It isn't used quite yet, but will be in the next commit!

code copied from slurmdbd_defs.c so we can use it elsewhere.

This way we can put more on here as startup was only a bool anyway.

in the form of assoc_mgr_post_tres_list

Bug 2872

Not sure if this could be done a different way.  At the moment I
wouldn't mind making a common structure to get the read and writes. Or
perhaps make use of a different array or such.

In the end this might be the best way to do it, but it isn't clear at
the moment since the num_* isn't sent to the database in any fashion.

    Bug 2782

This will most likely be reverted as well since we plan on moving the
Usage TRES to dynamic.

Bug 2782

Bug 2782

Not sure if this could be done a different way.  At the moment I
wouldn't mind making a filesystem specific structure to pass in. Or
perhaps make use of a different array or such.

In the end this might be the best way to do it, but it isn't clear at
the moment since the num_* isn't sent to the database in any fashion.

Bug 2782

This will most likely be reverted as well since we plan on moving the
Usage TRES to dynamic.

Bug 2782

At the moment I am wondering if this is needed or not.  Perhaps we might
want to revert this.

Bug 2782

Bug 2782

Bug 2782

Bug 2782

TRES.  This will hopefully be reverted in the future when TRES
information is returned from the slurmd.

Bug 2782

Bug 2782

char * nodelist.

Bug 2782

Bug 2782

Bug 2782

the future.

Bug 2782

to get ready for additions.

Hopefully no code change

Bug 2782

Bug 2782

Bug 2782

Bug 2782

Log the option is not supported using info() message:
$ sbatch -I tmp
sbatch: --immediate option is not supported for the sbatch command, ignored
Submitted batch job 1234

'sacctmgr show accounts withassoc cluster=x' now behaves like 'show users'
displaying always all the accounts on the enterprise. Non-matching filtered
account is now just displayed without associations.

Bug 4804

When Cray NHC is stuck for a job for more than 300 seconds and a
reconfigure is issued, job resources will be released in order to
not keep all the job nodes allocated when there's no reason to.
If then NHC finishes, resources will be released again causing memory
under-allocated errors.

This patch avoids releasing resources more than one time.

Bug 4801

reference commit 08425d9c

Update slurm.spec and slurm.spec-legacy as well

Also throws spurious errors of:
"slurmd: error: Domain socket directory /var/spool/slurmd: No such file or directory"
if you SlurmdSpoolDir is located elsewhere.

Bug 4289.

Add additional protection in slurmctld as well.

Bug 4826.

job_limits_check() uses the job desc to call _valid_pn_min_mem(). This
second function might adjust the following values (up to date):

cpus_per_task
pn_min_memory
min_cpus
max_cpus
pn_min_cpus

If the function returns success, these adjusted members need to be copied
back to the job_record. It turns out pn_min_cpus wasn't copied back,
thus the logs claimed to automatically increase pn_min_cpus but actually
the job record wasn't modified and the select plugin tried to allocate
wrong amount of resources.

Bug 4823.

Bug 4741.