/usr/share/doc/slurm-wlm-doc/html/preempt.html

<!DOCTYPE html>
<html lang="en-US">
<head>
    <meta charset="UTF-8" />
	<meta name="viewport" content="width=device-width" />

	<title>Slurm Workload Manager</title>

	<link rel="shortcut icon" href="favicon.ico" />

	<link rel="stylesheet" type="text/css" href="fonts.css" />
	<link rel="stylesheet" type="text/css" href="reset.css" />
	<link rel="stylesheet" type="text/css" href="style.css" />
	<link rel="stylesheet" type="text/css" href="slurm.css" />

	<script src="jquery.min.js"></script>
</head>

<body>

<div class="container container--main">

	<header class="site-header" role="banner">

		<div class="site-masthead">
			<h1 class="site-masthead__title site-masthead__title--slurm">
				<a href="/" rel="home">
					<span class="slurm-logo">Slurm Workload Manager</span>
				</a>
			</h1>
			<div class="site-masthead__title">
				<a href="https://www.schedmd.com/" rel="home">
					<span class="site-logo">SchedMD</span>
				</a>
			</div>

			<button class="site-masthead__trigger menu-trigger" type="button" role="button" aria-label="Toggle Navigation"><span class="menu-trigger__lines"></span></button>
		</div>


		<nav class="site-nav">
			<h2 class="site-nav__title">Navigation</h2>

			<div class="slurm-title">
				<div class="slurm-logo"><a href="/">Slurm Workload Manager</a></div>
				<div class="slurm-title__version">Version 17.11</div>
			</div>

			<ul class="site-nav__menu site-menu menu" role="navigation">
				<li class="site-menu__item">
				        <div>About</div>
					<ul>
						<li><a href="overview.html">Overview</a></li>
						<li><a href="news.html">What's New</a></li>
						<li><a href="team.html">Slurm Team</a></li>
						<li><a href="meetings.html">Meetings</a></li>
						<li><a href="testimonials.html">Testimonials</a></li>
						<li><a href="disclaimer.html">Legal Notices</a></li>
					</ul>
				</li>
				<li class="site-menu__item">
					<div>Using</div>
					<ul>
						<li><a href="tutorials.html">Tutorials</a></li>
						<li><a href="documentation.html">Documentation</a></li>
						<li><a href="faq.html">FAQ</a></li>
						<li><a href="publications.html">Publications</a></li>
					</ul>
				</li>
				<li class="site-menu__item">
					<div>Installing</div>
					<ul>
						<li><a href="download.html">Download</a></li>
						<li><a href="quickstart_admin.html">Installation Guide</a></li>
					</ul>
				</li>
				<li class="site-menu__item">
					<div>Getting Help</div>
					<ul>
						<li><a href="https://www.schedmd.com/services.php">Support</a></li>
						<li><a href="mail.html">Mailing Lists</a></li>
						<li><a href="https://www.schedmd.com/services.php">Training</a></li>
						<li><a href="troubleshoot.html">Troubleshooting</a></li>
					</ul>
				</li>
			</ul>

		</nav>

	</header>

	<div class="content" role="main">
		<section class="slurm-search">
			<div class="container" id="cse">
				<gcse:search></gcse:search>
			</div>
		</section>

		<div class="section">
			<div class="container">


<H1>Preemption</H1>

<P>
Slurm supports job preemption, the act of stopping one or more "low-priority"
jobs to let a "high-priority" job run.
Job preemption is implemented as a variation of Slurm's
<a href="gang_scheduling.html">Gang Scheduling</a> logic.
When a high-priority job has been allocated resources that have already been
allocated to one or more low priority jobs, the low priority job(s) are
preempted.
The low priority job(s) can resume once the high priority job completes.
Alternately, the low priority job(s) can be requeued and started using other
resources if so configured in newer versions of Slurm.
</P>
<P>
The job's partition priority or its Quality Of Service (QOS) can be used to
identify the which jobs can preempt or be preempted by other jobs.
Slurm offers the ability to configure the preemption mechanism used on a per
partition or per QOS basis.
For example, jobs in a low priority queue may get requeued,
while jobs in a medium priority queue may get suspended.
</P>
<H2>Configuration</H2>
<P>
There are several important configuration parameters relating to preemption:
</P>
<UL>
<LI>
<B>SelectType</B>: Slurm job preemption logic supports
nodes allocated by the <I>select/linear</I> plugin,
socket/core/CPU resources allocated by the <I>select/cons_res</I> plugin or
<I>select/cray</I> for Cray systems without ALPS.
</LI>
<LI>
<B>SelectTypeParameter</B>: Since resources may be getting over-allocated
with jobs (suspended jobs remain in memory), the resource selection
plugin should be configured to track the amount of memory used by each job to
ensure that memory page swapping does not occur. When <I>select/linear</I> is
chosen, we recommend setting <I>SelectTypeParameter=CR_Memory</I>. When
<I>select/cons_res</I> is chosen, we recommend including Memory as a resource
(ex. <I>SelectTypeParameter=CR_Core_Memory</I>).
<BR><B>NOTE:</B> Unless <I>PreemptMode=SUSPEND,GANG</I> these memory management
parameters are not critical.
</LI>
<LI>
<B>DefMemPerCPU</B>: Since job requests may not explicitly specify
a memory requirement, we also recommend configuring
<I>DefMemPerCPU</I> (default memory per allocated CPU) or
<I>DefMemPerNode</I> (default memory per allocated node).
It may also be desirable to configure
<I>MaxMemPerCPU</I> (maximum memory per allocated CPU) or
<I>MaxMemPerNode</I> (maximum memory per allocated node) in <I>slurm.conf</I>.
Users can use the <I>--mem</I> or <I>--mem-per-cpu</I> option
at job submission time to specify their memory requirements.
<BR><B>NOTE:</B> Unless <I>PreemptMode=SUSPEND,GANG</I> these memory management
parameters are not critical.
</LI>
<LI>
<B>GraceTime</B>: Specifies a time period for a job to execute after
it is selected to be preempted. This option can be specified by partition or
QOS using the <I>slurm.conf</I> file or database respectively. This option is
only honored if <I>PreemptMode=CANCEL</I>. The <I>GraceTime</I> is specified in
seconds and the default value is zero, which results in no preemption delay.
Once a job has been selected for preemption, its end time is set to the
current time plus <I>GraceTime</I>. The job is immediately sent SIGCONT and
SIGTERM signals in order to provide notification of its imminent termination.
This is followed by the SIGCONT, SIGTERM and SIGKILL signal sequence upon
reaching its new end time.
</LI>
<LI>
<B>JobAcctGatherType and JobAcctGatherFrequency</B>: The "maximum data segment
size" and "maximum virtual memory size" system limits will be configured for
each job to ensure that the job does not exceed its requested amount of memory.
If you wish to enable additional enforcement of memory limits, configure job
accounting with the <I>JobAcctGatherType</I> and <I>JobAcctGatherFrequency</I>
parameters. When accounting is enabled and a job exceeds its configured memory
limits, it will be canceled in order to prevent it from adversely effecting
other jobs sharing the same resources.
<BR><B>NOTE:</B> Unless <I>PreemptMode=SUSPEND,GANG</I> these memory management
parameters are not critical.
</LI>
<LI>
<B>PreemptMode</B>: Specifies the mechanism used to preempt low priority jobs.
The PreemptMode should be specified for the cluster as a whole, although
different values can be configured on each partition when
<I>PreemptType=preempt/partition_prio</I>. Note that when specified 
on a partition, a compatible mode must also be specified system-wide;
specifically if a PreemptMode is set to SUSPEND for any partition, then
the system-wide PreemptMode must include the GANG and SUSPEND parameters so the
module responsible for resuming jobs executes.
Configure to <I>CANCEL</I>, <I>CHECKPOINT</I>,
<I>SUSPEND</I> or <I>REQUEUE</I> depending on the desired action for low
priority jobs.
The <I>GANG</I> option must also be specified if gang scheduling is desired
or a PreemptMode of <I>SUSPEND</I> is used for any jobs.
<UL>
<LI>A value of <I>CANCEL</I> will always cancel the job.</LI>
<LI>A value of <I>CHECKPOINT</I> will checkpoint (if possible) or kill low
priority jobs.</LI>
Checkpointed jobs are not automatically restarted.
<LI>A value of <I>REQUEUE</I> will requeue (if possible) or kill low priority
jobs. Requeued jobs are permitted to be restarted on different resources.</LI>
<LI>A value of <I>SUSPEND</I> will suspend and resume jobs.
If PreemptType=preempt/partition_prio is configured then a value of
<I>SUSPEND</I> will suspend and automatically resume the low priority jobs.
If PreemptType=preempt/qos is configured, then the jobs sharing resources
will always time slice rather than one job remaining suspended.
The <I>SUSPEND</I> option must be used with the <I>GANG</I>
option (e.g. "PreemptMode=SUSPEND,GANG").</LI>
<LI>A value of <I>GANG</I> may be used with any of the above values and will
execute a module responsible for resuming jobs previously suspended for either
gang scheduling or job preemption with suspension.</LI>
</UL>
</LI>
<LI>
<B>PreemptType</B>: Configure to the desired mechanism used to identify
which jobs can preempt other jobs.
<UL>
<LI><I>preempt/none</I> indicates that jobs will not preempt each other
(default).</LI>
<LI><I>preempt/partition_prio</I> indicates that jobs from one partition
can preempt jobs from lower priority partitions.</LI>
<LI><I>preempt/qos</I> indicates that jobs from one Quality Of Service (QOS)
can preempt jobs from a lower QOS. These jobs can be in the same partition
or different partitions. PreemptMode must be set to CANCEL, CHECKPOINT,
REQUEUE or SUSPEND. This option requires the use of a database identifying
available QOS and their preemption rules. This option is not compatible with
PreemptMode=OFF and PreemptMode=SUSPEND is only supported by the
select/cons_res plugin.</LI>
</UL>
</LI>
<LI>
<B>PriorityTier</B>: Configure the partition's <I>PriorityTier</I> setting
relative to other partitions to control the preemptive behavior when
<I>PreemptType=preempt/partition_prio</I>.
This option is not relevant if <I>PreemptType=preempt/qos</I>.
If two jobs from two
different partitions are allocated to the same resources, the job in the
partition with the greater <I>PriorityTier</I> value will preempt the job in the
partition with the lesser <I>PriorityTier</I> value. If the <I>PriorityTier</I>
values of the two partitions are equal then no preemption will occur. The
default <I>PriorityTier</I> value is 1.
</LI>
<LI>
<B>OverSubscribe</B>: Configure the partition's <I>OverSubscribe</I> setting to
<I>FORCE</I> for all partitions in which job preemption using a suspend/resume
mechanism is used or <I>NO</I> otherwise.
The <I>FORCE</I> option supports an additional parameter that controls
how many jobs can oversubscribe a compute resource (FORCE[:max_share]). By default the
max_share value is 4. In order to preempt jobs (and not gang schedule them),
always set max_share to 1. To allow up to 2 jobs from this partition to be
allocated to a common resource (and gang scheduled), set
<I>OverSubscribe=FORCE:2</I>.
NOTE: <I>PreemptType=QOS</I> will permit one additional job to be run
on the partition if started due to job preemption. For example, a configuration
of <I>OverSubscribe=FORCE:1</I> will only permit one job per resources normally,
but a second job can be started if done so through preemption based upon QOS.
The use of <I>PreemptType=QOS</I> and <I>PreemptType=Suspend</I> only applies
with <I>SelectType=cons_res</I>.
</LI>
</UL>
<P>
To enable preemption after making the configuration changes described above,
restart Slurm if it is already running. Any change to the plugin settings in
Slurm requires a full restart of the daemons. If you just change the partition
<I>PriorityTier</I> or <I>OverSubscribe</I> setting, this can be updated with
<I>scontrol reconfig</I>.
</P>

<P>
If a job request restricts Slurm's ability to run jobs from multiple users or
accounts on a node by using the "--exclusive=user" or "--exclusive=mcs" job
options, that may prevent preemption of jobs to start higher priority jobs. If
preemption is used, it is generally advisable to disable the "--exclusive=user"
and "--exclusive=mcs" job options by using a job_submit plugin (set the value of
"shared" to "NO_VAL16").
</P>

<H2>Preemption Design and Operation</H2>

<P>
The select plugin will identify resources where a pending job can begin
execution.
When <I>PreemptMode</I> is configured to CANCEL, CHECKPOINT, SUSPEND or
REQUEUE, the select plugin will also preempt running jobs as needed to
initiate the pending job.
When <I>PreemptMode=SUSPEND,GANG</I> the select plugin will initiate the
pending job and rely upon the gang scheduling logic to perform job suspend
and resume as described below.
</P>
<P>
The select plugin is passed an ordered list of preemptable jobs to consider for
each pending job which is a candidate to start.
This list is sorted by either:
</P>
<ol>
<li>QOS priority,</li>
<li>Partition priority and job size (to favor preempting smaller jobs), or</li>
<li>Job start time (with <I>SchedulerParameters=preempt_youngest_first</I>).</li>
</ol>
<P>
The select plugin will determine if the pending job can start without preempting
any jobs and if so, starts the job using available resources.
Otherwise, the select plugin will simulate the preemption of each job in the
priority ordered list and test if the job can be started after each preemption.
Once the job can be started, the higher priority jobs in the preemption queue
will not be considered, but the jobs to be preempted in the original list may
be sub-optimal.
For example, to start an 8 node job, the ordered preemption candidates may be
2 node, 4 node and 8 node.
Preempting all three jobs would allow the pending job to start, but by reording
the preemption candidates it is possible to start the pending job after
preempting only one job.
To address this issue, the preemption candidates are re-ordered with the final
job requiring preemption placed first in the list and all of the other jobs
to be preempted ordered by the number of nodes in their allocation which overlap
the resources selected for the pending job.
In the example above, the 8 node job would be moved to the first position in
the list.
The process of simulating the preemption of each job in the priority ordered
list will then be repeated for the final decision of which jobs to preempt.
This two stage process may preempt jobs which are not strictly in preemption
priority order, but fewer jobs will be preempted than otherwise required.
See the SchedulerParameters configuration parameter options of
<I>preempt_reorder_count</I> and <I>preempt_strict_order</I> for preemption
tuning parameters.
</P>
<P>
When enabled, the gang scheduling logic (which is also supports job
preemption) keeps track of the resources allocated to all jobs.
For each partition an "active bitmap" is maintained that tracks all
concurrently running jobs in the Slurm cluster.
Each partition also maintains a job list for that partition, and a list of
"shadow" jobs.
The "shadow" jobs are high priority job allocations that "cast shadows" on the
active bitmaps of the low priority jobs.
Jobs caught in these "shadows" will be preempted.
</P>
<P>
Each time a new job is allocated to resources in a partition and begins
running, the gang scheduler adds a "shadow" of this job to all lower priority
partitions.
The active bitmap of these lower priority partitions are then rebuilt, with the shadow jobs added first.
Any existing jobs that were replaced by one or more "shadow" jobs are
suspended (preempted). Conversely, when a high priority running job completes,
it's "shadow" goes away and the active bitmaps of the lower priority
partitions are rebuilt to see if any suspended jobs can be resumed.
</P>
<P>
The gang scheduler plugin is designed to be <I>reactive</I> to the resource
allocation decisions made by the "select" plugins.
The "select" plugins have been enhanced to recognize when job preemption has
been configured, and to factor in the priority of each partition when selecting resources for a job.
When choosing resources for each job, the selector avoids resources that are
in use by other jobs (unless sharing has been configured, in which case it
does some load-balancing).
However, when job preemption is enabled, the select plugins may choose
resources that are already in use by jobs from partitions with a lower
priority setting, even when sharing is disabled in those partitions.
</P>
<P>
This leaves the gang scheduler in charge of controlling which jobs should run
on the over-allocated resources.
If <I>PreemptMode=SUSPEND</I>, jobs are suspended using the same internal
functions that support <I>scontrol suspend</I> and <I>scontrol resume</I>.
A good way to observe the operation of the gang scheduler is by running
<I>squeue -i&lt;time&gt;</I> in a terminal window.
</P>

<H2>A Simple Example</H2>

<P>
The following example is configured with <I>select/linear</I> and
<I>PreemptMode=SUSPEND,GANG</I>.
This example takes place on a cluster of 5 nodes:
</P>
<PRE>
[user@n16 ~]$ <B>sinfo</B>
PARTITION AVAIL  TIMELIMIT NODES  STATE NODELIST
active*      up   infinite     5   idle n[12-16]
hipri        up   infinite     5   idle n[12-16]
</PRE>
<P>
Here are the Partition settings:
</P>
<PRE>
[user@n16 ~]$ <B>grep PartitionName /shared/slurm/slurm.conf</B>
PartitionName=DEFAULT OverSubscribe=FORCE:1 Nodes=n[12-16]
PartitionName=active PriorityTier=1 Default=YES
PartitionName=hipri  PriorityTier=2
</PRE>
<P>
The <I>runit.pl</I> script launches a simple load-generating app that runs
for the given number of seconds. Submit 5 single-node <I>runit.pl</I> jobs to
run on all nodes:
</P>
<PRE>
[user@n16 ~]$ <B>sbatch -N1 ./runit.pl 300</B>
sbatch: Submitted batch job 485
[user@n16 ~]$ <B>sbatch -N1 ./runit.pl 300</B>
sbatch: Submitted batch job 486
[user@n16 ~]$ <B>sbatch -N1 ./runit.pl 300</B>
sbatch: Submitted batch job 487
[user@n16 ~]$ <B>sbatch -N1 ./runit.pl 300</B>
sbatch: Submitted batch job 488
[user@n16 ~]$ <B>sbatch -N1 ./runit.pl 300</B>
sbatch: Submitted batch job 489
[user@n16 ~]$ <B>squeue -Si</B>
JOBID PARTITION     NAME   USER  ST   TIME  NODES NODELIST
  485    active runit.pl   user   R   0:06      1 n12
  486    active runit.pl   user   R   0:06      1 n13
  487    active runit.pl   user   R   0:05      1 n14
  488    active runit.pl   user   R   0:05      1 n15
  489    active runit.pl   user   R   0:04      1 n16
</PRE>
<P>
Now submit a short-running 3-node job to the <I>hipri</I> partition:
</P>
<PRE>
[user@n16 ~]$ <B>sbatch -N3 -p hipri ./runit.pl 30</B>
sbatch: Submitted batch job 490
[user@n16 ~]$ <B>squeue -Si</B>
JOBID PARTITION     NAME   USER  ST   TIME  NODES NODELIST
  485    active runit.pl   user   S   0:27      1 n12
  486    active runit.pl   user   S   0:27      1 n13
  487    active runit.pl   user   S   0:26      1 n14
  488    active runit.pl   user   R   0:29      1 n15
  489    active runit.pl   user   R   0:28      1 n16
  490     hipri runit.pl   user   R   0:03      3 n[12-14]
</PRE>
<P>
Job 490 in the <I>hipri</I> partition preempted jobs 485, 486, and 487 from
the <I>active</I> partition. Jobs 488 and 489 in the <I>active</I> partition
remained running.
</P>
<P>
This state persisted until job 490 completed, at which point the preempted jobs
were resumed:
</P>
<PRE>
[user@n16 ~]$ <B>squeue</B>
JOBID PARTITION     NAME   USER  ST   TIME  NODES NODELIST
  485    active runit.pl   user   R   0:30      1 n12
  486    active runit.pl   user   R   0:30      1 n13
  487    active runit.pl   user   R   0:29      1 n14
  488    active runit.pl   user   R   0:59      1 n15
  489    active runit.pl   user   R   0:58      1 n16
</PRE>

<H2>Another Example</H2>
<P>
In this example we have three different partitions using three different
job preemption mechanisms.
</P>
<PRE>
# Excerpt from slurm.conf
PartitionName=low Nodes=linux Default=YES OverSubscribe=NO      PriorityTier=10 PreemptMode=requeue
PartitionName=med Nodes=linux Default=NO  OverSubscribe=FORCE:1 PriorityTier=20 PreemptMode=suspend
PartitionName=hi  Nodes=linux Default=NO  OverSubscribe=FORCE:1 PriorityTier=30 PreemptMode=off
</PRE>
<PRE>
$ sbatch tmp
Submitted batch job 94
$ sbatch -p med tmp
Submitted batch job 95
$ sbatch -p hi tmp
Submitted batch job 96
$ squeue
  JOBID PARTITION     NAME     USER  ST       TIME  NODES NODELIST(REASON)
     96        hi      tmp      moe   R       0:04      1 linux
     94       low      tmp      moe  PD       0:00      1 (Resources)
     95       med      tmp      moe   S       0:02      1 linux
(after job 96 completes)
$ squeue
  JOBID PARTITION     NAME     USER  ST       TIME  NODES NODELIST(REASON)
     94       low      tmp      moe  PD       0:00      1 (Resources)
     95       med      tmp      moe   R       0:24      1 linux
</PRE>

<H2>Another Example</H2>
<P>
In this example we have one partition on which we want to execute only one
job per resource (e.g. core) at a time except when a job submitted to the
partition from a high priority Quality Of Service (QOS) is submitted. In that
case, we want that second high priority job to be started and be gang scheduled
with the other jobs on overlapping resources.
</P>
<PRE>
# Excerpt from slurm.conf
PreemptMode=Suspend,Gang
PreemptType=preempt/qos
PartitionName=normal Nodes=linux Default=NO  OverSubscribe=FORCE:1
</PRE>

<H2><A NAME="future_work">Future Ideas</A></H2>

<P>
<B>More intelligence in the select plugins</B>: This implementation of
preemption relies on intelligent job placement by the <I>select</I> plugins.
</P><P>
Take the following example:
</P>
<PRE>
[user@n8 ~]$ <B>sinfo</B>
PARTITION AVAIL  TIMELIMIT NODES  STATE NODELIST
active*      up   infinite     5   idle n[1-5]
hipri        up   infinite     5   idle n[1-5]
[user@n8 ~]$ <B>sbatch -N1 -n2 ./sleepme 60</B>
sbatch: Submitted batch job 17
[user@n8 ~]$ <B>sbatch -N1 -n2 ./sleepme 60</B>
sbatch: Submitted batch job 18
[user@n8 ~]$ <B>sbatch -N1 -n2 ./sleepme 60</B>
sbatch: Submitted batch job 19
[user@n8 ~]$ <B>squeue</B>
  JOBID PARTITION     NAME     USER  ST       TIME  NODES NODELIST(REASON)
     17    active  sleepme  cholmes   R       0:03      1 n1
     18    active  sleepme  cholmes   R       0:03      1 n2
     19    active  sleepme  cholmes   R       0:02      1 n3
[user@n8 ~]$ <B>sbatch -N3 -n6 -p hipri ./sleepme 20</B>
sbatch: Submitted batch job 20
[user@n8 ~]$ <B>squeue -Si</B>
  JOBID PARTITION     NAME     USER  ST       TIME  NODES NODELIST(REASON)
     17    active  sleepme  cholmes   S       0:16      1 n1
     18    active  sleepme  cholmes   S       0:16      1 n2
     19    active  sleepme  cholmes   S       0:15      1 n3
     20     hipri  sleepme  cholmes   R       0:03      3 n[1-3]
[user@n8 ~]$ <B>sinfo</B>
PARTITION AVAIL  TIMELIMIT NODES  STATE NODELIST
active*      up   infinite     3  alloc n[1-3]
active*      up   infinite     2   idle n[4-5]
hipri        up   infinite     3  alloc n[1-3]
hipri        up   infinite     2   idle n[4-5]
</PRE>
<P>
It would be more ideal if the "hipri" job were placed on nodes n[3-5], which
would allow jobs 17 and 18 to continue running. However, a new "intelligent"
algorithm would have to include factors such as job size and required nodes in
order to support ideal placements such as this, which can quickly complicate
the design. Any and all help is welcome here!
</P>

<p style="text-align:center;">Last modified 20 September 2017</p>

			</div> <!-- END .container -->
		</div> <!-- END .section -->
	</div> <!-- END .content -->
</div> <!-- END .main -->

<footer class="site-footer" role="contentinfo">
	<nav class="footer-nav section">
		<div class="container">
			<p><a href="disclaimer.html" target="_blank" class="privacy">Legal Notices</a></p>
		</div>
	</nav>
</footer>

</body>
</html>
slurm-wlm-doc 17.11.2-1build1 / usr / share / doc / slurm-wlm-doc / html / preempt.html
