doc/html/resource_limits.shtml

<!--#include virtual="header.txt"-->

<h1>Resource Limits</h1>

<p>Familiarity with Slurm's <a href="accounting.html">Accounting</a> web page
is strongly recommended before use of this document.</p>

<h2 id="hierarchy">Hierarchy<a class="slurm_link" href="#hierarchy"></a></h2>

<p>Slurm's hierarchical limits are enforced in the following order
  with Job QOS and Partition QOS order being reversible by using the QOS
  flag 'OverPartQOS':</p>
<ol>
	<li>Partition QOS limit</li>
	<li>Job QOS limit</li>
	<li>User association</li>
	<li>Account association(s), ascending the hierarchy</li>
	<li>Root/Cluster association</li>
	<li>Partition limit</li>
	<li>None</li>
</ol>

<p>Note: If limits are defined at multiple points in this hierarchy,
the point in this list where the limit is first defined will be used.
Consider the following example:</p>
<ul>
<li>MaxJobs=20 and MaxSubmitJobs is undefined in the partition QOS</li>
<li>No limits are set in the job QOS and</li>
<li>MaxJobs=4 and MaxSubmitJobs=50 in the user association</li>
</ul>
<p>The limits in effect will be MaxJobs=20 and MaxSubmitJobs=50.</p>

<p>Note: The precedence order specified above is respected except for the
following limits: Max[Time|Wall], [Min|Max]Nodes. For these limits, even
if the job is enforced with QOS and/or Association limits, it can't
go over the limit imposed at Partition level, even if it listed at the bottom.
So the default for these 3 types of limits is that they are upper bound by the
Partition one. This Partition level bound can be ignored if
the respective QOS PartitionTimeLimit and/or Partition[Max|Min]Nodes flags
are set, then the job would be enforced the limits imposed at QOS
and/or association level respecting the order above.
<b>Grp*</b> limits are also an exception. A more restrictive limit at the
Account level will be enforced before a less restrictive limit at the User
level. This is due to the nature of the limit being enforced, requiring that
the limit at the highest level not be exceeded.
</p>

<h2 id="config">Configuration<a class="slurm_link" href="#config"></a></h2>

<p>Scheduling policy information must be stored in a database
as specified by the <b>AccountingStorageType</b> configuration parameter
in the <b>slurm.conf</b> configuration file.
Information can be recorded in a <a href="http://www.mysql.com/">MySQL</a> or
<a href="https://mariadb.org/">MariaDB</a> database.
For security and performance reasons, the use of
SlurmDBD (Slurm Database Daemon) as a front-end to the
database is strongly recommended.
SlurmDBD uses a Slurm authentication plugin (e.g. MUNGE).
SlurmDBD also uses an existing Slurm accounting storage plugin
to maximize code reuse.
SlurmDBD uses data caching and prioritization of pending requests
in order to optimize performance.
While SlurmDBD relies upon existing Slurm plugins for authentication
and database use, the other Slurm commands and daemons are not required
on the host where SlurmDBD is installed.
Only the <i>slurmdbd</i> and <i>slurm-plugins</i> RPMs are required
for SlurmDBD execution.</p>

<p>Both accounting and scheduling policies are configured based upon
an <i>association</i>. An <i>association</i> is a 4-tuple consisting
of the cluster name, account, user and (optionally) the Slurm
partition.
In order to enforce scheduling policy, set the value of
<b>AccountingStorageEnforce</b>.
This option contains a comma separated list of options you may want to
enforce.  The valid options are:
<ul>
<li>associations - This will prevent users from running jobs if
their <i>association</i> is not in the database. This option will
prevent users from accessing invalid accounts.
</li>
<li>limits - This will enforce limits set to associations.  By setting
  this option, the 'associations' option is also set.
</li>
<li>qos - This will require all jobs to specify (either overtly or by
default) a valid qos (Quality of Service).  QOS values are defined for
each association in the database.  By setting this option, the
'associations' option is also set.
</li>
<li>safe - This will ensure a job will only be launched when using an
  association or qos that has a TRES-minutes limit set if the job will be
  able to run to completion. Without this option set, jobs will be
  launched as long as their usage hasn't reached the TRES-minutes limit
  which can lead to jobs being launched but then killed when the limit is
  reached.
  With the 'safe' option set, a job won't be killed due to limits,
  even if the limits are changed after the job was started and the
  association or qos violates the updated limits.
  By setting this option, both the 'associations' option and the
  'limits' option are set automatically.
</li>
<li>wckeys - This will prevent users from running jobs under a wckey
  that they don't have access to.  By using this option, the
  'associations' option is also set.  The 'TrackWCKey' option is also
  set to true.
</li>
</ul>

<p><b>NOTE</b>: The association is a combination of cluster, account,
user names and optional partition name.
<br>
Without AccountingStorageEnforce being set (the default behavior)
jobs will be executed based upon policies configured in Slurm on each
cluster.
</p>

<h2 id="tools">Tools<a class="slurm_link" href="#tools"></a></h2>

<p>The tool used to manage accounting policy is <i>sacctmgr</i>.
It can be used to create and delete cluster, user, account,
and partition records plus their combined <i>association</i> record.
See <i>man sacctmgr</i> for details on this tools and examples of
its use.</p>

<p>Changes made to the scheduling policy are uploaded to
the Slurm control daemons on the various clusters and take effect
immediately. When an association is deleted, all running or pending
jobs which belong to that association are immediately canceled.
When limits are lowered, running jobs will not be canceled to
satisfy the new limits, but the new lower limits will be enforced.</p>

<h2 id="limits">Association specific limits and scheduling policies
<a class="slurm_link" href="#assoc"></a>
</h2>
<p>These represent the limits and scheduling policies relevant to Associations.
When dealing with Associations, most of these limits are available
not only for the user association, but also for each cluster and account.
Limits and policies are applied in the following order:
<br>
1. The option specified for the user association.
<br>
2. The option specified for the account.
<br>
3. The option specified for the cluster.
<br>
4. If nothing is configured at the above levels, no limit will be applied.
</p>

<p>These are just the limits and policies for Associations. For a more
complete description of the columns available to be displayed, see the
<a href="sacctmgr.html#SECTION_LIST/SHOW-ASSOCIATION-FORMAT-OPTIONS">
sacctmgr</a> man page.</p>

<dl>
<dt id="assoc_fairshare"><b>Fairshare</b>
<a class="slurm_link" href="#assoc_fairshare"></a></dt>
<dd>Integer value used for determining priority.
  Essentially this is the amount of claim this association and its
  children have to the above system. Can also be the string "parent",
  when used on a user this means that the parent association is used
  for fairshare.  If Fairshare=parent is set on an account, that
  account's children will be effectively re-parented for fairshare
  calculations to the first parent of their parent that is not
  Fairshare=parent.  Limits remain the same, only its fairshare value
  is affected.
</dd>

<dt id="assoc_grpjobs"><b>GrpJobs</b>
<a class="slurm_link" href="#assoc_grpjobs"></a></dt>
<dd>The total number of jobs able to run at any given
  time from an association and its children.  If
  this limit is reached, new jobs will be queued but only allowed to
  run after previous jobs complete from this group.
</dd>

<dt id="assoc_grpjobsaccrue"><b>GrpJobsAccrue</b>
<a class="slurm_link" href="#assoc_grpjobsaccrue"></a></dt>
<dd>The total number of pending jobs able to accrue age
  priority at any given time from an association and its children.  If
  this limit is reached, new jobs will be queued but not accrue age priority
  until after previous jobs are removed from pending in this group.
  This limit does not determine if the job can run or not, it only limits the
  age factor of the priority.
</dd>

<dt id="assoc_grpsubmitjobs"><b>GrpSubmitJobs</b>
<a class="slurm_link" href="#assoc_grpsubmitjobs"></a></dt>
<dd>The total number of jobs able to be submitted
  to the system at any given time from an association and its children.
  If this limit is reached, new submission requests will be
  denied until previous jobs complete from this group.
</dd>

<dt id="assoc_grptres"><b>GrpTRES</b>
<a class="slurm_link" href="#assoc_grptres"></a></dt>
<dd>The total count of TRES able to be used at any given
  time from jobs running from an association and its children. If
  this limit is reached, new jobs will be queued but only allowed to
  run after resources have been relinquished from this group.
</dd>

<dt id="assoc_grptresmins"><b>GrpTRESMins</b>
<a class="slurm_link" href="#assoc_grptresmins"></a></dt>
<dd>The total number of TRES minutes that can
  possibly be used by past, present and future jobs
  running from an association and its children. If any limit is reached,
  all running jobs with that TRES in this group will be killed, and no new
  jobs will be allowed to run.  This usage is decayed (at a rate of
  PriorityDecayHalfLife).  It can also be reset (according to
  PriorityUsageResetPeriod) in order to allow jobs to run against the
  association tree.
  The decay rate can be set per TRES with
  <a href="#tresdecayhalflife">TresDecayHalfLife</a>.
  This limit only applies when using the Priority Multifactor plugin.
</dd>

<dt id="assoc_grptresrunmins"><b>GrpTRESRunMins</b>
<a class="slurm_link" href="#assoc_grptresrunmins"></a></dt>
<dd>Used to limit the combined total number of TRES
  minutes used by all jobs running with an association and its
  children.  This takes into consideration time limit of
  running jobs and consumes it. If the limit is reached, no new jobs
  are started until other jobs finish to allow time to free up.
</dd>

<dt id="assoc_grpwall"><b>GrpWall</b>
<a class="slurm_link" href="#assoc_grpwall"></a></dt>
<dd>The maximum wall clock time running jobs are able
  to be allocated in aggregate for an association and its children.
  If this limit is reached, future jobs in this association will be
  queued until they are able to run inside the limit.
  This usage is decayed (at a rate of
  PriorityDecayHalfLife).  It can also be reset (according to
  PriorityUsageResetPeriod) in order to allow jobs to run against the
  association tree again.
</dd>

<dt id="assoc_maxjobs"><b>MaxJobs</b>
<a class="slurm_link" href="#assoc_maxjobs"></a></dt>
<dd>The total number of jobs able to run at any given
  time for the given association.  If this limit is reached, new jobs will
  be queued but only allowed to run after existing jobs in the association
  complete.
</dd>

<dt id="assoc_maxjobsaccrue"><b>MaxJobsAccrue</b>
<a class="slurm_link" href="#assoc_maxjobsaccrue"></a></dt>
<dd>The maximum number of pending jobs able to accrue age
  priority at any given time for the given association.  If this limit is
  reached, new jobs will be queued but will not accrue age priority
  until after existing jobs in the association are moved from a pending state.
  This limit does not determine if the job can run, it only limits the
  age factor of the priority.
</dd>

<dt id="assoc_maxsubmitjobs"><b>MaxSubmitJobs</b>
<a class="slurm_link" href="#assoc_maxsubmitjobs"></a></dt>
<dd>The maximum number of jobs able to be submitted
  to the system at any given time from the given association.  If
  this limit is reached, new submission requests will be denied until
  existing jobs in this association complete.
</dd>

<dt id="assoc_maxtresminsperjob"><b>MaxTRESMinsPerJob</b>
<a class="slurm_link" href="#assoc_maxtresminsperjob"></a></dt>
<dd>A limit of TRES minutes to be used by a job.
  If this limit is reached, the job will be killed if not running in
  Safe mode, otherwise the job will pend until enough time is given to
  complete the job.
</dd>

<dt id="assoc_maxtresperjob"><b>MaxTRESPerJob</b>
<a class="slurm_link" href="#assoc_maxtresperjob"></a></dt>
<dd>The maximum size in TRES any given job can
  have from the association.
</dd>

<dt id="assoc_maxtrespernode"><b>MaxTRESPerNode</b>
<a class="slurm_link" href="#assoc_maxtrespernode"></a></dt>
<dd>The maximum size in TRES each node in a job
  allocation can use.
</dd>

<!-- For future use
<li><b>MaxTRESRunMinsPerJob=</b> A limit of TRES minutes to be used by jobs
  running from the given association/QOS.  If this limit is
  reached the job will be killed will be allowed to run.
</li>
-->

<dt id="assoc_maxwalldurationperjob"><b>MaxWallDurationPerJob</b>
<a class="slurm_link" href="#assoc_maxwalldurationperjob"></a></dt>
<dd>The maximum wall clock time any individual job
  can run for in the given association.  If this limit is reached,
  the job will be denied at submission.
</dd>

<dt id="assoc_minpriothreshold"><b>MinPrioThreshold</b>
<a class="slurm_link" href="#assoc_minpriothreshold"></a></dt>
<dd>Minimum priority required to reserve resources
  in the given association. Used to override bf_min_prio_reserve.
  See <a href="slurm.conf.html#OPT_bf_min_prio_reserve=#">
  bf_min_prio_reserve</a> for details.
</dd>

<dt id="assoc_qos"><b>QOS</b>
<a class="slurm_link" href="#assoc_qos"></a></dt>
<dd>comma separated list of QOSs an association is
  able to run.
</dd>
</dl>

<p><b>NOTE</b>: When modifying a TRES field with <i>sacctmgr</i>, one must
specify which TRES to modify (see <a href="tres.html">TRES</a> for complete
list) as in the following examples: </p>
<pre>
SET:
sacctmgr modify user bob set GrpTRES=cpu=1500,mem=200,gres/gpu=50
UNSET:
sacctmgr modify user bob set GrpTRES=cpu=-1,mem=-1,gres/gpu=-1
</pre>


<h2 id="qos">QOS specific limits and scheduling policies
<a class="slurm_link" href="#qos"></a>
</h2>

<p>As noted <a href="#hierarchy">above</a>, the default behavior is that
a limit set on a Partition QOS will be applied before a limit on the job's
requested QOS. You can change this behavior with the <i>OverPartQOS</i>
flag.</p>

<p>Unless noted, if a job request breaches a given limit
on its own, the job will pend unless the job's QOS has the DenyOnLimit
flag set, which will cause the job to be denied at submission.  When
Grp limits are considered with respect to this flag the Grp limit
is treated as a Max limit.</p>

<dl>
<dt id="qos_gracetime"><b>GraceTime</b>
<a class="slurm_link" href="#qos_gracetime"></a></dt>
<dd>Preemption grace time to be extended to a job which
  has been selected for preemption in the format of
  &lt;hh&gt;:&lt;mm&gt;:&lt;ss&gt;. The default value is zero,
  meaning no preemption grace time is allowed on this QOS. This value
  is only meaningful for QOS PreemptMode=CANCEL and PreemptMode=REQUEUE.
</dd>

<dt id="qos_grpjobs"><b>GrpJobs</b>
<a class="slurm_link" href="#qos_grpjobs"></a></dt>
<dd>The total number of jobs able to run at any given time
  from a QOS. If this limit is reached, new jobs will be queued but only
  allowed to run after previous jobs complete from this group.
</dd>

<dt id="qos_grpjobsaccrue"><b>GrpJobsAccrue</b>
<a class="slurm_link" href="#qos_grpjobsaccrue"></a></dt>
<dd>The total number of pending jobs able to accrue age priority at any
  given time from a QOS. If this limit is reached, new jobs will be queued but
  will not accrue age based priority until after previous jobs are removed
  from pending in this group. This limit does not determine if the job can
  run or not, it only limits the age factor of the priority. This limit only
  applies to the job's QOS and not the partition's QOS.
</dd>

<dt id="qos_grpsubmitjobs"><b>GrpSubmitJobs</b>
<a class="slurm_link" href="#qos_grpsubmitjobs"></a></dt>
<dd>The total number of jobs able to be submitted to the system at any
  given time from a QOS. If this limit is reached, new submission requests
  will be denied until previous jobs complete from this group.
</dd>

<dt id="qos_grptres"><b>GrpTRES</b>
<a class="slurm_link" href="#qos_grptres"></a></dt>
<dd>The total count of TRES able to be used at any given time from jobs
  running from a QOS. If this limit is reached, new jobs will be queued but
  only allowed to run after resources have been relinquished from this group.
</dd>

<dt id="qos_grptresmins"><b>GrpTRESMins</b>
<a class="slurm_link" href="#qos_grptresmins"></a></dt>
<dd>The total number of TRES minutes that can possibly be used by past,
  present and future jobs running from a QOS. If any limit is reached,
  all running jobs with that TRES in this group will be killed, and no new
  jobs will be allowed to run.  This usage is decayed (at a rate of
  PriorityDecayHalfLife).  It can also be reset (according to
  PriorityUsageResetPeriod) in order to allow jobs to run against the
  QOS again.  QOS that have the NoDecay flag set do not decay GrpTRESMins,
  see <a href="qos.html#qos_other">QOS Options</a> for details.
  The decay rate can be set per TRES with
  <a href="#tresdecayhalflife">TresDecayHalfLife</a>.
  This limit only applies when using the Priority Multifactor plugin.
</dd>

<dt id="qos_grptresrunmins"><b>GrpTRESRunMins</b>
<a class="slurm_link" href="#qos_grptresrunmins"></a></dt>
<dd>Used to limit the combined total number of TRES
  minutes used by all jobs running with a QOS.  This takes into
  consideration the time limit of running jobs and consumes it.
  If the limit is reached, no new jobs are started until other jobs
  finish to allow time to free up.
</dd>

<dt id="qos_grpwall"><b>GrpWall</b>
<a class="slurm_link" href="#qos_grpwall"></a></dt>
<dd>The maximum wall clock time running jobs are able
  to be allocated in aggregate for a QOS. If this limit is reached,
  future jobs in this QOS will be queued until they are able to run
  inside the limit. This usage is decayed (at a rate of
  PriorityDecayHalfLife).  It can also be reset (according to
  PriorityUsageResetPeriod) in order to allow jobs to run against the
  QOS again.  QOS that have the NoDecay flag set do not decay GrpWall.
  See <a href="qos.html#qos_other">QOS Options</a> for details.
</dd>

<dt id="qos_limitfactor"><b>LimitFactor</b>
<a class="slurm_link" href="#qos_limitfactor"></a></dt>
<dd>A float that is factored into an associations [Grp|Max]TRES limits.
  For example, if the LimitFactor is 2, then an association with a GrpTRES of
  30 CPUs would be allowed to allocate 60 CPUs when running under this QOS.

  <b>NOTE</b>: This factor is only applied to associations running in this
  QOS and is not applied to any limits in the QOS itself.
</dd>

<dt id="qos_maxjobsaccruepa"><b>MaxJobsAccruePerAccount</b>
<a class="slurm_link" href="#qos_maxjobsaccruepa"></a></dt>
<dd>The maximum number of pending jobs an
  account (or sub-account) can have accruing age priority at any given time.
  This limit does not determine if the job can run, it only limits the
  age factor of the priority.
</dd>

<dt id="qos_maxjobsaccruepu"><b>MaxJobsAccruePerUser</b>
<a class="slurm_link" href="#qos_maxjobsaccruepu"></a></dt>
<dd>The maximum number of pending jobs a
  user can have accruing age priority at any given time.
  This limit does not determine if the job can run, it only limits the
  age factor of the priority.
</dd>

<dt id="qos_maxjobspa"><b>MaxJobsPerAccount</b>
<a class="slurm_link" href="#qos_maxjobspa"></a></dt>
<dd>The maximum number of jobs an account (or sub-account) can have running at
  a given time.
</dd>

<dt id="qos_maxjobspu"><b>MaxJobsPerUser</b>
<a class="slurm_link" href="#qos_maxjobspu"></a></dt>
<dd>The maximum number of jobs a user can
  have running at a given time.
</dd>

<dt id="qos_maxsubmitjobspa"><b>MaxSubmitJobsPerAccount</b>
<a class="slurm_link" href="#qos_maxsubmitjobspa"></a></dt>
<dd>The maximum number of jobs an account (or sub-account) can have running and
  pending at a given time.
</dd>

<dt id="qos_maxsubmitjobspu"><b>MaxSubmitJobsPerUser</b>
<a class="slurm_link" href="#qos_maxsubmitjobspu"></a></dt>
<dd>The maximum number of jobs a user can
  have running and pending at a given time.
</dd>

<dt id="qos_maxtresminsperjob"><b>MaxTRESMinsPerJob</b>
<a class="slurm_link" href="#qos_maxtresminsperjob"></a></dt>
<dd>Maximum number of TRES minutes each job is able to use.
</dd>

<dt id="qos_maxtrespa"><b>MaxTRESPerAccount</b>
<a class="slurm_link" href="#qos_maxtrespa"></a></dt>
<dd>The maximum number of TRES an account can
  allocate at a given time.
</dd>

<dt id="qos_maxtrespj"><b>MaxTRESPerJob</b>
<a class="slurm_link" href="#qos_maxtrespj"></a></dt>
<dd>The maximum number of TRES each job is able to use.
</dd>

<dt id="qos_maxtrespn"><b>MaxTRESPerNode</b>
<a class="slurm_link" href="#qos_maxtrespn"></a></dt>
<dd>The maximum number of TRES each node in a job allocation can use.
</dd>

<dt id="qos_maxtrespu"><b>MaxTRESPerUser</b>
<a class="slurm_link" href="#qos_maxtrespu"></a></dt>
<dd>The maximum number of TRES a user can
  allocate at a given time.
</dd>

<dt id="qos_maxwalldurationpj"><b>MaxWallDurationPerJob</b>
<a class="slurm_link" href="#qos_maxwalldurationpj"></a></dt>
<dd>Maximum wall clock time each job is able to use. Format is &lt;min&gt;
  or &lt;min&gt;:&lt;sec&gt; or &lt;hr&gt;:&lt;min&gt;:&lt;sec&gt; or
  &lt;days&gt;-&lt;hr&gt;:&lt;min&gt;:&lt;sec&gt; or &lt;days&gt;-&lt;hr&gt;.
  The value is recorded in minutes with rounding as needed.
</dd>

<dt id="qos_minpriothreshold"><b>MinPrioThreshold</b>
<a class="slurm_link" href="#qos_minpriothreshold"></a></dt>
<dd>Minimum priority required to reserve resources when scheduling.
</dd>

<dt id="qos_mintresperjob"><b>MinTRESPerJob</b>
<a class="slurm_link" href="#qos_mintresperjob"></a></dt>
<dd>The minimum size in TRES any given job can
  have when using the requested QOS.
</dd>

<dt id="qos_usagefactor"><b>UsageFactor</b>
<a class="slurm_link" href="#qos_usagefactor"></a></dt>
<dd>A float that is factored into a job's TRES usage (e.g. RawUsage,
  TRESMins, TRESRunMins). For example, if the usagefactor was 2, for every
  TRESBillingUnit second a job ran it would count for 2. If the usagefactor
  was .5, every second would only count for half of the time.
  A setting of 0 would add no timed usage from the job.

  The usage factor only applies to the job's QOS and not the partition QOS.
  <br>
  If the UsageFactorSafe flag is set and AccountingStorageEnforce includes
  <i>Safe</i>, jobs will only be able to run if the job can run to completion
  with the UsageFactor applied, and won't be killed due to limits.
  <br>
  If the UsageFactorSafe flag is not set and AccountingStorageEnforce includes
  <i>Safe</i>, a job will be able to be scheduled without the UsageFactor
  applied and won't be killed due to limits.
  <br>
  If the UsageFactorSafe flag is not set and AccountingStorageEnforce does
  not include <i>Safe</i>, a job will be scheduled as long as the limits are
  not reached, but could be killed due to limits.
  <br>
  See <a href="slurm.conf.html#OPT_AccountingStorageEnforce">
  AccountingStorageEnforce</a> in the slurm.conf man page.
  <br>
  <b>NOTE</b>: A job's TRESRunMins usage is booked using the QOS UsageFactor in
  effect when the job starts, and is given back with that same factor when
  the job ends. Modifying a QOS's UsageFactor does not rescale the TRESRunMins
  usage of jobs that are already running; RawUsage and GrpWall continue to
  accrue with the current factor. However, when slurmctld is restarted or
  reconfigured, the usage of all running jobs is booked again with the
  UsageFactor in effect at that time.
</dd>
</dl>


<p>The <b>MaxNodes</b> and <b>MaxTime</b> options already exist in
Slurm's configuration on a per-partition basis, but the above options
provide the ability to impose limits on a per-user basis.  The
<b>MaxJobs</b> option provides an entirely new mechanism for Slurm to
control the workload any individual may place on a cluster in order to
achieve some balance between users.</p>

<p>When assigning limits to a QOS to use for a Partition QOS,
keep in mind that those limits are enforced at the QOS level, not
individually for each partition.  For example, if a QOS has a
<b>GrpTRES=cpu=20</b> limit defined and the QOS is assigned to two
unique partitions, users will be limited to 20 CPUs for the QOS
rather than being allowed 20 CPUs for each partition.</p>

<p>Fair-share scheduling is based upon the hierarchical account
data maintained in the Slurm database.  More information can be found
in the <a href="priority_multifactor.html">priority/multifactor</a>
plugin description.</p>

<h3 id="gres_limits">Specific limits over GRES
<a class="slurm_link" href="#gres_limits"></a>
</h3>
<p> When a GRES has a type associated with it and a limit is applied
  over this specific type (e.g. <i>MaxTRESPerUser=gres/gpu:tesla=1</i>) if a
  user requests a generic gres, the type's limit will not be enforced. In this
  situation an additional lua job submit plugin to check the user request may
  become useful. For example, if one requests <i>--gres=gpu:2</i> having a
  limit set of <i>MaxTRESPerUser=gres/gpu:tesla=1</i>, the limit won't be
  enforced so it will still be possible to get two teslas.
</p>
<p>
  This is due to a design limitation. The only way to enforce such a limit
  is to combine the specification of the limit with a job submit plugin that
  forces the user to always request a specific type model.
</p>
<p>
  An example of basic lua job submit plugin function could be:
</p>
<pre>
function slurm_job_submit(job_desc, part_list, submit_uid)
   gres_request = ""
   t = {job_desc.tres_per_job,
	job_desc.tres_per_socket,
	job_desc.tres_per_task,
	job_desc.tres_per_node}
   for k in pairs(t) do
	gres_request = gres_request .. t[k] .. ","
   end
   if (gres_request ~= nil)
   then
      for g in gres_request:gmatch("[^,]+")
      do
	 bad = string.match(g,'^gres/gpu[:=]*[0-9]*$')
	 if (bad ~= nil)
	 then
	    slurm.log_info("User specified gpu GRES without type: %s", bad)
	    slurm.user_msg("You must always specify a type when requesting gpu GRES")
	    return slurm.ERROR
	 end
      end
   end
end
</pre>
<p> Having this script and the limit in place will force the users to always
  specify a gpu with its type, thus enforcing the limits for each specific
  model.
</p>

<p>When <b>TRESBillingWeights</b> are defined for a partition, both typed and
non-typed resources should be included. For example, if you have 'tesla' GPUs
in one partition and you only define the billing weights for the 'tesla' typed
GPU resource, then those weights will not be applied to the generic GPUs.</p>

<p>It is also advisable to set <b>AccountingStorageTRES</b> for both generic
  and specific gres types, otherwise requests that ask for the generic instance
  of a gres won't be accounted for. For example, to track generic GPUs and
  Tesla GPUs, you would set this in your slurm.conf:
</p>
<pre>
  AccountingStorageTRES=gres/gpu,gres/gpu:tesla
</pre>

<p>
  See <a href="tres.html">Trackable Resources TRES</a> for details.
</p>

<h2 id="tresdecayhalflife">Controlling how GrpTRESMins usage decays
<a class="slurm_link" href="#tresdecayhalflife"></a>
</h2>

<p>Four settings affect whether a GrpTRESMins allocation replenishes, and they
do different jobs:</p>

<ul>
<li><b>PriorityDecayHalfLife</b> (slurm.conf, cluster wide) is the default for
  every usage counter: fairshare RawUsage, GrpWall, and GrpTRESMins.</li>
<li><b>TresDecayHalfLife</b> (per association or QOS, per TRES) overrides that
  half-life for GrpTRESMins only, and only for the TRES named. The value is
  <b>in seconds</b> &mdash; note that both GrpTRESMins and
  PriorityDecayHalfLife are expressed in minutes, so do not carry a value
  across without converting it. A longer half-life replenishes the quota more
  slowly, which is the reverse of most limits; set a TRES to 0 and its
  GrpTRESMins becomes an allocation that never replenishes. Use it when one
  resource should renew on a different schedule from the rest, or not at all,
  while fairshare keeps decaying normally. See
  <a href="sacctmgr.html">sacctmgr</a> for the full syntax.</li>
<li><b>NoDecay</b> (QOS flag) stops the whole QOS record decaying &mdash; its
  GrpTRESMins, GrpWall and RawUsage. Association (account and user) fairshare
  usage still decays normally, so job priority is unaffected. Use it when the
  whole QOS is a fixed allocation.</li>
<li><b>PriorityUsageResetPeriod</b> (slurm.conf, cluster wide) zeroes every
  counter on a calendar boundary instead of decaying it, for "N hours per
  month" schemes. It still resets a TRES whose TresDecayHalfLife is 0, so the
  two do not normally belong together.</li>
</ul>

<p>To give an account a CPU allocation that does not come back:</p>
<pre>
  sacctmgr modify account name=science set GrpTRESMins=cpu=100000
  sacctmgr modify account name=science set TresDecayHalfLife=cpu=0
</pre>

<p>Watch what has been used with
<span class="commandline">sshare -o Account,User,GrpTRESMins,GrpTRESRaw</span>,
and see the half-lives the controller is applying with
<span class="commandline">scontrol show assoc_mgr</span>. Once the allocation is
gone the account's pending jobs hold with reason
<i>AssocGrpCPUMinutesLimit</i>, and stay held until an administrator raises the
limit or clears the usage; jobs already running when the budget is crossed are
killed instead, unless <b>AccountingStorageEnforce</b> includes <b>Safe</b>.
Clearing the usage means <b>RawUsage=0</b>, which also
clears fairshare usage and every other TRES on that record, and for an account
every association beneath it; there is no per-TRES reset. None of this is
enforced unless <b>AccountingStorageEnforce</b> includes <b>limits</b> and
<b>PriorityType=priority/multifactor</b>.</p>

<h2 id="reasons">Job Reason Codes
<a class="slurm_link" href="#reasons"></a>
</h2>

<p>When a pending job is evaluated by the scheduler but found to exceed a
configured resource limit, a corresponding reason will be assigned to the job.
More details can be found on the <a href="job_reason_codes.html">Job Reason
Codes</a> page. More details about scheduling can be found in the
<a href="sched_config.html"> Scheduling Configuration Guide</a>.</p>

<!--#include virtual="footer.txt"-->