would only track gres/gpu, now it will track both gres/gpu and
gres/gpu:tesla as separate gres if configured like
AccountingStorageTRES=gres/gpu,gres/gpu:tesla

Continuation of 31225a82

Bug 2430

We want to introduce a new behavior in the way slurmd uses the
HealthCheckProgram. The idea is to avoid a race condition between the
first HealthCheckProgram run and the node accepting jobs. The slurmd
daemon will initialize and then loop on HealthCheckProgram execution
before registering with slurmctld. It will stay in this loop until
the HealthCheckProgram returns successfully (the node is still DOWN).

On our clusters we are using NHC as an HealthCheckProgram. NHC drains
the node if it fails and remove the drain if it is successfull, this
behavior fits well with our purpose. This behavior permits us to start
slurmd at boot without setting up a complex boot sequence in the init
system, slurmd just wait for the node to be ready before registering.

The HealthCheckProgram is not run during slurmd startup if
HealthCheckInteval is 0.

Bug 2507

Step GRES value changed from type "int" to "int64_t" to support larger
values. Previous logic could fail in step allocation values over 32-bits.
Other GRES values are 64-bit.

slurmstepd to close potential open ones.

It was pointed out the slurmd using acct_gather_energy/ipmi links to
freeipmi which could possibly open /dev/ipmi0 without the close on exec
flag set as root while launching a step leaving it open in the users app.

What this does is sets the flag on the first 256 to mitigate the concern.

Reported by Maksym Planeta.

Bug 2506

Previous logic tested whatever the job's partition pointer indicated
rather than the partition we are trying to run the job in. This bug
was introduced in Slurm version 15.08.5, Nov 16, 2015, commit
94f0e948
bug 2499

Check that PowerSave mode configured for node_features/knl_cray plugin.
    It is required to reconfigure and reboot nodes. Fatal error if
    not configured.

src/common/mapping.h was the one place outside of slurm/*h that used this,
just remove it from there.

Replace macro with #ifdef __cplusplus in slurm/*h in case anyone is linking
C++ against libslurm.

Macro hasn't been used consistently for three+ years, and is protecting against
compilation by non-ANSI C compilers which has not been a concern for quite some
time. Cleanup formatting of function declarations while here.

No change to logic.

Insure that a job is completely launched before trying to suspend it.
Previous logic would start suspend logic early in the life of the
slurmstepd process, after it's listening socket was open but before
the tasks were launched. This defers the suspend logic until after
all prologs and setup completes and the tasks are launched. This is
important in the case of gang scheduling, in which newly launched
jobs can be immediately suspended.
bug 2494

Default value is 1. Weight is uint32_t so this check was always succeeding.

Add note to slurm.conf man page about setting "--cpu_bind=no" as part
of SallocDefaultCommand if a TaskPlugin is in use.

Revert call to getaddrinfo, restoring gethostbyaddr (introduced in Slurm
    16.05.0pre1) which was failing on some systems. Specifically test7.2
    was failing on some systems with getaddrinfo() returning an error of
    "System error: Resource temporarily unavailable". Partial reversion
    of commit 89621f65

was also given.

Split partition's "Priority" field into "PriorityTier" (used to order
    partitions for scheduling and preemption) plus "PriorityJobFactor" (used by
    priority/multifactor plugin in calculating job priority, which is used to
    order jobs within a partition for scheduling).
bug 2479

a partition.

This also reverts most of commit fa331e30 as well as commit bd9fa830
which would try to set the pn_min_cpus every time a job was updated.
If a job didn't request node counts then they were hosed.

This commit takes away the magic which was screwing things up.  Now the
person gets what they asked for without magic changing things.

Bug 2302
Bug 2742
Bug 2478

erroneously.

This whole process could probably be done better by keeping track of
old values and new values and only calling one function instead of a
pre and post function, but that can probably wait for future generations
of the code as it works now and is probably adequate for the time being.

Bug 2352

BurstBuffer/cray - Defer job cancellation or time limit while "pre-run"
    operation in progress to avoid inconsistent state due to multiple calls
    to job termination functions.
bug 2454

a new account and making it a default all at once.

Bug 2428

Match acct_gather_energy/rapl plugin. Bug 2397.

Control whether the scheduler will continue to try to run jobs in
a partition if a higher priority job is stuck due to an association
limit.

Can cause starvation for larger jobs, but will improve throughput and
utilization for systems that have extensively divvyed up their resources
through association/QOS limits.

Bug 2388 and 2452.

Bug 2453