This removes a duplicated test statement which appears identically twice.

This adds support for the SLURM_CLUSTERS environment variable also for sprio.
It also makes the test for the priority plugin type dependent on whether
running with multiple cluster support or not.

On our frontend host we support multiple clusters (Cray and non-Cray) by
setting the SLURM_CLUSTERS environment variable accordingly.

In order to use scontrol (e.g. for hold/release of a user job) from a
frontend host to control jobs on a remote Cray system, we need support for
the SLURM_CLUSTERS environment variable also in scontrol.

The current code erases the old nice value (both negative and positive) when a job is
put on hold so that the job has a 0 nice component upon release.

This interaction causes difficulties if the nice value set at submission time had been
set there for a reason, for instance when
 * a system administrator has allowed to set a negative nice value;
 * the user wanted to keep this as a low-priority job and wants his/her other jobs
   to go first (indenpendent of the hold option);
 * the nice value is used for other semantics - at our site for instance, we use it
   for computed "base priority values" that are computed by looking at how much of
   their quota a given group has already (over)used.

Here is an example which illustrates the loss of original nice values:

  [2011-03-31T09:47:53] sched: update_job: setting priority to 0 for job_id 55
  [2011-03-31T09:47:53] sched: update_job: setting priority to 0 for job_id 66
  [2011-03-31T09:47:53] sched: update_job: setting priority to 0 for job_id 77
  [2011-03-31T09:47:54] sched: update_job: setting priority to 0 for job_id 88
  [2011-03-31T09:47:54] sched: update_job: setting priority to 0 for job_id 99
  [2011-03-31T09:47:54] sched: update_job: setting priority to 0 for job_id 110

This is from user 'kraused' whose project 's310' is within the allocated quota and thus
has an initial nice value of -542 (set via the job_submit/lua plugin).

However, by putting his jobs on hold, he has lost this advantage:

  JOBID     USER   PRIORITY        AGE  FAIRSHARE    JOBSIZE  PARTITION        QOS   NICE
     55  kraused      15181        153          0       5028      10000          0      0
     66  kraused      15181        153          0       5028      10000          0      0
     77  kraused      15181        153          0       5028      10000          0      0
     88  kraused      15178        150          0       5028      10000          0      0
     99  kraused      15178        150          0       5028      10000          0      0
    110  kraused      15178        150          0       5028      10000          0      0

I believe that resetting the nice value has been there for a reason, thus the patch prevents
reset of current nice value only if the operation is not user/administrator hold.

This fixes a problem when trying to move a pending job from one partitition to another
while not supplying any other parameters:
 * if a partition value is present, the job is pending and no min_nodes are supplied,
   the job_specs->min_nodes get set from the detail_ptr value;
 * this causes subsequent tests for job_specs->min_nodes ==/!= NO_VAL to fail.

The following illustrates the behaviour, the example is taken from our system:
  palu2:0 ~>scontrol update jobid=3944 partition=night
  slurm_update error: Requested operation not supported on this system

  slurmctld.log
  [2011-04-06T14:39:51] update_job: setting partition to night for job_id 3944
  [2011-04-06T14:39:51] Change of size for job 3944 not supported
  [2011-04-06T14:39:51] updating accounting
  [2011-04-06T14:39:51] _slurm_rpc_update_job JobId=3944 uid=21215: Requested operation not supported on this system

==> The 'Change of size for job 3944' reveals that the !select_g_job_expand_allow() case was triggered,
    after setting the job_specs->min_nodes due to supplying a job_specs->partition.

Fix:
====
 Since the test for select_g_job_expand_allow() is not dependent on the job state, moved it up, before
 the test for the job_specs->partition. At the same time, also moved the equality for INFINITE/NO_VAL
 min_nodes values to the same place.
 Tests for job_specs->min_nodes below the job_specs->partition setting depend on the job state, 
 - the 'Reset min and max node counts as needed, insure consistency' requires pending state;
 - the other remaining test is only for IS_JOB_RUNNING/SUSPENDED.

This patch avoids that the priority is not recalculated on 'scontrol release',
which  happens when an authorized operator releases a job, or if the job is
released via e.g. the job_submit plugin.

The patch reorders the tests in update_job() to 
 * test first if the job has been held by the user and, only if not,
 * test whether an authorized operator changed the priority or
   the updated priority is being reduced.

Due to earlier permission checks, we have either
 * job_ptr->user_id == uid or 
 * authorized,
where in both cases the release-user-hold operation is authorized.

This fix is related to an earlier one and was observed when trying to 'scontrol release'
a job previously submitted via 'sbatch --hold' by the same user.

Within the job_submit/lua plugin, the user gets automatically assigned a partition. So,
even if no submitter uid checks are usually expected, it can happen in the process of
releasing a job, that a part_check is performed.

In this case, the error message was

[2011-03-30T18:37:17] _part_access_check: uid 4294967294 access to partition usup denied, bad group
[2011-03-30T18:37:17] error: _slurm_rpc_update_job JobId=12856 uid=21215: User's group not permitted to use this partition

and like before (in scontrol_update_job()), was fixed by supplying the UID of the requesting user.

We build slurm with --enable-memory-leak-debug and encountered twice the same core
dump when user 'root' was trying to run jobs during a maintenance session. 

The root user is not in the accounting database, which explains the errors seen
below. The gdb session shows that in this invocation 

palu7:0 log>stat /var/crash/palu7-slurmstepd-6602.core 
...
Modify: 2011-04-04 19:34:44.000000000 +0200

slurmctld.log
[2011-04-04T19:34:44] _slurm_rpc_submit_batch_job JobId=3254 usec=1773
[2011-04-04T19:34:44] ALPS RESERVATION #5, JobId 3254: BASIL -n 1920 -N 0 -d 1 -m 1333
[2011-04-04T19:34:44] sched: Allocate JobId=3254 NodeList=nid000[03-13,18-29,32-88] #CPUs=1920
[2011-04-04T19:34:44] error: slurmd error 4005 running JobId=3254 on front_end=palu7: User not found on host
[2011-04-04T19:34:44] update_front_end: set state of palu7 to DRAINING
[2011-04-04T19:34:44] completing job 3254
[2011-04-04T19:34:44] Requeue JobId=3254 due to node failure
[2011-04-04T19:34:44] sched: job_complete for JobId=3254 successful
[2011-04-04T19:34:44] requeue batch job 3254
[2011-04-04T20:28:43] sched: Cancel of JobId=3254 by UID=0, usec=57285

(gdb) core-file palu7-slurmstepd-6602.core 
[New Thread 6604]
Core was generated by `/opt/slurm/2.3.0/sbin/slurmstepd'.
Program terminated with signal 11, Segmentation fault.
#0  main (argc=1, argv=0x7fffd65a1fd8) at slurmstepd.c:413
413             jobacct_gather_g_destroy(job->jobacct);
(gdb) print job
$1 = (slurmd_job_t *) 0x0
(gdb) list
408
409     #ifdef MEMORY_LEAK_DEBUG
410     static void
411     _step_cleanup(slurmd_job_t *job, slurm_msg_t *msg, int rc)
412     {
413             jobacct_gather_g_destroy(job->jobacct);
414             if (!job->batch)
415                     job_destroy(job);
416             /*
417              * The message cannot be freed until the jobstep is complete
(gdb) print msg
$2 = (slurm_msg_t *) 0x916008
(gdb) print rc
$3 = -1
(gdb) 

The patch tests for a NULL job argument for the calls that need to dereference the job pointer.

This avoids meaningless error messages that warn about a zero reservation ID:

 [2011-04-07T15:31:26] _slurm_rpc_submit_batch_job JobId=2870 usec=33390
                       ... a minute later the user decides to scancel the queued job:
 [2011-04-07T15:32:34] error: JobId=2870 has invalid (ZERO) resId
 [2011-04-07T15:32:34] sched: Cancel of JobId=2870 by UID=21770, usec=230

To keep things simple, that test has been removed.

(The patch is in particular also necessary since now job_signal() may trigger
 a basil_release() of a pending job which has no ALPS reservation yet.)

On rosa we experienced severe problems when jobs got killed via scancel or
as a result of job timeout. Job cleanup took several minutes, created stray
processes that consumed resources on the slurmd node, keeping the system 
for long spans unable from scheduling.

This problem did not show up on the smaller 2-cabinet XE system (which also
runs a more recent ALPS version). The fix for the problem is to keep new
script lines from starting by sending apkill only after formally releasing
the reservation.

For all signals whose default disposition is to terminate or to dump core,
the reservation is released before signalling the aprun job steps. This
prevents a race condition where further aprun lines get executed while the
apkill of the current aprun line in the job script is in progress.

We did a before/after test on rosa under full load and the problem disappeared.

    Patch from Andrej N. Gritsenko.

Most of the changes were to support hostname with 4 digit suffix.
Some other tests are failing, but will leave for now

they are satisfied.

    rather than the original dependencies. From Dan Rusak, Bull.

add some more logging information

   Submitted by Martin Perry / Rod Schultz, Bull.

    avoid purging active jobs.

    event triggers.

    errors after slurmctld restarts.

Fix so slurmctld will pack correctly 2.1 step information. (Only needed if a 2.1 client is talking to a 2.2 slurmctld.)