.4 to fail.

it would require many changes to the slurm.conf files used in testing
and the functionality being tested here should work the same on
non-cray systems anyway (if it works on a non-cray, the funcitonality
would be fine on a real cray system too).

it would require many changes to the slurm.conf files used in testing
and the functionality being tested here should work the same on
non-cray systems anyway (if it works on a non-cray, the funcitonality
would be fine on a real cray system too).

Change test from select/cray (also used for testing on non-cray
systems) to switch/cray (only used on real cray systems)

This make the test work properly if the default partition configuration
includes a configuration of "OverSubscribe=Exclusive"

Some conditions were resulting in an srun error about no SGI job
container

This fixes a potential task layout problem. Specifically if
a cluster allocates by core and each core contains 2 CPUs and
a job requests 3 tasks and one core on each of two nodes is
available and the first node has one GPU available and the
job requests --gpus-per-task=1 then without this patch
cons_tres would try to put 2 tasks on the first node (one
per CPU). This adds a check of the GPU count in order to
prevent that. Observed sportatically when running regression
tests39.[10-15] in immediate succession on a Cray system
or a system configured with Epilog that takes a few seconds
to complete.

Updated ControlAddr to point to 127.0.0.1 rather than 123.4.5.6

Bug 6794

Bug 6768.

Bug 6779.

Bug 6718.

Replace duplicated text with links, so text that should be the same
does not get out of sync.
Merge out-of-sync text together to get the best of both.
Fix error where line starting with 'nvml' was being omitted.
Minor grammatical and wording fixes.
Improve spacing of paragraphs.
Escape some missed `-` characters.
Remove some statements that are no longer true.
Sundry other minor changes.

An update in gres.conf.5 was not propagated to gres.shtml. Remove the
possibility of that happening again by simply referring the user to
the original doc via a link for more info.

Bug 4717

Bug 6501.