Moab Workload Manager

Appendix F: Moab Parameters

See the Parameters Overview in the Moab Admin Manual for further information about specifying parameters.

Index: A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

ACCOUNTCFG[<ACCOUNTID>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
General Credential Flags, CHARGERATE, PRIORITY, ENABLEPROFILING, MEMBERULIST, TRIGGER, PDEF, PLIST, QDEF, QLIST, usage limit, or a fairness usage limit specification (FSCAP, FSTARGET, and FSWEIGHT).
---
Specifies account specific attributes. See the account overview for general information and the job flag overview for a description of legal flag values.
ACCOUNTCFG[projectX] MAXJOB=50 QDEF=highprio

Up to 50 jobs submitted under the account ID projectX will be allowed to execute simultaneously and will be assigned the QOS highprio by default.

   
ACCOUNTINGINTERFACEURL
<URL> where protocol can be one of exec or file
---
Specifies the interface to use for real-time export of Moab accounting/auditing information. See Exporting Events in Real-Time for more information.
ACCOUNTINGINTERFACEURL exec:///$TOOLSDIR/dumpacc.pl
   
ACCOUNTWEIGHT
<INTEGER>
1
Specifies the priority weight to be applied to the specified account priority. See Credential (CRED) Factor.
ACCOUNTWEIGHT 100
   
ADMIN1, ADMIN2, ADMIN3
Space delimited list of user names
root
Deprecated. Use ADMINCFG. Users listed under the parameter ADMIN1 are allowed to perform any scheduling function. They have full control over the scheduler and access to all data. The first user listed in the ADMIN1 user list is considered to be the 'primary admin' and is the ID under which Moab must be started and run.  Valid values include user names or the keyword 'ALL'. Again, these parameters are deprecated; use ADMINCFG.
ADMIN1 moabuser steve scott jenny
All users listed have full access to Moab control commands and Moab data. Moab must be started by and run under the 'moabuser' user id since moabuser is the primary admin.

   
ADMINCFG[X]
One or more <ATTR>=<VALUE> pairs where <ATTR> is one of the following: ENABLEPROXY, USERS, SERVICES, or NAME
---
Allows a site to configure which services and users belong to a particular level of administration. Note: The first user listed in the ADMINCFG[1] users list is considered to be the 'primary admin'.
ADMINCFG[1] USERS=root,john 
ADMINCFG[1] SERVICES=ALL 
ADMINCFG[1] NAME=batchadmin

ADMINCFG[3] USERS=bob,carol,smoore 
ADMINCFG[3] SERVICES=mjobctl:cancel:adjustprio:adjusthold,mcredctl,runjob
ADMINCFG[3] NAME=helpdesk
Members of the batchadmin admin role are allowed to run all commands. Members of the helpdesk role are allowed to cancel jobs, adjust job priority, and adjust job holds. They are also able to view and modify credential objects (ie, users, groups, accounts, etc) See the security overview for more details.
   
ADMINHOSTS
Space delimited list of host names.
---
If specified, the ADMINHOSTS parameter allows a site to specify a subset of trusted hosts. All administrative commands (level 1-3) will be rejected unless they are received from one of the hosts listed.
ADMINHOSTS hostA hostB
   
AGGREGATENODEACTIONS
<BOOLEAN>
FALSE
Consolidates queued node actions into as few actions as possible to reduce communication burden with resource manager. Node actions are queued until the AGGREGATENODEACTIONSTIME setting.
Note This may delay some node actions. When reprovisioning, the system job may expire before the provision action occurs; while the action will still occur, the job will not show it.
AGGREGATENODEACTIONS TRUE

Queues node actions together when possible.

   
AGGREGATENODEACTIONSTIME
<SECONDS>
60
The delay time for the AGGREGATENODEACTIONS parameter to aggregate requests before sending job batches.
AGGREGATENODEACTIONSTIME 120

Sets the AGGREGATENODEACTIONS delay to two minutes..

   
ALLOWROOTJOBS
<BOOLEAN>
FALSE
Specifies whether batch jobs from the root user (UID=0) are allowed to be execute. Note: The resource manager must also support root jobs.
ALLOWROOTJOBS TRUE

Jobs from the root user can execute.

   
ALLOWVMMIGRATION
<BOOLEAN>
FALSE
Enables Moab to migrate VMs.
ALLOWVMMIGRATION TRUE
   
ALWAYSEVALUATEALLJOBS
<BOOLEAN>
FALSE
When scheduling priority jobs, Moab stops scheduling when it encounters the first job that cannot run and cannot get a reservation. ALWAYSEVALUATEALLJOBS directs Moab to continue scheduling until all priority jobs (jobs that do not violate any limits) are evaluated.
ALWAYSEVALUATEALLJOBS TRUE
   
AMCFG
One or more key-value pairs as described in the Allocation Manager Configuration Overview.
N/A
Specifies the interface and policy configuration for the scheduler-allocation manager interface. Described in detail in the Allocation Manager Configuration Overview.
AMCFG[bank] SERVER=gold://master.ufl.edu JOBFAILUREACTION=IGNORE TIMEOUT=15
   
APPLICATIONLIST
Space delimited list of generic resources.
N/A
Specifies which generic resources represent actual applications on the cluster/grid. See 12.4 Managing Consumable Generic Resources for more information.
NODECFG[node01] GRES=calclab:1,powerhouse:1 RCSOFTWARE=calclab:1,powerhouse:1
NODECFG[node02] GRES=calclab:1,powerhouse:1 RCSOFTWARE=calclab:1,powerhouse:1
APPLICATIONLIST calclab,powerhouse

The generic resources 'calclab' and 'powerhouse' will now be recognized and treated as application software.

   
ASSIGNVLANFEATURES
<BOOLEAN>
FALSE
When set to TRUE, this forces all VMs to be contained in VLANs.
ASSIGNVLANFEATURES TRUE
If a VLAN is not requested when a VM is created, the VM is assigned to one.
   
ATTRATTRWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to jobs with the specified job attribute. See Attribute (ATTR) Factor.
ATTRATTRWEIGHT 100
   
ATTRGRESWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to jobs requesting the specified generic resource. See Attribute (ATTR) Factor.
ATTRGRESWEIGHT 200
   
ATTRSTATEWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to jobs with the specified job state. See Attribute (ATTR) Factor.
ATTRSTATEWEIGHT 200
   
ATTRWEIGHT
<INTEGER>
1
Specifies the priority component weight to be applied to the ATTR subcomponents. See Attribute (ATTR) Factor.
ATTRWEIGHT      2
ATTRSTATEWEIGHT 200
   
BACKFILLDEPTH
<INTEGER>
0 (no limit)
Specifies the number of idle jobs to evaluate for backfill. The backfill algorithm will evaluate the top <X> priority jobs for scheduling. By default, all jobs are evaluated.
BACKFILLDEPTH 128
Evaluate only the top 128 highest priority idle jobs for consideration for backfill.
   
BACKFILLMETRIC
One of the following: PROCS, PROCSECONDS, SECONDS, or NODES
PROCS
Specifies the criteria used by the backfill algorithm to determine the 'best' jobs to backfill. Only applicable when using BESTFIT or GREEDY backfill algorithms.
BACKFILLMETRIC PROCSECONDS
   
BACKFILLPOLICY
One of the following: FIRSTFIT, BESTFIT, GREEDY, PREEMPT, or NONE
FIRSTFIT
Specifies which backfill algorithm will be used. See Configuring Backfill for more information.
BACKFILLPOLICY  BESTFIT
   
BFCHUNKDURATION
[[[DD:]HH:]MM:]SS
0 (chunking disabled)
Specifies the duration during which freed resources will be aggregated for use by larger jobs. Used in conjunction with BFCHUNKSIZE. See Configuring Backfill for more information.
BFCHUNKDURATION 00:05:00
BFCHUNKSIZE     4
Aggregate backfillable resources for up to 5 minutes, making resources available only to jobs of size 4 or larger.
   
BFCHUNKSIZE
<INTEGER>
0 (chunking disabled)
Specifies the minimum job size which can utilize chunked resources. Used in conjunction with BFCHUNKDURATION. See Configuring Backfill for more information.
BFCHUNKDURATION 00:05:00
BFCHUNKSIZE     4
Aggregate backfillable resources for up to 5 minutes, making resources available only to jobs of size 4 or larger.
   
BFMINVIRTUALWALLTIME
[[[DD:]HH:]MM:]SS
---

Specifies the minimum job wallclock time for virtual scaling (optimistic-like backfilling.) Any job with a wallclock time less than this setting will not be virtually scaled. The value specified relates to a job's original walltime and not its virtually-scaled walltime.

BFMINVIRTUALWALLTIME  00:01:30
   
BFPRIORITYPOLICY
One of RANDOM, DURATION, or HWDURATION
---
Specifies policy to use when prioritizing backfill jobs for preemption
BFPRIORITYPOLICY  DURATION

Use length of job in determining which backfill job to preempt.

   
BFVIRTUALWALLTIMECONFLICTPOLICY
One of the following: PREEMPT
---
Specifies how to handle scheduling conflicts when a virtually scaled job "expands" to its original wallclock time. This occurs when the job is within one scheduling iteration - RMPOLLINTERVAL - of its virtually scaled wallclock time expiring.
BFVIRTUALWALLTIMECONFLICTPOLICY  PREEMPT
   
BFVIRTUALWALLTIMESCALINGFACTOR
<DOUBLE>
0 (virtual scaling disabled)
Specifies the factor by which eligible jobs' wallclock time is virtually scaled (optimistic-like backfilling).
BFVIRTUALWALLTIMESCALINGFACTOR .4
   
BLOCKLIST
DEPEND
NONE
Specifies the additional non-default criteria which are used to determine if a job should be filtered out before idle usage is updated and idle usage policies are enforced.
BLOCKLIST DEPEND
   
BYPASSCAP
<INTEGER>
0
Specifies the max weighted value allowed from the bypass count subfactor when determining a job's priority (see Priority Factors for more information).
BYPASSWEIGHT 5000
BYPASSCAP    30000
   
BYPASSWEIGHT
<INTEGER>
0
Specifies the weight to be applied to a job's backfill bypass count when determining a job's priority (see Priority Factors for more information).
BYPASSWEIGHT 5000
   
CHECKPOINTDIR
<STRING>
---
Specifies the directory for temporary job checkpoint files (usually of the form "jobid.cp"). This is not the directory for Moab's checkpoint file (.moab.ck).
CHECKPOINTDIR  /tmp/moabcheckpoint
   
CHECKPOINTEXPIRATIONTIME
[[[DD:]HH:]MM:]SS or UNLIMITED
3,000,000 seconds
Specifies how 'stale' checkpoint data can be before it is ignored and purged.
CHECKPOINTEXPIRATIONTIME 1:00:00:00

Expire checkpoint data which has been stale for over one day.

   
CHECKPOINTFILE
<STRING>
moab.ck
Name (absolute or relative) of the Moab checkpoint file.
CHECKPOINTFILE /var/adm/moab/moab.ck
Maintain the Moab checkpoint file in the file specified.
   
CHECKPOINTINTERVAL
[[[DD:]HH:]MM:]SS
00:05:00
Time between automatic Moab checkpoints.
CHECKPOINTINTERVAL 00:15:00
Moab should checkpoint state information every 15 minutes.
   
CHECKPOINTWITHDATABASE
<BOOLEAN>
FALSE
If set to TRUE, Moab stores checkpoint information to a database rather than to the .moab.ck flat text file.
CHECKPOINTWITHDATABASE    TRUE
   
CHILDSTDERRCHECK
<BOOLEAN>
FALSE
If set to TRUE, child processes Moab executes are considered failed if their standard error stream contains the text "ERROR".
CHILDSTDERRCHECK    TRUE
   
CLASSCFG[<CLASSID>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
General Credential Flags, DEFAULT.ATTR, DEFAULT.DISK, DEFAULT.FEATURES, DEFAULT.GRES, DEFAULT.MEM, DEFAULT.NODESET, DEFAULT.PROC, ENABLEPROFILING, EXCL.FEATURES, EXCLUDEUSERLIST, HOSTLIST, JOBEPILOG, JOBPROLOG, JOBTRIGGER, MAXPROCPERNODE, MAX.NODE, MAX.PROC, MAXPROC, MAX.WCLIMIT, MIN.NODE, MIN.PROC, MIN.TPN, MIN.WCLIMIT, PARTITION, PRIORITY, PRIORITYF, QDEF, QLIST, REQ.FEATURES, REQUIREDACCOUNTLIST, REQUIREDUSERLIST, RESFAILPOLICY, SYSPRIO, TRIGGER, WCOVERRUN, usage limit, or fairshare usage limit specification.
---
Specifies class specific attributes (see Credential Overview for details).
CLASSCFG[batch] MAXJOB=50 QDEF=highprio
Up to 50 jobs submitted to the class batch will be allowed to execute simultaneously and will be assigned the QOS highprio by default.
   
CLASSWEIGHT
<INTEGER>
1
Specifies the weight to be applied to the class priority of each job (see Credential (CRED) Factor and credential priority).
CLASSWEIGHT 10
   
CLIENTCFG[<X>]
One or more of <ATTR>-<VALUE> pairs where <X> indicates the specified peer and <ATTR> is one of the following: AUTH, AUTHCMD, AUTHTYPE, HOST, KEY, or DEFAULTSUBMITPARTITION.
---
Specifies the shared secret key and authentication method which Moab will use to communicate with the named peer daemon. See Security Overview for more information. Note: The AUTHTYPE and KEY attributes of this parameter may only be specified in the moab-private.cfg config file.
CLIENTCFG[silverB] KEY=apple7 AUTH=admin1
Moab will use the session key apple7 for peer authentication and for encrypting and decrypting messages sent from silverB. Also, client connections from this interface will be authorized at an admin 1 level.
   
CLIENTMAXCONNECTIONS
<INTEGER>
128
Changes the maximum number of connections that can simultaneously connect to Moab. The value can be increased during runtime, but it cannot be decreased. The value cannot be reduced below the default value of 128.
CLIENTMAXCONNECTIONS 256
Doubles the maximum number of connections.
   
CLIENTMAXPRIMARYRETRY
<integer> or INFINITY
1
Specifies how many times the client will attempt to retry its connection to the primary server if Moab is not available.
CLIENTMAXPRIMARYRETRY 5
CLIENTMAXPRIMARYRETRYTIMEOUT 1000
The client will attempt to retry its connection to the primary server 5 times with 1 second intervals before giving up. Note: If INFINITY is specified, Moab will attempt 2140000000 times.
   
CLIENTMAXPRIMARYRETRYTIMEOUT
<integer> (milliseconds)
2 seconds
Specifies how much time to wait until the client will attempt to retry its connection to the primary server if Moab is not available.
CLIENTMAXPRIMARYRETRY 3
CLIENTMAXPRIMARYRETRYTIMEOUT 500
The client will attempt to retry its connection to the primary server 3 times with 1/2 second intervals before giving up.
   
CLIENTTIMEOUT
[[[DD:]HH:]MM:]SS
00:00:30
Time which Moab client commands will wait for a response from the Moab server. See Client Configuration for more information. Note: May also be specified as an environment variable.
CLIENTTIMEOUT 00:15:00

Moab clients will wait up to 15 minutes for a response from the server before timing out.

   
CREDDISCOVERY
TRUE
FALSE
Specifies that Moab should create otherwise unknown credentials when it discovers them in the statistics files.
CREDDISCOVERY TRUE
   
CREDWEIGHT
<INTEGER>
1
Specifies the credential component weight associated with the credential priority. See Credential (CRED) Factor for more information.
CREDWEIGHT 2
   
DATASTAGEHOLDTYPE
Any Job Hold type
DEFER
Specifies what to do if a job's data staging operations fail.
DATASTAGEHOLDTYPE  BATCH
   
DEADLINEPOLICY
One of CANCEL, HOLD, IGNORE, or RETRY
NONE
Specifies what to do when a requested deadline cannot be reached (see Job Deadlines).
DEADLINEPOLICY  IGNORE
   
DEFAULTCLASSLIST
Space delimited list of one or more <STRING>'s.
---
Specifies the default classes supported on each node for RM systems which do not provide this information.
DEFAULTCLASSLIST serial parallel
   
DEFAULTSUBMITLANGUAGE
One of LSF, PBS, LL or SLURM.
PBS
Specifies the default job language to use when interpreting commandline arguments and command file scripts associated with the msub command.
DEFAULTSUBMITLANGUAGE LSF
   
DEFAULTSUBMITPARTITION
See parameter CLIENTCFG[].
---
If a user submits a job using msub which does not specify host, feature, or partition constraints, then the msub client will insert the specified default submit partition into the newly submitted job as a hard requirement.
CLIENTCFG[DEFAULT] DEFAULTSUBMITPARTITION=partition1
   
DEFERCOUNT
<INTEGER>
24
Specifies the number of times a job can be deferred before it will be placed in batch hold.
DEFERCOUNT 12
   
DEFERSTARTCOUNT
<INTEGER>
1
Specifies the number of times a job will be allowed to fail in its start attempts before being deferred. JOBRETRYTIME overrides DEFERSTARTCOUNT; DEFERSTARTCOUNT only begins when the JOBRETRYTIME window elapses. Note: A job's startcount will increase each time a start request is made to the resource manager regardless of whether or not this request succeeded. This means start count increases if job starts fail or if jobs are started and then rejected by the resource manager.
DEFERSTARTCOUNT 3
   
DEFERTIME
[[[DD:]HH:]MM:]SS
1:00:00
Specifies the amount of time a job will be held in the deferred state before being released back to the Idle job queue. Note: A job's defer time will be restarted if Moab is restarted.
DEFERTIME 0:05:00
   
DELETESTAGEOUTFILES
<BOOLEAN>
FALSE
Specifies whether the scheduler should delete explicitly specified stageout files after they are successfully staged. By default, such files are not deleted but are left on the nodes where the job ran.
DELETESTAGEOUTFILES TRUE

Example of an explicit stageout request

msub x=MSTAGEOUT:ssh://source_node/tmp/file,file:///results_folder
job.cmd

With this parameter set to true, /tmp/file on source_node is deleted after it is copied to the specified destination (file:///results_folder). If the parameter is not set, or if it is set to false, /tmp/file remains on source_node after the job terminates.

   
DEPENDFAILUREPOLICY
HOLD or CANCEL
HOLD
Specifies what happens to a job if its dependencies cannot be fulfilled; that is, what happens when a job depends on another job to complete successfully but the other job fails.
DEPENDFAILUREPOLICY CANCEL
If job A is submitted with depend=afterok:B and job B fails, job A is cancelled.
   
DIAGCACHESTARTITERATION
<INTEGER>
---
Specifies which iteration Moab begins using the cache to return results for the showq and mdiag -j commands. By default, running these commands immediately after reboot causes Moab to return an empty job queue rather than wait for the jobs to load from checkpoint. DIAGCACHESTARTITERATION allows you to use the non-cache blocking commands, which wait for all jobs to load from checkpoint before Moab begins using the memory cache.
DIAGCACHESTARTITERATION 4
showq and mdiag -j commands will begin using the cache when the 4th Moab iteration begins.
   
DIRECTORYSERVER
<HOST>[:<PORT>]
---
Specifies the interface for the directory server.
DIRECTORYSERVER calli3.icluster.org:4702
   
DISABLEEXCHLIST
<BOOLEAN>
FALSE
If the resource manager rejects a job and the value is set to TRUE, then the node is not added to the job's exclude host list.
DISABLEEXCHLIST TRUE
   
DISABLEINTERACTIVEJOBS
<BOOLEAN>
FALSE
Disallows interactive jobs submitted with musb -I.
DISABLEINTERACTIVEJOBS TRUE

Note: It is possible for users to submit interactive jobs directly to a resource manager, which can bypass the DISABLEINTERACTIVEJOBS parameter. However, some resource managers (such as TORQUE) will check with Moab before allowing an interactive job.

   
DISABLEREGEXCACHING
<BOOLEAN>
FALSE
Turns off regular expression caching. Turning off regular expression caching preserves memory with hostlist reservations and speeds up start time.
DISABLEREGEXCACHING TRUE
   
DISABLESAMECREDPREEMPTION
Comma delimited list of one or more credentials: ACCT, CLASS, GROUP, QOS, or USER
---
This parameter prevents specified credentials from preempting its own jobs.
DISABLESAMECREDPREEMPTION  QOS,USER
   
DISABLESAMEQOSPREEMPTION
<BOOLEAN>
FALSE
Specifies whether or not a preemptor job can preempt a preemptee job which possesses the same QoS.
DISABLESAMEQOSPREEMPTION TRUE
   
DISABLESCHEDULING
<BOOLEAN>
FALSE
Specifies whether or not the scheduler will schedule jobs. If set to TRUE, Moab will continue to update node and job state but will not start, preempt, or otherwise modify jobs. The command mschedctl -r will clear this parameter and resume normal scheduling.
DISABLESCHEDULING FALSE
   
DISABLESLAVEJOBSUBMIT
<BOOLEAN>
TRUE
This parameter can be added to the moab.cfg file on a slave Moab server (in a grid configuration) to prevent users from submitting jobs to the master Moab server from the slave Moab server. Some grid configurations allow the user to submit jobs on the slave that are migrated to the master and submitted from the master. Other grid configurations do not allow the jobs to be migrated to the master from the slave, in which case, jobs submitted from the slave remain idle on the slave and never run. This parameter will reject the job submissions on the slave to prevent the submission of jobs that will never run.
DISABLESLAVEJOBSUBMIT TRUE

example (node04 is a slave and node06 is the master)

[[email protected] moab-slurm]$ echo sleep 100 | msub
ERROR:    cannot submit job from slave
   
DISABLETHRESHOLDTRIGGERS
<BOOLEAN>
FALSE
This makes Moab not fire threshold-based triggers, but will log the intended action to the event logs. Similar to DISABLEVMDECISIONS.
DISABLETHRESHOLDTRIGGERS TRUE
   
DISABLEUICOMPRESSION
<BOOLEAN>
FALSE
Disables compression on information sent between the server and client.
DISABLEUICOMPRESSION TRUE
   
DISABLEVMDECISIONS
<BOOLEAN>
FALSE
This makes Moab not take any automatic decisions with respect to VM's, namely powering on/off nodes and migrating VMs. Intended actions will instead be logged in the event logs. Similar to DISABLETHRESHOLDTRIGGERS.
DISABLEVMDECISIONS TRUE
   
DISKWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to the amount of dedicated disk space required per task by a job (in MB).
RESWEIGHT  10
DISKWEIGHT 100

If a job requires 12 tasks and 512 MB per task of dedicated local disk space, Moab will increase the job's priority by 10 * 100 * 12 * 512

   
DISPLAYFLAGS
One or more of the following values (space delimited):

ACCOUNTCENTRIC, HIDEBLOCKED, NODECENTRIC, or USEBLOCKING

---
Specifies flags that control how Moab client commands display various information.
ACCOUNTCENTRIC will display account information in showq, rather than group information.
HIDEBLOCKED will prevent showq from listing information about blocked jobs which are not owned by the user if the user is not an admin.
NODECENTRIC will display node allocation information instead of processor allocation information in showq.
USEBLOCKING disables threading for commands that support it; those commands include showq, mdiag -n, and mdiag -j.
DISPLAYFLAGS NODECENTRIC
   
DISPLAYPROXYUSERASUSER
<BOOLEAN>
FALSE
If set to TRUE, Moab shows the proxy users instead of the real user on some queries of system jobs that have proxy users. Commands affected include mjobctl -q diag and checkjob.
DISPLAYPROXYUSERASUSER TRUE
   
DONTCANCELINTERACTIVEHJOBS
<BOOLEAN>
FALSE
If set to TRUE, Moab does not cancel interactive jobs that are held.
DONTCANCELINTERACTIVEHJOBS TRUE
   
DONTENFORCEPEERJOBLIMITS
<BOOLEAN>
FALSE
If set to TRUE, only the scheduler that is running the job can cancel the job or enforce other limits.
DONTENFORCEPEERJOBLIMITS TRUE
   
EMULATIONMODE
{ SLURM }
---
Specifies whether or not the scheduler will perform the automatic setup of a particular resource manager environment.
EMULATIONMODE SLURM

Moab will perform the automated setup steps as if it were interfacing with a slurm resource manager (automatic QOS creation).

   
ENABLEFSVIOLATIONPREEMPTION
<BOOLEAN>
FALSE
If set to TRUE, Moab will allow jobs within the same class/queue to preempt when the preemptee is violating a fairshare target and the preemptor is not.
ENABLEFSVIOLATIONPREEMPTION TRUE
   
ENABLEHIGHTHROUGHPUT
<BOOLEAN>
FALSE
Configures Moab so that it will accept msub submissions, start jobs, process triggers, etc., in a manner which minimizes their processing time. The downside is that Moab will return minimal information about these jobs at submit time (no job ID is returned). It is recommended that jobs be submitted with a "-N <JOBNAME>" argument so users can keep track of their jobs.
ENABLEHIGHTROUGHPUT TRUE

Moab can now accept hundreds of jobs per second using msub instead of 20-30.

   
ENABLEJOBARRAYS
<BOOLEAN>
True
If set to TRUE, job arrays will be enabled.
ENABLEJOBARRAYS TRUE
   
ENABLENEGJOBPRIORITY
<BOOLEAN>
FALSE
If set to TRUE, the scheduler allows job priority value to range from -INFINITY to MMAX_PRIO; otherwise, job priority values are given a lower bound of '1'. For more information, see REJECTNEGPRIOJOBS.
ENABLENEGJOBPRIORITY TRUE

Job priority may range from -INFINITY to MMAX_PRIO.

   
ENABLENODEADDRLOOKUP
<BOOLEAN>
FALSE
If set to TRUE, the scheduler will use the default host name service lookup mechanism (i.e., /etc/hosts, DNS, NIS, etc) to determine the IP address of the nodes reported by the resource manager. This information is used to correlate information reported by multi-homed hosts.
ENABLENODEADDRLOOKUP TRUE
   
ENABLEPOSUSERPRIORITY
<BOOLEAN>
FALSE
If set to TRUE, the scheduler will allow users to specify positive job priority values which will be honored. In other words, users can specify a priority that falls in the range of -1024 to +1023, inclusive. If set to FALSE (the default), user priority values are given an upper bound of '0' when users request a positive priority.
ENABLEPOSUSERPRIORITY TRUE

Users may now specify positive job priorities and have them take effect (e.g. msub -p 100 job.cmd).

   
ENABLESPVIOLATIONPREEMPTION
<BOOLEAN>
FALSE
If set to TRUE, Moab will allow jobs within the same class/queue to preempt when the preemptee is violating a soft usage policy and the preemptor is not.
ENABLESPVIOLATIONPREEMPTION TRUE
   
ENABLESTARTESTIMATESTATS
<BOOLEAN>
FALSE
If set to TRUE, Moab will collect job start time estimation stats for all submitted jobs and will use those statistics to improve future estimates. Resource availability queries which can utilize this info can be made using the showstart command. Current estimation statistics can be viewed using the mdiag -S command. To flush current statistics, use the mschedctl -f command. Note: This parameter can be resource intensive so sites with large job queues (>10,000 jobs) may want to consider leaving this set to FALSE. See Considerations for Large Clusters for more details.
ENABLESTARTESTIMATESTATS TRUE
   
ENFORCEACCOUNTACCESS
<BOOLEAN>
FALSE
Specifies whether or not Moab will enforce account access constraints without an allocation manager.
ENFORCEACCOUNTACCESS TRUE
   
ENFORCEGRESACCESS
<BOOLEAN>
FALSE
If a user submits a job with a non-existent gres (e.g. in the case of a typo) and ENFORCEGREACCESS is not set in moab.cfg, or is set to FALSE, then the requested gres will be created (but will not exist on any nodes) and the job will be deferred (similar to ENFORCEACCOUNTACCESS).
ENFORCEGRESACCESS TRUE
   
ENFORCEINTERNALCHARGING
<BOOLEAN>
FALSE
Specifies whether or not the scheduler will enforce internal credit tracking (see Internal Charging).
ENFORCEINTERNALCHARGING TRUE
   
EVENTSERVER
<HOST>[:<PORT>]
---
Specifies the interface for the event server.
EVENTSERVER  calli3.icluster.org:4702
   
FEATURENODETYPEHEADER
<STRING>
---
Specifies the header used to specify node type via node features (ie, LL features or PBS node attributes).
FEATURENODETYPEHEADER xnt

Moab will interpret all node features with the leading string xnt as a nodetype specification - as used by Gold and other allocation managers, and assign the associated value to the node. i.e., xntFast.

   
FEATUREPARTITIONHEADER
<STRING>
---
Specifies the header used to specify node partition via node features (ie, LL features or PBS node attributes).
FEATUREPARTITIONHEADER xpt

Moab will interpret all node features with the leading string xpt as a partition specification and assign the associated value to the node. i.e., xptGold.

   
FEATUREPROCSPEEDHEADER
<STRING>
---
Specifies the header used to extract node processor speed via node features (i.e., LL features or PBS node attributes). Note: Adding a trailing '$' character will specifies that only features with a trailing number be interpreted. For example, the header 'sp$' will match 'sp450' but not 'sport'.
FEATUREPROCSPEEDHEADER xps

Moab will interpret all node features with the leading string xps as a processor speed specification and assign the associated value to the node. i.e., xps950.

   
FEATURERACKHEADER
<STRING>
---
Specifies the header used to extract node rack index via node features (i.e., LL features or PBS node attributes). Note: Adding a trailing '$' character will specifies that only features with a trailing number be interpreted. For example, the header 'rack$' will match 'rack4' but not 'racket'.
FEATURERACKHEADER rack

Moab will interpret all node features with the leading string rack as a rack index specification and assign the associated value to the node. i.e., rack16.

   
FEATURESLOTHEADER
<STRING>
---
Specifies the header used to extract node slot index via node features (i.e., LL features or PBS node attributes). Note: Adding a trailing '$' character will specifies that only features with a trailing number be interpreted. For example, the header 'slot$' will match 'slot12' but not 'slotted'.
FEATURESLOTHEADER slot

Moab will interpret all node features with the leading string slot as a slot index specification and assign the associated value to the node. i.e., slot16.

   
FEEDBACKPROGRAM
<STRING>
---
Specifies the name of the program to be run at the completion of each job. If not fully qualified, Moab will attempt to locate this program in the 'tools' subdirectory.
FEEDBACKPROGRAM /var/moab/fb.pl

Moab will run the specified program at the completion of each job.

   
FILEREQUESTISJOBCENTRIC
<BOOLEAN>
FALSE
Specifies whether a job's file request is a total request for the job or a per task request.
FILEREQUESTISJOBCENTRIC TRUE

Moab will treat file requests as a total request per job.

   
FILTERCMDFILE
<BOOLEAN>
TRUE
Running the msub command performs the following operations on the submission script:
  • Replace all comments with spaces (excludes Resource Manager directives)
  • Strip empty lines
  • Replace \r with \n
  • Lock job to a PBS resource manager if $PBS is found in the script
Note Include the FILTERCMDFILE parameter in the moab.cfg file that resides on the clients.
FILTERCMDFILE FALSE

Running the msub command does not perform the actions detailed earlier.

   
FORCERSVSUBTYPE
<BOOLEAN>
FALSE
Specifies that admin reservations must have a subtype associated with them.
FORCERSVSUBTYPE TRUE

Moab will require all admin reservations to include a subtype.

   
FSACCOUNTWEIGHT
<INTEGER>
0
Specifies the weight assigned to the account subcomponent of the fairshare component of priority.
FSACCOUNTWEIGHT 10
   
FSCAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum allowed absolute value for a job's total pre-weighted fairshare component.
FSCAP 10.0

Moab will bound a job's pre-weighted fairshare component by the range +/- 10.0.

   
FSCLASSWEIGHT
<INTEGER>
0
Specifies the weight assigned to the class subcomponent of the fairshare component of priority.
FSCLASSWEIGHT 10
   
FSDECAY
<DOUBLE>
1.0
Specifies decay rate applied to past fairshare interval when computing effective fairshare usage. Values may be in the range of 0.01 to 1.0. A smaller value causes more rapid decay causing aged usage to contribute less to the overall effective fairshare usage. A value of 1.0 indicates that no decay will occur and all fairshare intervals will be weighted equally when determining effective fairshare usage. See Fairshare Overview.
FSPOLICY   DEDICATEDPS
FSDECAY    0.8
FSDEPTH    8
Moab will apply a decay rate of 0.8 to all fairshare windows.
   
FSDEPTH
<INTEGER>
8
Note: The number of available fairshare windows is bounded by the MAX_FSDEPTH value (32 in Moab). See Fairshare Overview.
FSDEPTH 12
   
FSENABLECAPPRIORITY
<BOOLEAN>
FALSE
Fairshare priority will increase to target and stop.
FSENABLECAPPRIORITY TRUE
   
FSGROUPWEIGHT
<INTEGER>
0
 
FSGROUPWEIGHT 4
   
FSINTERVAL
[[[DD:]HH:]MM:]SS
12:00:00
Specifies the length of each fairshare window.
FSINTERVAL 12:00:00
Track fairshare usage in 12 hour blocks.
   
FSJPUWEIGHT
<INTEGER>
0
Specifies the fairshare weight assigned to jobs per user.
FSJPUWEIGHT 10
   
FSMOSTSPECIFICLIMIT
<BOOLEAN>
FALSE
When checking policy usage limits in a fairshare tree, if the most specific policy limit is passed then do not check the same policy again at higher levels in the tree. For example, if a user has a MaxProc policy limit then do not check the MaxProc policy limit on the account for this same user.
FSMOSTSPECIFICLIMIT  TRUE
   
FSPOLICY
<POLICY>[*] where <POLICY> is one of the following: DEDICATEDPS, DEDICATEDPES, UTILIZEDPS, PDEDICATEDPS, or SDEDICATEDPES.
---
Specifies the unit of tracking fairshare usage. DEDICATEDPS tracks dedicated processor seconds. DEDICATEDPES tracks dedicated processor-equivalent seconds. UTILIZEDPS tracks the number of utilized processor seconds. SDEDICATEDPES tracks dedicated processor-equivalent seconds scaled by the speed of the node. PDEDICATEDPS tracks dedicated processor seconds scaled by the processor speed of the node. If the optional '%' (percentage) character is specified, percentage based fairshare will be used. See Fairshare Overview and Fairshare Consumption Metrics or more information.
FSPOLICY DEDICATEDPES

Moab will track fairshare usage by dedicated process-equivalent seconds.

   
FSPPUWEIGHT
<INTEGER>
0
Specifies the fairshare weight assigned to processors per user.
FSPPUWEIGHT 10
   
FSPSPUWEIGHT
<INTEGER>
0
Specifies the fairshare weight assigned to processor-seconds per user.
FSPSPUWEIGHT 10
   
FSQOSWEIGHT
<INTEGER>
0
Specifies the priority weight assigned to the QOS fairshare subcomponent.
FSQOSWEIGHT 16
   
FSTARGETISABSOLUTE
<BOOLEAN>
FALSE
Specifies whether Moab should base fairshare targets off of delivered cycles or up/available cycles.
FSTARGETISABSOLUTE TRUE
   
FSTREE
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
SHARES or MEMBERLIST
---
Specifies the share tree distribution for job fairshare prioritization (see Hierarchical Share Tree Overview).
FSTREE[geo]  SHARES=16  MEMBERLIST=geo103,geo313,geo422
   
FSTREEACLPOLICY
OFF, PARENT, or FULL
FULL
Specifies how Moab should interpret credential membership when building the FSTREE (see Hierarchical Share Tree Overview).
FSTREEACLPOLICY PARENT

Credentials will be given access to their parent node when applicable.

   
FSTREEISREQUIRED
<BOOLEAN>
FALSE
Specifies whether a job must have an applicable node in the partition's FSTREE in order to execute within that partition (see Hierarchical Share Tree Overview).
FSTREEISREQUIRED TRUE

Jobs must have an applicable node in the FSTREE in order to execute.

   
FSTREEUSERISREQUIRED
<BOOLEAN>
FALSE
Specifies whether the user must be given explicit access to a branch in the FSTREE (see Hierarchical Share Tree Overview).
FSTREEUSERISREQUIRED TRUE

Users must be given explicit access to FSTREE nodes in order to gain access to the FSTREE.

   
FSUSERWEIGHT
<INTEGER>
0
Specifies the priority weight assigned to the user fairshare subfactor.
FSUSERWEIGHT 8
   
FSWEIGHT
<INTEGER>
1
Specifies the priority weight assigned to the summation of the fairshare subfactors (see Priority Factor and Fairshare overviews).
FSWEIGHT 500
   
GEVENTCFG[<GEVENT>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is ACTION, ECOUNT, REARM, or SEVERITY. See Responding to Generic Events for details on values you can assign to each attribute.
---
Specifies how the scheduler should behave when various cluster events are detected. See the Generic Events Overview for more information.
GEVENTCFG[hitemp] ACTION=avoid,record,notify  REARM=00:10:00
GEVENT[nodeerror] SEVERITY=3

If a hitemp event is detected, Moab adjusts the node allocation policy to minimize the allocation of the node. Moab also sends emails to cluster administrators and reports the event in the Moab event log

   
GEVENTTIMEOUT
[HH[:MM[:SS]]]
00:30:00
Clears generic events after the specified duration.
GEVENTTIMEOUT 00:30:00
   
GRESCFG[<GRES>]
List of zero or more space delimited <ATTR >=<VALUE> pairs where <ATTR> is one of the following:
TYPE or STARTDELAY
---
Specifies associations of generic resources into resource groups. See 12.6 Managing Consumable Generic Resources for more information.
GRESCFG[scsi1] TYPE=fastio
GRESCFG[scsi2] TYPE=fastio
GRESCFG[scsi3] TYPE=fastio

The generic resources scsi1, scsi2, and scsi3 are all associated with the generic resource type fastio.

   
GRESTOJOBATTR
Comma delimited list of generic resources
---
The list of generic resources will also be interpreted as JOB features. See 7.1.5 Managing Reservations.
GRESTOJOBATTR  matlab,ccs
Jobs which request the generic resources matlab or ccs will have a corresponding job attribute assigned to them.
   
GROUPCFG[<GROUPID>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
General Credential Flags, PRIORITY, ENABLEPROFILING, TRIGGER, QLIST, QDEF, PLIST, PDEF, FLAGS, usage limits, or a fairshare usage limit specification.
---
Specifies group specific attributes. See the flag overview for a description of legal flag values.
GROUPCFG[staff] MAXJOB=50 QDEF=highprio
Up to 50 jobs submitted by members of the group staff will be allowed to execute simultaneously and will be assigned the QOS highprio by default.
   
GROUPWEIGHT
<INTEGER>
1
Specifies the priority weight assigned to the specified group priority (See Credential (CRED) Factor).
GROUPWEIGHT 20
   
GUARANTEEDPREEMPTION
<BOOLEAN>
FALSE
This parameter relies on the resource manager to tell Moab that the job can't start now. When a job preempts, Moab will attempt to start the job. If the resource manager reports back that it can't start the job because the preemptees are still cleaning up on the nodes, the preemptor will hold on to its reservation and keep trying to start for JOBRETRYTIME. If the job doesn't start in JOBRETRYTIME, it will be deferred and will lose its reservation. When it comes out of deferral, Moab will attempt to run the job elsewhere.
GUARANTEEDPREEMPTION TRUE
   
HALOCKCHECKTIME
<SECONDS>
9
Specifies how frequently the Moab fallback (or secondary) server checks the lock file timestamp. See the high availability overview and configuration sections for more information.
HALOCKCHECKTIME 9
The Moab fallback server will check the lock file timestamp every 9 seconds.
   
HALOCKUPDATETIME
<SECONDS>
3
Specifies how frequently the primary Moab server updates the timestamp on the lock file. See the high availability overview and configuration sections for more information.
HALOCKUPDATETIME 3
The primary Moab server will update the lock file timestamp every 3 seconds.
   
HAPOLLINTERVAL
[[[DD:]HH:]MM:]SS
30
Specifies the amount of time between subsequent high availability pings. See High Availability Overview for more info.
HAPOLLINTERVAL 00:00:15
The Moab fallback server will check the health of the Moab primary every 15 seconds.
   
HIDEVIRTUALNODES
<BOOLEAN>
---
Enables VM management; also used to reveal hypervisors.
HIDEVIRTUALNODES TRANSPARENT
   
IDCFG[X]
One or more of the following attribute/value pairs: BLOCKEDCREDLIST, CREATECRED, CREATECREDURL, REFRESHPERIOD, RESETCREDLIST, SERVER or UPDATEREFRESHONFAILURE.
---
This parameter enables the identity manager interface allowing credential, policy, and usage information to be shared with an external information service.
IDCFG[info] SERVER=exec:///opt/moab/dbquery.pl REFRESHPERIOD=hour

Moab will refresh credential info every hour using the specified script.

   
IGNOREMDATASTAGING
<BOOLEAN>
FALSE
When set to TRUE, Moab will ignore any resource manager specific data staging on a job and assume the resource manager is processing the request. Currently, this only applies to PBS.
IGNORERMDATASTAGING TRUE
   
IGNORECLASSES
[!]<CLASS>[,<CLASS>]...
---
By default, if using the TORQUE resource manager, jobs from all listed classes are ignored and not scheduled, tracked, or otherwise processed by Moab. If the not (i.e., '!') character is specified, only jobs from listed classes are processed. See the Moab Side-by-Side Analysis for more information.
IGNORECLASSES dque,batch

Moab will ignore jobs from classes dque and batch.

   
IGNOREJOBS
[!]<JOBID>[,<JOBID>]...
---
By default, listed jobs are ignored and not scheduled, tracked, or otherwise processed by Moab. If the not (i.e., '!') character is specified, only listed jobs are processed. See the Moab Side-by-Side Analysis for more information.
IGNOREJOBS !14221,14223

Moab will ignore jobs all classes except 14221 and 14223.

   
IGNORENODES
[!]<NODE>[,<NODE>]...
---
By default, all listed nodes are ignored and not scheduled, tracked, or otherwise processed by Moab. If the not (i.e., '!') character is specified, only listed nodes are processed. See the Moab Side-by-Side Analysis for more information.
IGNORENODES !host3,host4

Moab will only process nodes host3 and host4.

   
IGNOREPREEMPTEEPRIORITY
<BOOLEAN>
FALSE
By default, preemptor jobs can only preempt preemptee jobs if the preemptor has a higher job priority than the preemptee. When this parameter is set to true, the priority constraint is removed allowing any preemptor to preempt any preemptees once it reaches the top of the eligible job queue.
IGNOREPREEMPTEEPRIORITY TRUE

A preemptor job can preempt any preemptee jobs when it is at the top of the eligible job queue.

   
IGNOREUSERS
[!]<USERNAME>[,<USERNAME>]...
---
By default, if using the TORQUE resource manager, jobs from all listed users are ignored and not scheduled, tracked, or otherwise processed by Moab. If the not (i.e., '!') character is specified, only jobs from listed users are processed. (See the Moab Side-by-Side Analysis for more information.)
IGNOREUSERS testuser1,annapolis

Moab will ignore jobs from users testuser1 and annapolis.

   
#INCLUDE
<STRING>
---
Specifies another file which contains more configuration parameters. If <STRING> is not an absolute path, Moab will search its home directory for the file.
#INCLUDE moab.acct

 Moab will process the parameters in moab.acct as well as moab.cfg

   
INSTANTSTAGE
<BOOLEAN>
FALSE
Specifies whether Moab should instantly stage jobs to the underlying resource manager when a job is submitted through msub.
INSTANTSTAGE TRUE
   
INVALIDFSTREEMSG
"<STRING>"
"no valid fstree node found"
Specifies the error message that should be attached to jobs that cannot run because of a fairshare tree configuration violation.
INVALIDFSTREEMSG        "account is invalid for requested partition"
   
JOBACTIONONNODEFAILURE
CANCEL, FAIL, HOLD, IGNORE, NOTIFY, or REQUEUE
---
Specifies the action to take if Moab detects that a node allocated to an active job has failed (state is down). By default, Moab only reports this information via diagnostic commands. If this parameter is set, Moab will cancel or requeue the active job. See Reallocating Resources When Failures Occur for more information.

Note: The HOLD value is only applicable when using checkpointing.
JOBACTIONONNODEFAILURE REQUEUE

Moab will requeue active jobs which have allocated nodes which have failed during the execution of the job.

   
JOBACTIONONNODEFAILUREDURATION
<INTEGER>
300
Specifies how long a node must be marked as down before Moab performs the failure action. It eliminates failure actions resulting from short-lived network failures or glitches that cause nodes to appear down briefly when they are functional.
JOBACTIONONNODEFAILUREDURATION 60

Moab waits for a node to be down for 60 seconds before performing the failure action.

   
JOBAGGREGATIONTIME
[[[DD:]HH:]MM:]SS
0
Specifies the minimum amount of time the scheduler should wait after receiving a job event until it should process that event. This parameter allows sites with bursty job submissions to process job events in groups decreasing total job scheduling cycles and allowing the scheduler to make more intelligent choices by aggregating job submissions and choosing between the jobs. See Considerations for Large Clusters.
JOBAGGREGATIONTIME 00:00:04
RMPOLLINTERVAL     00:00:30

Moab will wait 4 seconds between scheduling cycles when job events have been received and will wait 30 seconds between scheduling cycles otherwise

   
JOBCFG
<ATTR>=<VAL> where <ATTR> is one of FLAGS, GRES, NODERANGE, PRIORITYF, PROCRANGE, QOS, RARCH, REQFEATURES, ROPSYS, or SELECT
---
Specifies attributes for jobs which satisfy the specified profile. The SELECT attribute allows users to specify the job template by using msub -l template=.

The JOBCFG parameter supports the following attributes:
NONE, ACCOUNT, ACTION, AUTOSIZE, CLASS, CPULIMIT, DESCRIPTION, DGRES, FAILUREPOLICY, GROUP, IFLAGS, JOBSCRIPT MEM (for MEM=<value>), MEMORY (for MEMORY=$LEARN), NODEACCESSPOLICY, NODEMOD, PARTITION, PREF, QOS, RESTARTABLE, RM, RMSERVICEJOB, SELECT, STAGEIN, SOFTWARE, SRM, TEMPLIMIT, TFLAGS, USER, VMUSAGE, WALLTIME, WORK

It also supports the following Wiki attributes:
ARGS, DMEM, DDISK, DWAP, ERROR, EXEC, EXITCODE, GATTR, GEVENT, IWD, JNAME, NAME, PARTITIONMASK, PRIORITYF, RDISK, RSWAP, RAGRES, RCGRES, TASKPERNODE, TRIGGER, VARIABLE, NULL

Note: The index to the JOBCFG parameter can either be an admin-chosen job template name or the exact name of the job reported by one or more workload queries. See Wiki Attributes and Job Template Extensions.

JOBCFG[sql] REQFEATURES=sqlnode QOS=service

When the sql job is detected, it will have the specified default qos and node feature attributes set.

   
JOBCPURGETIME
[[[DD:]HH:]MM:]SS
00:05:00
Specifies the amount of time Moab will preserve detailed information about a completed job (see JOBPURGETIME, showq -c and checkjob).
JOBCPURGETIME 02:00:00

Moab will maintain detailed job information for two hours after a job has completed.

   
JOBCTRUNCATENLCP
<BOOLEAN>
TRUE
Specifies whether Moab will store only the first node of the node list for a completed job in the checkpoint file.
JOBCTRUNCATENLCP TRUE

JOBCTRUNCATENLCP reduces the amount of memory Moab uses to store completed job information.

   
JOBEXTENDSTARTWALLTIME
<BOOLEAN>
---
Extends the job walltime when Moab starts the job up to the lesser of the maximum or the next reservation (rounded down to the nearest minute).
JOBEXTENDSTARTWALLTIME TRUE

Submit job with a minimum wallclock limit and a walltime; for example:

echo sleep 500 | msub -A ee -l
nodes=5,minwclimit=5:00,walltime=30:00,partition=g02
At job start, Moab recognizes the nodes assigned to the specified job and extends the walltime for the job (one time at job start) up to the lesser of the maximum walltime requested or the least amount of time available for any of the nodes until the next reservation on that node.

   
JOBFAILRETRYCOUNT
<INTEGER>
0

Specifies the number of times a job is requeued and restarted by Moab if the job fails (if the job itself returns a non-zero exit code). Some types of jobs may succeed if automatically retried several times in short succession. This parameter was created with these types of jobs in mind. Note that the job in question must also be restartable (the job needs to have the "RESTARTABLE" flag set on it) and the RM managing the job must support requeuing and starting completed jobs.

If a job fails too many times, and reaches the number of retries given by JobFailRetryCount, then a UserHold is placed on the job and a message is attached to it signifying that the job has a "restart count violation."

JOBFAILRETRYCOUNT  7
Any job with a RESTARTABLE flag is requeued, if it fails, up to 7 times before a UserHold is placed on it.
   
JOBIDWEIGHT
<INTEGER>
---
Specifies the weight to be applied to the job's id. See Attribute (ATTR) Factor.
JOBIDWEIGHT -1
Later jobs' priority will be negatively affected.
   
JOBMATCHCFG
<ATTR>=<VAL> where <ATTR> is one of JMIN, JMAX, JDEF, JSET, or JSTAT
---
Specifies the job templates which must be matched and which will be applied in the case of a match.
JOBMATCHCFG[sql] JMIN=interactive JSTAT=istat
   
JOBMAXHOLDTIME
[[[DD:]HH:]MM:]SS
---
Specifies the amount of time a job can be held before it is canceled automatically.
JOBMAXHOLDTIME 02:00:00

Moab will keep jobs in any HOLD state for 2 hours before canceling them.

   
JOBMAXNODECOUNT
<INTEGER>
1024
Specifies the maximum number of nodes which can be allocated to a job. After changing this parameter, Moab must be restarted.  Note: This value cannot exceed either MMAX_NODE or MMAX_TASK_PER_JOB. If larger values are required, these values must also be increased. Moab must be restarted before changes to this command will take affect. The command mdiag -S will indicate if any job node count overflows have occurred. See Consideration for Large Clusters.
JOBMAXNODECOUNT 4000
   
JOBMAXOVERRUN
[[[[DD:]HH:]MM:]SS,][[[DD:]HH:]MM:]SS
(no soft limit), 10 minutes (hard limit)
Soft and hard limit of the amount of time Moab will allow a job to exceed its wallclock limit before it first sends a mail to the primary admin (soft limit) and then terminates the job (hard limit). See WCVIOLATIONACTION or Resource Usage Limit Overview.
JOBMAXOVERRUN 15:00,1:00:00

Jobs may exceed their wallclock limit by up to 1 hour, but Moab will send an email to the primary administrator when a job exceeds its walltime by 15 minutes.

   
JOBMAXPREEMPTPERITERATION
<INTEGER>
0 (No Limit)
Maximum number of jobs allowed to be preempted per iteration.
JOBMAXPREEMPTPERITERATION 10
   
JOBMAXSTARTPERITERATION
<INTEGER>
0 (No Limit)
Maximum number of jobs allowed to start per iteration.
JOBMAXSTARTPERITERATION  10
   
JOBMAXSTARTTIME
[[[DD:]HH:]MM:]SS
-1 (NO LIMIT)
length of time a job is allowed to remain in a 'starting' state. If a 'started' job does not transition to a running state within this amount of time, Moab will cancel the job, believing a system failure has occurred. 
JOBMAXSTARTTIME 2:00:00

Jobs may attempt to start for up to 2 hours before being canceled by the scheduler

   
JOBMIGRATEPOLICY
One of the following: IMMEDIATE, JUSTINTIME, or AUTO
AUTO
Upon using the msub command to submit a job, you can allocate the job to immediately (IMMEDIATE) migrate to the resource manager, or you can instruct Moab to only migrate the job to the resource manager when it is ready to run (JUSTINTIME). Specifying AUTO allows MOAB to determine on a per-job basis whether to use IMMEDIATE or JUSTINTIME.
JOBMIGRATEPOLICY JUSTINTIME
   
JOBNAMEWEIGHT
<INTEGER>
---
Specifies the weight to be applied to the job's name if the Name contains an integer. See Attribute (ATTR) Factor.
JOBNAMEWEIGHT 1
   
JOBNODEMATCHPOLICY
EXACTNODE or EXACTPROC
---
Specifies additional constraints on how compute nodes are to be selected. EXACTNODE indicates that Moab should select as many nodes as requested even if it could pack multiple tasks onto the same node. EXACTPROC indicates that Moab should select only nodes with exactly the number of processors configured as are requested per node even if nodes with excess processors are available.
JOBNODEMATCHPOLICY EXACTNODE

In a PBS/Native job with resource specification 'nodes=<x>:ppn=<y>', Moab will allocate exactly <y> task on each of <x> distinct nodes.

   
JOBPREEMPTMAXACTIVETIME
[[[DD:]HH:]MM:]SS
0
The amount of time in which a job may be eligible for preemption. See Job Preemption.
JOBPREEMPTMAXACTIVETIME 00:05:00

A job is preemptable for the first 5 minutes of its run time.

   
JOBPREEMPTMINACTIVETIME
[[[DD:]HH:]MM:]SS
0
The minimum amount of time a job must be active before being considered eligible for preemption. See Job Preemption.
JOBPREEMPTMINACTIVETIME 00:05:00

A job must execute for 5 minutes before Moab will consider it eligible for preemption.

   
JOBPRIOACCRUALPOLICY
ACCRUE or RESET
ACCRUE
Specifies how Moab should track the dynamic aspects of a job's priority. ACCRUE indicates that the job will accrue queuetime based priority from the time it is submitted unless it violates any of the policies not specified in JOBPRIOEXCEPTIONS. RESET indicates that it will accrue priority from the time it is submitted unless it violates any of the JOBPRIOEXCEPTIONS. However, with RESET, if the job does violate JOBPRIOEXCEPTIONS then its queuetime based priority will be reset to 0.

Note: the following old JOBPRIOACCRUALPOLICY values have been deprecated and should be adjusted to the following values:

  • QUEUEPOLICY = ACCRUE and JOBPRIOEXCEPTIONS SOFTPOLICY,HARDPOLICY
  • QUEUEPOLICYRESET = RESET and JOBPRIOEXCEPTIONS SOFTPOLICY,HARDPOLICY
  • ALWAYS = ACCRUE and JOBPRIOEXCEPTIONS ALL
  • FULLPOLICY = ACCRUE and JOBPRIOEXCEPTIONS NONE
  • FULLPOLICYRESET = RESET and JOBPRIOEXCEPTIONS NONE
JOBPRIOACCRUALPOLICY   RESET

Moab will adjust the job's dynamic priority subcomponents, i.e., QUEUETIME, XFACTOR, and TARGETQUEUETIME, etc. each iteration that the job does not violate any JOBPRIOEXCEPTIONS, if it is found in violation, its queuetime will be reset to 0.

   
JOBPRIOEXCEPTIONS
Comma delimited list of any of the following: DEFER, DEPENDS, SOFTPOLICY, HARDPOLICY, IDLEPOLICY, USERHOLD, BATCHHOLD, and SYSTEMHOLD (ALL or NONE can also be specified on their own)
NONE
Specifies exceptions for calculating a job's dynamic priority (QUEUETIME, XFACTOR, TARGETQUEUETIME). Normally, when a job violates a policy, is placed on hold, or has an unsatisfied dependency, it will not accrue priority. Exceptions can be configured to allow a job to accrue priority inspite of any of these violations. With DEPENDS a job will increase in priority even if there exists an unsatisfied dependency. With SOFTPOLICY, HARDPOLICY, or IDLEPOLICY a job can accrue priority despite violating a specific limit. With DEFER, USERHOLD, BATCHHOLD, or SYSTEMHOLD a job can accrue priority despite being on hold.
JOBPRIOEXCEPTIONS BATCHHOLD,SYSTEMHOLD,DEPENDS

Jobs will accrue priority in spite of batchholds, systemholds, or unsatisfied dependencies.

   
JOBPRIOF
<ATTRIBUTE>[<VALUE>]=<PRIORITY> where <ATTRIBUTE> is one of ATTR, GRES or STATE
---
Specifies attribute priority weights for jobs with specific attributes, generic resource requests, or states. State values must be one of the standard Moab job states. See Attribute-Based Job Prioritization.
JOBPRIOF         STATE[Running]=100  STATE[Suspended]=1000  ATTR[PREEMPTEE]=200  GRES[biocalc]=5
ATTRATTRWEIGHT   1
ATTRSTATEWEIGHT  1
Moab will adjust the job's dynamic priority subcomponents.
   
JOBPURGETIME
[[[DD:]HH:]MM:]SS
0 (purge immediately if the resource manager does not report the job)
The amount of time Moab will keep a job record which is no longer reported by the resource manager. Useful when using a resource manager which drops information about a job due to internal failures. See JOBCPURGETIME.
JOBPURGETIME 00:05:00

Moab will maintain a job record for 5 minutes after the last update regarding that object received from the resource manager.

   
JOBREJECTPOLICY
One or more of CANCEL, HOLD, IGNORE (beta), MAIL, or RETRY
HOLD
Specifies the action to take when the scheduler determines that a job can never run. CANCEL issues a call to the resource manager to cancel the job. HOLD places a batch hold on the job preventing the job from being further evaluated until released by an administrator. (Note: Administrators can dynamically alter job attributes and possibly fix the job with mjobctl -m.) With IGNORE (currently in beta), the scheduler will allow the job to exist within the resource manager queue but will neither process it nor report it. MAIL will send email to both the admin and the user when rejected jobs are detected. If RETRY is set, then Moab will allow the job to remain idle and will only attempt to start the job when the policy violation is resolved.  Any combination of attributes may be specified. See QOSREJECTPOLICY.
JOBREJECTPOLICY  MAIL,CANCEL
   
JOBREMOVEENVVARLIST
Comma-delimited list of strings
---

Moab will remove the specified environment variables from the job's environment before migrating the job to its destination resource manager. This is useful when jobs submit themselves from one cluster to another with the full environment.

Note This parameter is currently only supported with torque resource managers.
JOBREMOVEENVVARLIST PBS_SERVER,TZ

Moab will remove the environment variables PBS_SERVER and TZ before submitting jobs.

   
JOBRETRYTIME
[[[DD:]HH:]MM:]SS
60 seconds
Period of time Moab will continue to attempt to start a job which has failed to start due to transient failures or which has successfully started and was then rejected by the resource manager due to transient failures. See Reservation Policies, RESERVATIONRETRYTIME, and JOBRETRYTIME for more information.
JOBRETRYTIME   00:05:00

Moab will try for up to 5 minutes to restart jobs if the job start has failed due to transient errors.

   
JOBSYNCTIME
[[[DD:]HH:]MM:]:SS
00:10:00
Specifies the length of time after which Moab will synchronize a job's expected state with an unexpected reported state. IMPORTANT Note: Moab will not allow a job to run while its expected state does not match the state reported by the resource manager.
JOBSYNCTIME 00:01:00
   
LIMITEDJOBCP
<BOOLEAN>
FALSE
Specifies whether there should be limited job checkpointing (see Consideration for Large Clusters).
LIMITEDJOBCP TRUE

Moab will only maintain scheduler checkpoint information for jobs with explicitly modified job attributes. Some minor job performance and usage statistics may be lost.

   
LIMITEDNODECP
<BOOLEAN>
FALSE
Specifies whether there should be limited node checkpointing (see Consideration for Large Clusters).
LIMITEDNODECP TRUE

Moab will only maintain scheduler checkpoint information for nodes with explicitly modified job attributes. (some minor node performance and usage statistics may be lost)

   
LOADALLJOBCP
<BOOLEAN>
FALSE
Specifies whether Moab should load, during startup, all non-completed jobs in the checkpoint files regardless of whether or not their corresponding resource managers are active. For example, this allows source peers to continue showing remote jobs in the queue based on checkpointed info, even though the destination peer is offline.
LOADALLJOBCP TRUE

Moab will load, at startup, all non-completed jobs from all checkpoint files.

   
LOCKFILE
<STRING>
---
Specifies the path for the lock (pid) file used by Moab.
LOCKFILE /var/spool/moab/lock
   
LOGDIR
<STRING>
log
Specifies the directory in which log files will be maintained. If specified as a relative path, LOGDIR will be relative to $(MOABHOMEDIR) See Logging Overview for more information.
LOGDIR /var/spool/moab

Moab will record its log files directly into the /var/spool/moab directory

   
LOGFACILITY
Colon delimited list of one or more of the following:  fCORE, fSCHED, fSOCK, fUI, fLL, fCONFIG, fSTAT, fSIM, fSTRUCT, fFS, fCKPT, fBANK, fRM, fPBS, fWIKI, fALL
fALL
Specifies which types of events to log (see Logging Overview).
LOGFACILITY fRM:fPBS

Moab will log only events involving general resource manager or PBS interface activities.

   
LOGFILE
<STRING>
moab.log
Name of the Moab log file. This file is maintained in the directory pointed to by <LOGDIR> unless <LOGFILE> is an absolute path (see Logging Overview)
LOGFILE moab.test.log

Log information will be written to the file moab.test.log located in the directory pointed to by the LOGDIR parameter.

   
LOGFILEMAXSIZE
<INTEGER>
10000000
Maximum allowed size (in bytes) of the log file before it will be rolled (see Logging Overview).
LOGFILEMAXSIZE 50000000

Log files will be rolled when they reach 50 MB in size

   
LOGFILEROLLDEPTH
<INTEGER>
3
Number of old log files to maintain (i.e., when full, moab.log will be renamed moab.log.1, moab.log.1 will be renamed moab.log.2, ...). See Logging Overview.
LOGFILEROLLDEPTH 5

Moab will maintain and roll the last 5 log files.

   
LOGLEVEL
<INTEGER> (0-9)
0
Specifies the verbosity of Moab logging where 9 is the most verbose (Note: each logging level is approximately an order of magnitude more verbose than the previous level). See Logging Overview.
LOGLEVEL 4

Moab will write all Moab log messages with a threshold of 4 or lower to the moab.log file.

   
LOGLEVELOVERRIDE
<BOOLEAN>
FALSE
When this parameter is on, if someone runs a command with --loglevel=<x>, that loglevel, if higher than the current loglevel, is used on the scheduler side for the duration of the command. All logs produced during that time are put into a separate log file (this creates a "gap" in the normal logs). This can be very useful for debugging, but it is recommend that this be used only when diagnosing a specific problem so that users can't affect performance by submitting multiple --loglevel commands.
Note This parameter does not work with threaded commands (such as showq, mdiag -n, and mdiag -j).
LOGLEVELOVERRIDE    TRUE
   
LOGROLLACTION
<STRING>
---
Specifies a script to run when the logs roll. The script is run as a trigger and can be viewed using mdiag -T. For example, a script can be specified that always moves the first rolled log file, moab.log.1, to an archive directory for longer term storage.
LOGROLLACTION /usr/local/tools/logroll.pl
   
MAILPROGRAM
[ <Full_Path_To_Mail_Command> | DEFAULT | NONE ][@<DEFAULTMAILDOMAIN>]
NONE
If set to NONE, no mail will be sent.  If set to Default:, the default mail program, /usr/bin/sendmail, will be used. Note: By default, Moab mail notification is disabled. To enable, MAILPROGRAM must be set to Default: or to the locally available mail program. If the default mail domain is set, emails will be routed to this domain unless a per-user domain is specified using the EMAILADDRESS attribute of the USERCFG parameter. See Notify Admins.
MAILPROGRAM /usr/local/bin/sendmail
   
MAXGMETRIC
<INTEGER>
10
Specifies how many generic metrics Moab should manage.
MAXGMETRIC 20
   
MAXGRES
<INTEGER>
NONE
Specifies how many generic resources Moab should manage.
MAXGRES 1024
   
MAXJOB
<INTEGER>
4096
Specifies the maximum number of simultaneous jobs which can be evaluated by the scheduler. If additional jobs are submitted to the resource manager, Moab will ignore these jobs until previously submitted jobs complete. Note: Moab must be restarted for any changes to this parameter to take affect. The command mdiag -S will indicate if any job overflows have occurred.
MAXJOB 45000
   
MAXRSVPERNODE
<INTEGER>
24
Specifies the maximum number of reservations on a node.

For large SMP systems (>512 processors/node), Adaptive Computing advises adjusting the value to approximately twice the average sum of admin, standing, and job reservations present.

A second number, led by a comma, can also be specified to set a maximum number of reservations for nodes that are part of the SHARED partition.

Moab must be restarted for any changes to this parameter to take effect. The command mdiag -S indicates whether any node reservation overflows have occurred. See Considerations for Large Clusters.
Caution Do not lower the MAXRSVPERNODE value while there are active jobs in the queue. This can lead to queue instability and certain jobs could become stuck or disconnected from the system.
MAXRSVPERNODE 64

64 is the maximum number of reservations on a single node.

MAXRSVPERNODE 100,7000

100 is the maximum number of reservations on a single node, and 7000 is the maximum number of reservations for global nodes.

   
MEMREFRESHINTERVAL
[[[DD:]HH:]MM:]:SS | job:<COUNT>
---
Specifies the time interval or total job query count at which Moab will perform garbage collection to free memory associated with resource manager API's which possess memory leaks (i.e., Loadleveler, etc).
# free memory associated with leaky RM API
MEMREFRESHINTERVAL 24:00:00

Moab will perform garbage collection once a day.

   
MEMWEIGHT
<INTEGER>
0
Specifies the coefficient to be multiplied by a job's MEM (dedicated memory in MB) factor. See Resource Priority Overview.
RESWEIGHT 10
MEMWEIGHT 1000

Each job's priority will be increased by 10 * 1000 * <request memory>.

   
MINADMINSTIME
<INTEGER>
60 seconds
Specifies the minimum time a job will be suspended if suspended by an administrator or by a scheduler policy.
MINADMINSTIME 00:10:00
Each job suspended by administrators or policies will stay in the suspended state for at least 10 minutes .
   
MISSINGDEPENDENCYACTION
CANCEL, HOLD, or RUN
HOLD
Controls what Moab does with a dependent job when its dependency job cannot be found when Moab evaluates the dependent job for scheduling. This only affects jobs whose dependent job cannot be found.
MISSINGDEPENDENCYACTION CANCEL
Any job that has a dependent job that cannot be found is cancelled.
   
MSUBQUERYINTERVAL
<INTEGER>
5 seconds

Specifies the length of the interval (in seconds) between job queries when using msub -K. Jobs submitted with the -K option query the scheduler every MSUBQUERYINTERVAL seconds until the job is completed.

MSUBQUERYINTERVAL can exist as an environment variable. Any value in moab.cfg overrides the environment variable.

Note: If MSUBQUERYINTERVAL is set to 0, the -K option will be disabled. Jobs will still submit correctly, but the client will not continue to check on the job.

MSUBQUERYINTERVAL  60

If a user uses the msub -K command, the client remains open and queries the server every 60 seconds until the job completes.

   
NODEACCESSPOLICY
One of the following: SHARED, SHAREDONLY, SINGLEJOB, SINGLETASK , SINGLEUSER, or UNIQUEUSER
SHARED
Specifies how node resources will be shared by various tasks (See the Node Access Overview for more information).
NODEACCESSPOLICY SINGLEUSER

Moab will allow resources on a node to be used by more than one job provided that the job's are all owned by the same user.

   
NODEALLOCATIONPOLICY
One of the following:
FIRSTAVAILABLE, LASTAVAILABLE, MINRESOURCE, CPULOAD, LOCAL, CONTIGUOUS, MAXBALANCE, PRIORITY, or FASTEST
LASTAVAILABLE
Specifies how Moab should allocate available resources to jobs. See Node Allocation Overview for more information.
NODEALLOCATIONPOLICY MINRESOURCE

Moab will apply the node allocation policy MINRESOURCE to all jobs by default.

   
NODEALLOCRESFAILUREPOLICY
One of the following: CANCEL, HOLD, IGNORE, MIGRATE, NOTIFY, or REQUEUE
NONE
Specifies how Moab should handle active jobs which experience node failures during execution. See the RESFAILPOLICY resource manager extension or the Node Availability Overview.
NODEALLOCRESFAILUREPOLICY REQUEUE

Moab will requeue jobs which have allocated nodes fail during execution.

   
NODEAVAILABILITYPOLICY
<POLICY>[:<RESOURCETYPE>] ...

where POLICY is one of COMBINED, DEDICATED, or UTILIZED
and RESOURCETYPE is one of PROC, MEM, SWAP, or DISK

COMBINED
Specifies how available node resources are reported. Moab uses the following calculations to determine the amount of available resources:

Dedicated (use what Moab has scheduled to be used):
   Available = Configured - Dedicated
Utilized (use what the resource manager is reporting is being used):
   Available = Configured - Utilized
Combined (use the larger of dedicated and utilized):
   Available = Configured - (MAX(Dedicated,Utilized))

Moab marks a node as busy when it has no available processors, so NODEAVAILABILTYPOLICY, by affecting how many processors are reported as available, also affects node state. See Node Availability Policies for more information.
NODEAVAILABILITYPOLICY DEDICATED:PROCS COMBINED:MEM

Moab will ignore resource utilization information in locating available processors for jobs but will use both dedicated and utilized memory information in determining memory availability.

   
NODEBUSYSTATEDELAYTIME
[[[DD:]HH:]MM:]SS
0:01:00 (one minute)
Length of time Moab will assume busy nodes will remain unavailable for scheduling if a system reservation is not explicitly created for the node.
NODEBUSYSTATEDELAYTIME 0:30:00

Moab will assume busy nodes are not available for scheduling for at least 30 minutes from the current time. Thus, these nodes will never be allocated to starting jobs. Also, these nodes will only be available for reservations starting more than 30 minutes in the future.

   
NODECATCREDLIST

<LABEL>=<NODECAT>[,<NODECAT>]...[ <LABEL>=<NODECAT>[,<NODECAT>]...]... where
<LABEL> is any string and <NODECAT> is one of the defined node categories.

---
If specified, Moab will generate node category groupings and each iteration will assign usage of matching resources to pseudo-credentials with a name matching the specified label. See the Node Categorization section of the Admin manual for more information.
NODECATCREDLIST down=BatchFailure,HardwareFailure,NetworkFailure idle=Idle

Moab will create a down user, group, account, class, and QoS and will associate BatchFailure, HardwareFailure, and NetworkFailure resources with these credentials. Additionally, Moab will assign all Idle resources to matching idle credentials.

   
NODECFG[X]
List of space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
ACCESS, CHARGERATE, FEATURES, FLAGS, GRES, LOGLEVEL, MAXJOB, MAXJOBPERUSER, MAXLOAD, MAXPE, NODETYPE, OSLIST, PARTITION, POOL, POWERPOLICY, PRIORITY, PRIORITYF, PROCSPEED, RACK, RADISK, SLOT, SPEED, or TRIGGER
---
Specifies node-specific attributes for the node indicated in the array field. See the General Node Administration Overview for more information.
NODECFG[nodeA] MAXJOB=2 SPEED=1.2

Moab will only allow two simultaneous jobs to run on node nodeA and will assign a relative machine speed of 1.2 to this node.

   
NODEDOWNSTATEDELAYTIME
[[[DD:]HH:]MM:]SS
-1 (never)
Length of time Moab will assume down, drained (offline), or corrupt nodes will remain unavailable for scheduling if a system reservation is not explicitly created for the node. The default specification of "-1" causes Moab to never create job reservations on down nodes. See Node Availability for more information.
NODEDOWNSTATEDELAYTIME 0:30:00

Moab will assume down, drained, and corrupt nodes are not available for scheduling for at least 30 minutes from the current time. Thus, these nodes will never be allocated to starting jobs. Also, these nodes will only be available for reservations starting more than 30 minutes in the future.

   
NODEDOWNTIME
[[[DD:]HH:]MM:]SS
---
The maximum time a previously reported node remains unreported by a resource manager before the node is considered to be in the down state. This can happen if communication with a resource manager or a peer server is lost for more than the specified length of time, or if there is communication with the resource manager but it fails to report the node status.
NODEDOWNTIME 10:00

If Moab loses communication with the resource manager for more than ten minutes, it sets the state of all nodes belonging to that resource manager to DOWN.

   
NODEDRAINSTATEDELAYTIME
[[[DD:]HH:]MM:]SS
3:00:00 (three hours)
Length of time Moab will assume drained nodes will remain unavailable for scheduling if a system reservation is not explicitly created for the node. Specifying "-1" will cause Moab to never create job reservations on drained nodes. See Node Availability for more information.
NODEDRAINSTATEDELAYTIME 0:30:00

Moab will assume down, drained, and corrupt nodes are not available for scheduling for at least 30 minutes from the current time. Thus, these nodes will never be allocated to starting jobs. Also, these nodes will only be available for reservations starting more than 30 minutes in the future.

   
NODEFAILURERESERVETIME
[[[DD:]HH:]MM:]SS
0:05:00
Duration of reservation Moab will place on any node in which it detects a failure from the resource manager (0 indicates no reservation will be placed on the node). See Node Availability for more information. See also RMCFG[] NODEFAILURERSVPROFILE.
NODEFAILURERESERVETIME 10:00

Moab will reserve failed nodes for 10:00

   
NODEIDFORMAT
<STRING>
*$N*
Specifies how a node id can be processed to extract possible node, rack, slot, and cluster index information. The value of the parameter may include the markers '$C' (cluster index), '$N' (node index), '$R' (rack index), or '$S' (slot index) separated by '* (asterisk - representing any number of non-numeric characters) or other characters to indicate this encoding. See Node Selection for more information on use of node, rack, and slot indices.
NODEIDFORMAT *$R*$S

Moab will extract rack and slot information from the cluster node ids (ie, tg-13s08).

   
NODELOADPOLICY
One of the following: ADJUSTSTATE or ADJUSTPROCS
ADJUSTSTATE
Specifies whether a node's load affects its state or its available processors. The ADJUSTSTATE policy specifies that Moab should mark the node busy when the node's maximum load is reached. The ADJUSTPROCS policy causes the node's available procs to be equivalent to MIN(ConfiguredProcs - DedicatedProcs,MaxLoad - CurrentLoad) Note: NODELOADPOLICY only affects a node if MAXLOAD has been set.
NODELOADPOLICY ADJUSTSTATE

Moab will mark a node busy if its measured load exceeds its maximum cpu load limit.

   
NODEMAXLOAD
<DOUBLE>
0.0
Specifies that maximum cpu load on a idle or running node. If the node's load reaches or exceeds this value, Moab will mark the node busy (See Load Balancing Features)
NODEMAXLOAD 0.75

Moab will adjust the state of all idle and running nodes with a load >= .75 to the state busy.

   
NODEMEMOVERCOMMITFACTOR
<DOUBLE>
---
The parameter overcommits available and configured memory and swap on a node by the specified factor (for example: mem/swap * factor). Used to show that the node has more mem and swap than it really does. Only works for PBS RM types.
NODEMEMOVERCOMMITFACTOR .5

Moab will overcommit the memory and swap of the node by a factor of 0.5.

   
NODEPOLLFREQUENCY
<INTEGER>
0 (Poll Always)
Specifies the number of scheduling iterations between scheduler initiated node manager queries. If set to '-2, Moab will never query the node manager daemons. If set to '-1', Moab will only query on the first iteration. Note: this parameter is most often used with OpenPBS and PBSPro. It is not required when using TORQUE, LoadLeveler, LSF, or SGE as the resource managers.
NODEPOLLFREQUENCY 5

Moab will update node manager based information every 5 scheduling iterations.

   
NODEPURGETIME
[[[DD:]HH:]MM:]SS
--- (never purge node info)
The amount of time Moab will keep a node record which is no longer reported by the resource manager. Useful when using a resource manager which drops information about a node due to internal failures.
NODEPURGETIME 00:05:00

Moab will maintain a node record for 5 minutes after the last update regarding that object received from the resource manager.

   
NODESETATTRIBUTE
one of ARCH, CLASS, FEATURE, MEMORY, or PROCSPEED
---
Specifies the type of node attribute by which node set boundaries will be established. See Node Set Overview.
NODESETPOLICY     ONEOF
NODESETATTRIBUTE  PROCSPEED

Moab will create node sets containing nodes with common processor speeds

   
NODESETDELAY
[[[DD:]HH:]MM:]SS
0:00:00
Specifies the length of time Moab will delay a job if adequate idle resources are available but are not available within node set constraints. Setting NODESETDELAY to any non-zero value will force Moab to always use nodesets. A value of zero will cause Moab to use nodesets on a best effort basis. Note: This attribute is deprecated - use NODESETISOPTIONAL instead. See Node Set Overview for more information.
NODESETATTRIBUTE PROCSPEED
NODESETDELAY     0:00:00

Moab will create node sets containing nodes with common processor speeds. Moab will allocate resources within the nodeset if possible. If no resources exist within a single nodeset, Moab will attempt to run the job using any available resources.

   
NODESETISOPTIONAL
<BOOLEAN>
TRUE
Specifies whether or not Moab will start a job if a requested node set cannot be satisfied. See Node Set Overview.
NODESETISOPTIONAL TRUE

Moab will not block a job from running if its node set cannot be satisfied.

   
NODESETLIST
<ATTR>[{ :,|}<ATTR>]...
---
Specifies the list of node attribute values which will be considered for establishing node sets. See Node Set Overview.
NODESETPOLICY     ONEOF
NODESETATTRIBUTE  FEATURE
NODESETLIST       switchA,switchB

Moab will allocate nodes to jobs either using only nodes with the switchA feature or using only nodes with the switchB feature.

   
NODESETPLUS
<STRING>
---
Specifies how Moab distributes jobs among nodesets. The two values for this parameter are DELAY and SPANEVENLY. See Node Set Overview.
NODESETPLUS SPANEVENLY

Moab attempts to fit all jobs on a single nodeset or to span them evenly across a number of nodesets, unless doing so would delay a job beyond the requested NODESETDELAY.

   
NODESETPOLICY
ANYOF, FIRSTOF, or ONEOF
---
Specifies how nodes will be allocated to the job from the various node set generated. See Node Set Overview.
NODESETPOLICY     ONEOF
NODESETATTRIBUTE  NETWORK

Moab will create node sets containing nodes with common network interfaces.

   
NODESETPRIORITYTYPE
one of AFFINITY, BESTFIT, WORSTFIT, BESTRESOURCE, or MINLOSS
MINLOSS
Specifies how resource sets will be selected when more than one feasible resource can can be found. See Node Set Overview.
NODESETPRIORITYTYPE BESTRESOURCE
NODESETATTRIBUTE    PROCSPEED

Moab will select the resource set containing the fastest nodes available.

   
NODESETTOLERANCE
<FLOAT>
0.0 (Exact match only)
Specifies the tolerance for selection of mixed processor speed nodes. A tolerance of X allows a range of processors to be selected subject to the constraint .

(Speed.Max - Speed.Min) / Speed.Min <= X

Note: Tolerances are only applicable when NODESETATTRIBUTE is set to PROCSPEED.
See Node Set Overview.

NODESETATTRIBUTE PROCSPEED
NODESETTOLERANCE 0.5

Moab will only allocate nodes with up to a 50% procspeed difference.

   
NODESYNCTIME
[[[DD:]HH:]MM:]SS
00:10:00
Specifies the length of time after which Moab will sync up a node's expected state with an unexpected reported state. IMPORTANT Note: Moab will not start new jobs on a node with an expected state which does not match the state reported by the resource manager. 
NODESYNCTIME 1:00:00
   
NODETOJOBATTRMAP
Comma delimited list of node features
---
Job requesting the listed node features will be assigned a corresponding job attribute. These job attributes can be used to enable reservation access, adjust job priority or enable other capabilities.
NODETOJOBATTRMAP  fast,big

Jobs requesting node feature fast or big will be assigned a corresponding job attribute.

   
NODEUNTRACKEDRESDELAYTIME
[[[DD:]HH:]MM:]SS
0:00:00
Length of time Moab will assume untracked generic resources will remain unavailable for scheduling if a system reservation is not explicitly created for the node.
NODEUNTRACKEDRESDELAYTIME 0:30:00

Moab will assume untracked generic resources are not available for scheduling for at least 30 minutes from the current time. Thus, these nodes will never be allocated to starting jobs. Also, these nodes will only be available for reservations starting more than 30 minutes in the future.

If NODEUNTRACKEDRESDELAYTIME is enabled and there is an untracked resource preventing a job from running, then the job remains in the idle queue instead of being deferred.

   
NODEWEIGHT
<INTEGER>
0
Specifies the weight which will be applied to a job's requested node count before this value is added to the job's cumulative priority. Note : this weight currently only applies when a nodecount is specified by the user job. If the job only specifies tasks or processors, no node factor will be applied to the job's total priority. This will be rectified in future versions.
NODEWEIGHT 1000
   
NOJOBHOLDNORESOURCES
<BOOLEAN>
FALSE
If TRUE, Moab does not place a hold on jobs that don't have feasible resources. For example, suppose there are 20 processors available for ClassA and 50 processors for the entire system. If a job requests 21 or more processors from ClassA, or 51 or more processors from the entire system, Moab idles the job (instead of putting a hold on it) until the resources become available.
NOJOBHOLDNORESOURCES TRUE
   
NOTIFICATIONPROGRAM
<STRING>
---
Specifies the name of the program to handle all notification call-outs.
NOTIFICATIONPROGRAM   tools/notifyme.pl
   
NOWAITPREEMPTION
<BOOLEAN>
---
Generally when a job is trying to preempt another, it just waits for the original jobs that it chose to preempt to end. If this parameter is on, the preemptor will continue trying to preempt jobs until it can get in.
NOWAITPREEMPTION   TRUE
   
OPTIMIZEDCHECKPOINTING
<BOOLEAN>
FALSE
If TRUE, Moab checkpoints VPCs on creation, modification and shutdown only, instead of in its normal checkpointing interval. OPTIMIZEDCHECKPOINTING only works when using a database.
OPTIMIZEDCHECKPOINTING TRUE
   
OSCREDLOOKUP
NEVER
---
Disables all Moab OS credential lookups, including UID, GID, user to group mappings, and any other OS specific information.
OSCREDLOOKUP NEVER
   
PARALLOCATIONPOLICY
One of BestFit, BestFitP, FirstStart, FirstCompletion, LoadBalance, LoadBalanceP, or RoundRobin
FirstStart
Specifies the approach to use to allocate resources when more than one eligible partition can be found. See Grid Scheduling Policies for more information.
PARALLOCATIONPOLICY  LOADBALANCE

New jobs will be started on the most lightly allocated partition.

   
PARCFG
NODEPOWEROFFDURATION, NODEPOWERONDURATION, or one or more key-value pairs as described in the Partition Overview.
---
Specifies the attributes, policies, and constraints for the given partition.
PARCFG[oldcluster] MAX.WCLIMIT=12:00:00

Moab will not allow jobs to run on the oldcluster partition which has a wallclock limit in excess of 12 hours.

   
PBSACCOUNTINGDIR
<PATH>
---
When specified, Moab will write out job events in standard PBS/TORQUE tracejob format to the specified directory using the standard PBS/TORQUE log file naming convention.
PBSACCOUNTINGDIR /var/spool/torque/sched_logs/

Job events will be written to the specified directory (can be consumed by PBS's tracejob command).

   
PEWEIGHT
<INTEGER>
0
Specifies the coefficient to be multiplied by a job's PE (processor equivalent) priority factor.
RESWEIGHT 10
PEWEIGHT  100

Each job's priority will be increased by 10 * 100 * its PE factor.

   
PREEMPTPOLICY
one of the following:
CANCEL, REQUEUE, SUSPEND, or CHECKPOINT
REQUEUE
Specifies how preemptable jobs will be preempted.

Note: If this policy is set to REQUEUE, preemptible jobs should be marked as RESTARTABLE. If this policy is set to SUSPEND, preemptible jobs shold be marked as SUSPENDABLE. Note: Moab uses preemption escalation to preempt resources if the specified preemption facility is not applicable. This means if the policy is set to SUSPEND and the job is not SUSPENDABLE, Moab may attempt to requeue or even cancel the job.

PREEMPTPOLICY CHECKPOINT

Jobs that are to be preempted will be checkpointed and restarted at a later time.

   
PREEMPTPRIOJOBSELECTWEIGHT
<DOUBLE>
256.0
Adjusts the impact of job run priority versus job size in selecting which jobs to preempt. Setting the value higher increases the impact of priority; setting it lower decreases the impact.

If set to 0, the cost of the job is the job's size.
If set to -1, the cost of the job is the wasted resource percentage.

PREEMPTPRIOJOBSELECTWEIGHT 220.5
   
PREEMPTRTIMEWEIGHT
<DOUBLE>
0
If set to to anything other than 0, a job's remaining time is added into the calculation of which jobs will be preempted. If a positive weight is specified, jobs with a longer remaining time are favored. If a negative weight is specified, jobs with a shorter remaining time are favored.
PREEMPTRTWEIGHT 1
   
PREEMPTSEARCHDEPTH
<INTEGER>
unlimited
Specifies how many preemptible jobs will be evaluated as potential targets for serial job preemptors. See Preemption Overview for more information.
PREEMPTSEARCHDEPTH 8

Serial job preemptors will only consider the first 8 feasible preemptee jobs when determining the best action to take.

   
PRIORITYTARGETDURATION
[[[DD:]HH:]MM:]SS
---
Specifies the ideal job duration which will maximize the value of the WALLTIMEWEIGHT priority factor. If specified, this factor will be calculated as the distance from the ideal. Consequently, in most cases, the associated subcomponent weight should be set to a negative value.
WALLTIMEWEIGHT         -2500
PRIORITYTARGETDURATION  1:00:00
   
PRIORITYTARGETPROCCOUNT
<INTEGER>{+|-|%}
---
Specifies the ideal job requested proc count which will maximize the value of the PROCWEIGHT priority factor. If specified, this factor will be calculated as the distance from the ideal (proc count - ideal = coefficient of PROCWEIGHT). Consequently, in most cases, the associated subcomponent weight should be set to a negative value.
PROCWEIGHT               -1000
PRIORITYTARGETPROCCOUNT  64
   
PROCWEIGHT
<INTEGER>
0
Specifies the coefficient to be multiplied by a job's requested processor count priority factor.
PROCWEIGHT 2500
   
PROFILECOUNT
<INTEGER>
600
Specifies the number of statistical profiles to maintain.
PROFILECOUNT 300
   
PROFILEDURATION
[[[DD:]HH:]MM:]SS
00:30:00
Specifies the duration of each statistical profile. The duration cannot be more than 24 hours, and any specified duration must be a factor of 24. For example, factors of 1/4, 1/2, 1, 2, 3, 4, 6, 8, 12, and 24 are acceptable durations.
PROFILEDURATION 24:00:00
   
PURGETIME
[[[DD:]HH:]MM:]SS
0
The amount of time Moab will keep a job or node record for an object no longer reported by the resource manager. Useful when using a resource manager which 'drops' information about a node or job due to internal failures.Note: This parameter is superseded by JOBPURGETIME and NODEPURGETIME.
PURGETIME 00:05:00

Moab will maintain a job or node record for 5 minutes after the last update regarding that object received from the resource manager.

   
QOSCFG[<QOSID>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
General Credential Flags, PRIORITY, ENABLEPROFILING, FSTARGET, MEMBERULIST, QTWEIGHT, QTTARGET, XFWEIGHT, XFTARGET, PREEMPTMINTIME, PREEMPTMAXTIME, PREEMPTQTTHRESHOLD, PREEMPTXFTHRESHOLD, RMAXPROC, RMAXDURATION RSVQTTHRESHOLD, RSVXFTHRESHOLD, ACLBLTHRESHOLD, ACLQTTHRESHOLD, ACLXFTHRESHOLD, PLIST, PDEF, QFLAGS, TRIGGER, or a usage limit.
---
Specifies QOS specific attributes. See the flag overview for a description of legal flag values.See the QOS Overview section for further details.
QOSCFG[commercial] PRIORITY=1000 MAXJOB=4 MAXPROC=80

Moab will increase the priority of jobs using QOS commercial, and will allow up to 4 simultaneous QOS commercial jobs with up to 80 total allocated processors.

   
QOSISOPTIONAL
<BOOLEAN>
FALSE
An entity's default QOS will be the first QOS specified in the QLIST parameter. When this parameter is set to TRUE the default QOS for the associated credential (user, account, class, etc.) will not be automatically set to the first QOS specified in the QLIST.
QOSISOPTIONAL TRUE
USERCFG[bob]  QLIST=high,low

Moab will set the QOSList for user "bob" to high and low but will not set the QDEF. Should "bob" decide to submit to a particular QOS he will have to do so manually.

   
QOSREJECTPOLICY
One or more of CANCEL, HOLD, IGNORE, or MAIL
HOLD (IGNORE for SLURM users)
Specifies the action to take when Moab determines that a job cannot access a requested QoS. CANCEL issues a call to the resource manager to cancel the job. HOLD places a batch hold on the job preventing the job from being further evaluated until released by an administrator. (Note: Administrators can dynamically alter job attributes and possibly fix the job with mjobctl -m.) With IGNORE, Moab will ignore the QoS request and schedule the job using the default QoS for that job. MAIL will send email to both the admin and the user when QoS request violations are detected. Most combinations of attributes may be specified; however, if both MAIL and IGNORE are specified, Moab will not implement MAIL. Similarly, while CANCEL and HOLD are mutually exclusive, CANCEL will supersede HOLD if both are specified. (see JOBREJECTPOLICY).
QOSREJECTPOLICY  MAIL,CANCEL
   
QOSWEIGHT
<INTEGER>
1
Specifies the weight to be applied to the qos priority of each job (see Credential (CRED) Factor).
QOSWEIGHT 10
   
QUEUETIMECAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum allowed absolute pre-weighted queuetime priority factor.
QUEUETIMECAP    10000
QUEUETIMEWEIGHT 10

A job that has been queued for 40 minutes will have its queuetime priority factor calculated as 'Priority = QUEUETIMEWEIGHT * MIN(10000,40)'.

   
QUEUETIMEWEIGHT
<INTEGER>
1
Specifies multiplier applied to a job's queue time (in minutes) to determine the job's queuetime priority factor.
QUEUETIMEWEIGHT 20

A job that has been queued for 4:20:00 will have a queuetime priority factor of 20 * 260.

   
RECORDEVENTLIST
One or more comma (',') or plus ('+') separated events of GEVENT, ALLSCHEDCOMMAND, JOBCANCEL, JOBCHECKPOINT, JOBEND, JOBFAILURE, JOBMIGRATE, JOBMODIFY, JOBPREEMPT, JOBREJECT, JOBRESUME, JOBSTART, JOBSUBMIT, NODEDOWN, NODEFAILURE, NODEUP, QOSVIOLATION, RMDOWN, RMPOLLEND, RMPOLLSTART, RMUP, RSVCANCEL, RSVCREATE, RSVEND, RSVMODIFY, RSVSTART, SCHEDCOMMAND, SCHEDCYCLEEND, SCHEDCYCLESTART, SCHEDPAUSE, SCHEDSTART, SCHEDSTOP, VMCREATE, VMDESTROY, VMMIGRATE, or ALL
JOBSTART, JOBCANCEL, JOBEND, JOBFAILURE, SCHEDPAUSE, SCHEDSTART, SCHEDSTOP, TRIGEND, TRIGFAILURE, TRIGSTART
Specifies which events should be recorded in the appropriate event file found in Moab's stats/ directory. These events are recorded for both local and remotely staged jobs. (See Event Log Overview) Note: If a plus character is included in the list, the specified events will be added to the default list; otherwise, the specified list will replace the default list.
RECORDEVENTLIST JOBSTART,JOBCANCEL,JOBEND

When a local and/or remote job starts, is canceled, or ends, the respective event will be recorded.

   
REJECTINFEASIBLEJOBS
<BOOLEAN>
FALSE
If zero feasible nodes are found for a job among the currently available nodes on the cluster, the scheduler rejects the job. See JOBREJECTPOLICY for more information.
REJECTINFEASIBLEJOBS TRUE
JOBREJECTPOLICY      MAIL,CANCEL

Any job with zero feasible nodes for execution will be rejected.

   
REJECTNEGPRIOJOBS
<BOOLEAN>
TRUE
If enabled, the scheduler will refuse to start any job with a negative priority. See Job Priority Overview and ENABLENEGJOBPRIORITY for more information.
ENABLENEGJOBPRIORITY TRUE 
REJECTNEGPRIOJOBS    TRUE

Any job with a priority less than zero will be rejected.

   
REMAPCLASS
<ClassID>
---
Specifies which class/queue will be remapped based on the processors, nodes, and node features requested and the resource limits of each class. See Remap Class Overview for more information.
REMAPCLASS batch
CLASSCFG[small]  MAX.PROC=2
CLASSCFG[medium] MAX.PROC=16
CLASSCFG[large]  MAX.PROC=1024

Class batch will be remapped based on the number of processors requested.
   
REMAPCLASSLIST
Comma delimited list of class names
---
Specifies the order in which classes will be searched when attempting to remap a class. Only classes included in the list will be searched and Moab will select the first class with matches.  Note: If no REMAPCLASSLIST is specified, Moab will search all classes and will search them in the order they are discovered. See Remap Class Overview for more information.
REMAPCLASS     batch
REMAPCLASSLIST short,medium,long

Class batch will be re-mapped to one of the listed classes.

   
REMOTEFAILTRANSIENT
<BOOLEAN>
FALSE
Only applicable to Moab configurations with multiple resource managers able to run jobs (such as in a grid environment). When Moab attempts to migrate a job to one of these resource managers, a remote failure may occur. For example, a destination peer in a grid that has an error accepting a job results in a remote error, and the job is rejected. REMOTEFAILTRANSIENT controls how Moab reacts to remote errors. By default, Moab considers such an error permanent and does not try to migrate the same job to that resource manager again. If REMOTEFAILTRANSIENT is set to TRUE, then Moab considers such an error as transient and will not exclude the erring resource manager in future migration attempts.
REMOTEFAILTRANSIENT    TRUE
   
REMOVETRIGOUTPUTFILES
<BOOLEAN>
FALSE
When Moab launches external trigger actions, the standard output and error of those trigger actions are redirected to files located in Moab's spool directory. By default, these files are cleaned every 24 hours. (Files older than 24 hours are removed.) If, however, you wish to have Moab immediately remove the spool files after they are no longer needed, set RemoveTrigOutputFiles to TRUE.
REMOVETRIGOUTPUTFILES  TRUE
   
RESCAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum allowed absolute pre-weighted job resource priority factor.
RESCAP 1000

The total resource priority factor component of a job will be bound by +/- 1000

   
RESERVATIONDEPTH[X]
<INTEGER>
1
Specifies the number of priority reservations which are allowed in the associated reservation bucket. Note: The array index, X, is the bucket label and can be any string up to 64 characters. This label should be synchronized with the RESERVATIONQOSLIST parameter. See Reservation Policies.
RESERVATIONDEPTH[bigmem]   4
RESERVATIONQOSLIST[bigmem] special,fast,joshua

Jobs with QOS's of special, fast, or joshua can have a cumulative total of up to 4 priority reservations.

   
RESERVATIONPOLICY
One of the following: CURRENTHIGHEST, HIGHEST, NEVER
CURRENTHIGHEST
Specifies how Moab reservations will be handled. (See also RESERVATIONDEPTH) See Reservation Policies.
RESERVATIONPOLICY          CURRENTHIGHEST
RESERVATIONDEPTH[DEFAULT]  2

Moab will maintain reservations for only the two currently highest priority jobs.

   
RESERVATIONQOSLIST[X]
One or more QOS values or [ALL]
[ALL]
Specifies which QOS credentials have access to the associated reservation bucket Note: The array index, X, is the bucket label and can be any string up to 64 characters. This label should be synchronized with the RESERVATIONDEPTH parameter.  See Reservation Policies.
RESERVATIONDEPTH[big]   4
RESERVATIONQOSLIST[big] hi,low,med

Jobs with QOS's of hi, low, or med can have a cumulative total of up to 4 priority reservations.

   
RESERVATIONRETRYTIME
[[[DD:]HH:]MM:]SS
60 seconds
Period of time Moab will continue to attempt to allocate resources to start a job after the time resources should be made available. This parameter takes into account resource manager node state race conditions, nodes with residual high load, network glitches, etc. See Reservation Policies and JOBRETRYTIME.
RESERVATIONRETRYTIME   00:05:00

Moab will try for up to 5 minutes to maintain immediate reservations if the reservations are blocked due to node state, network, or batch system based race conditions.

   
RESOURCELIMITMULTIPLIER[<PARID>]
<RESOURCE>:<MULTIPLIER>[,...]

Where <RESOURCE> is one of the following:
NODE, PROC, JOBPROC, MEM, JOBMEM, SWAP, DISK, or WALLTIME
1.0
If set to less than one, then the hard limit will be the specified limit and the soft limit will be the specified limit multiplied by the multiplier. If set to a value greater than one, then the specified limit will be the soft limit and the hard limit will be the specified limit multiplied by the multiplier. See Resource Usage Limits.
RESOURCELIMITMULTIPLER  PROC:1.1,MEM:2.0

Sets hard limit for PROC at 1.1 times the PROC soft limit, and the hard limit of MEM to 2.0 times the MEM soft limit.

   
RESOURCELIMITPOLICY
<RESOURCE>:[<SPOLICY>,]<HPOLICY>
:[<SACTION>,]<HACTION>
[:[<SVIOLATIONTIME>,]<HVIOLATIONTIME>]...

Where RESOURCE is one of PROC, JOBPROC, JOBMEM, DISK, SWAP, MEM or WALLTIME, where *POLICY is one of ALWAYS, EXTENDEDVIOLATION, or BLOCKEDWORKLOADONLY and where *ACTION is one of CANCEL, CHECKPOINT, NOTIFY, REQUEUE, SIGNAL, or SUSPEND.

No limit enforcement.
Specifies how the scheduler should handle jobs which utilize more resources than they request. See Resource Usage Limits.
RESOURCELIMITPOLICY  MEM:ALWAYS,BLOCKEDWORKLOADONLY:REQUEUE,CANCEL

Moab will cancel all jobs which exceed their requested memory limits.

   
RESOURCELIST
<node>[[,<node>]...]
---
Specifies which resources will have their information gathered from the NATIVE RM interface.
RESOURCELIST  node01,node05,node06

Moab will query the native rm for information about nodes node01, node05, and node06.

   
RESTARTINTERVAL
[[[DD:]HH:]MM:]SS
---
Causes Moab daemon to recycle/restart when the given interval of time has transpired.
RESTARTINTERVAL  20:00:00

Moab daemon will automatically restart every 20 hours.

   
RESOURCEQUERYDEPTH
<INTEGER>
3
Maximum number of options which will be returned in response to an mshow -a resource query.
RESOURCEQUERYDEPTH  1

The 'mshow -a' command will return at most one valid collection of resources.

   
RESWEIGHT
<INTEGER>
1
All resource priority components are multiplied by this value before being added to the total job priority. See Job Prioritization.
RESWEIGHT  5
MEMWEIGHT  10
PROCWEIGHT 100
SWAPWEIGHT 0
RESCAP     2000

The job priority resource factor will be calculated as MIN(2000,5 * (10 * JobMemory + 100 * JobProc)).

   
RMCFG
One or more key-value pairs as described in the Resource Manager Configuration Overview
N/A
Specifies the interface and policy configuration for the scheduler-resource manager interface. Described in detail in the Resource Manager Configuration Overview.
RMCFG[TORQUE3] TYPE=PBS

The PBS server will be used for resource management.

   
RMMSGIGNORE
<BOOLEAN>
FALSE
Specifies whether or not Moab should adjust node state based on generic resource manager failure messages. See RM Health Check for more info.
RMMSGIGNORE TRUE
Moab will load and report resource manager failure messages but will not adjust node state as a result of them.
   
RMPOLLINTERVAL
[<MINPOLLTIME>,]<MAXPOLLTIME> where poll time is specified as [[[DD:]HH:]MM:]SS
00:00:30
Specifies interval between RM polls. The poll interval will be no less than MINPOLLTIME and no more than MAXPOLLTIME. A single interval interval (interpreted by Moab as the maximum interval) can be used, or you can specify a minimum and maximum interval. If using a single interval, Moab sometimes has iterations of less than the specified interval.
RMPOLLINTERVAL 30,45

Moab will refresh its resource manager information between a minimum of 30 seconds and a maximum of 45 seconds. Note: This parameter specifies the default global poll interval for all resource managers.

   
RMRETRYTIMECAP
[[[DD:]HH:]MM:]SS
1:00:00

Moab attempts to contact RMs that are in state 'corrupt' (not down). If the attempt is unsuccessful, Moab tries again later. If the second attempt is unsuccessful, Moab increases the gap (the gap grows exponentially) between communication attempts. RMRETRYTIMECAP puts a cap on the length between connection attempts.

RMRETRYTIMECAP 24:00:00
Moab stops increasing the gap between connection attempts once the retry gap reaches 24 hours.
   
RSVCTLPOLICY
ADMINONLY or ANY
ADMINONLY
Specifies who can create admin reservations.
RSVCTLPOLICY ANY
Any valid user can create an arbitrary admin reservation.
   
RSVLIMITPOLICY
HARD or SOFT
---
Specifies what limits should be enforced when creating reservations.
RSVLIMITPOLICY  HARD
Moab will limit reservation creation based on the HARD limits configured.
   
RSVPROFILE[X]

One or more of the following <ATTR>=<VALUE> pairs:

Allowed:

TRIGGER
ACL (ACCOUNTLIST, CLASSLIST, GROUPLIST, MAXTIME, QOSLIST, USERLIST)
HostExp (HOSTLIST)
Features (NODEFEATURES)
FLAGS
TASKCOUNT
RSVACCESSLIST

Note: Lists of more than one ACL value cannot be whitespace delimited. Such lists must be delimited with either the comma, pipe, or colon characters.

Not allowed:

ACCESS
CHARGEACCOUNT

DAYS
DEPTH
ENDTIME
OWNER
PARTITION
PERIOD
PRIORITY
RESOURCES
STARTTIME
TPN
---
Specifies attributes of a reservation profile using syntax similar to that for specifying a standing reservation. See Using Reservation Profiles for details.
RSVPROFILE[fast] USERLIST=john,steve
RSVPROFILE[fast] QOSLIST=high,low
RSVPROFILE[fast] TRIGGER=ETYPE=start,OFFSET=5:00,ATYPE=exec,ACTION="/opt/moab/rp.pl"

Moab will create a reservation profile including trigger and ACL information.

   
RSVREALLOCPOLICY
One of FAILURE, NEVER, OPTIMAL, PRESTARTFAILURE, PRESTARTOPTIMAL, or REMAP
NEVER
Specifies the policy Moab uses to dynamically reallocate nodes to an existing reservation. See Reservation Resource Allocation Policy for more information.
RSVREALLOCPOLICY failure

Should any node go down within a reservation Moab will attempt to allocate a replacement. Moab will NOT allocate a busy or otherwise conflicting node. If Moab should allocate any node for replacement the IGNRSV flag must be used when creating the reservation.

   
RSVSEARCHALGO
LONG or WIDE
LONG
When Moab is determining when and where a job can run, it either searches for the most resources (WIDE) or the longest range of resources (LONG). In almost all cases, searching for the longest range is ideal and returns the soonest starttime. In some rare cases, however, a particular job may need to search for the most resources. In those cases sites can configure this parameter to prevent the starvation of large jobs that fail to hold onto their reservation starttimes.
RSVSEARCHALGO WIDE
   
SCHEDCFG
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following: CHARGEMETRICPOLICY, CHARGERATEPOLICY, FBSERVER, FLAGS, MAXRECORDEDCJOBID, MINJOBID, HTTPSERVERPORT, MODE, RECOVERYACTION, SERVER, or TRIGGER
---
Specifies scheduler policy and interface configuration
SCHEDCFG[zylem3] SERVER=geronimo.scc.com:3422 MODE=NORMAL
Moab will execute in NORMAL mode on the host geronimo.scc.com.
   
SERVERHOST
<HOSTNAME>
---
Deprecated. Hostname of machine on which moab will run. See SCHEDCFG for replacement parameter.
SERVERHOST geronimo.scc.edu
Moab will execute on the host geronimo.scc.edu.
   
SERVERMODE
One of the following:
INTERACTIVE, MONITOR, NORMAL, SIMULATION, or SLAVE
NORMAL
Deprecated. Specifies how Moab interacts with the outside world. See SCHEDCFG for replacement parameter.
SERVERMODE SIMULATION
   
SERVERNAME
<STRING>
<SERVERHOST>
Specifies the name the scheduler will use to refer to itself in communication with peer daemons. See SCHEDCFG for replacement parameter.
SERVERNAME moabA
   
SERVERPORT
<INTEGER> (range: 1-64000)
40559
Port on which Moab will open its user interface socket. See SCHEDCFG for replacement parameter.
SERVERPORT 30003
Moab will listen for client socket connections on port 30003.
   
SERVICEWEIGHT
<INTEGER>
1
Specifies the service component weight associated with the service factors. See Service (SERV) Factor for more information.
SERVICEWEIGHT 2
   
SHOWMIGRATEDJOBSASIDLE
<BOOLEAN>
FALSE
By default, migrated jobs in the grid will show as blocked. This is to prevent jobs from counting against the idle policies of multiple clusters rather than just the cluster to which the job was migrated.
SHOWMIGRATEDJOBSASIDLE TRUE

When set to TRUE, migrated jobs will show as idle and will count against the idle policies of the cluster showing the job as migrated.

   
SIMAUTOSHUTDOWN
<BOOLEAN>
TRUE
If TRUE, the scheduler will end simulations when the active queue and idle queue become empty.
SIMAUTOSHUTDOWN TRUE

The simulation will end as soon as there are no jobs running and no idle jobs which could run.

   
SIMCPUSCALINGPERCENT
<INTEGER>
100 (no scaling)
Specifies whether to increase or decrease the runtime and wallclock limit of each job in the workload trace file.

   
SIMDEFAULTJOBFLAGS
One or more of the following:
ADVRES, HOSTLIST, RESTARTABLE, PREEMPTEE , DEDICATED, PREEMPTOR
---
Cause Moab to force the specified job flags on all jobs supplied in the workload trace file.
SIMDEFAULTJOBFLAGS DEDICATED

Moab will set the DEDICATED job flag on all jobs loaded from the workload trace file.

   
SIMEXITITERATION
<INTEGER>
0 (no exit iteration)
Iteration on which a Moab simulation will create a simulation summary and exit.
SIMEXITITERATION 36000
   
SIMFLAGS
Comma delimited list of zero or more of the following:
IGNHOSTLIST, IGNCLASS, IGNQOS, IGNMODE, IGNFEATURES, IGNGRES
---
Controls how Moab handles trace based simulation information (see Simulation Overview).
SIMFLAGS IGNHOSTLIST,IGNQOS

Moab will ignore hostlist and qos requests specified in the workload trace file.

   
SIMIGNOREJOBFLAGS
One or more of the following:
ADVRES, HOSTLIST, RESTARTABLE, PREEMPTEE , DEDICATED, PREEMPTOR
---
Moab will ignore specified job flags if supplied in the workload trace file (see Simulation Overview).
SIMIGNOREJOBFLAGS DEDICATED

Moab will ignore the DEDICATED job flag if specified in any job trace.

   
SIMINITIALQUEUEDEPTH
<INTEGER>
16
Specifies how many jobs the simulator will initially place in the idle job queue (see Simulation Overview).
SCHEDCFG[sim1]         MODE=SIMULATION
SIMINITIALQUEUEDEPTH   64
SIMJOBSUBMISSIONPOLICY CONSTANTJOBDEPTH

Moab will initially place 64 idle jobs in the queue and, because of the specified queue policy, will attempt to maintain this many jobs in the idle queue throughout the duration of the simulation.

   
SIMJOBSUBMISSIONPOLICY
One of the following:
NORMAL, CONSTANTJOBDEPTH, CONSTANTPSDEPTH, or REPLAY
CONSTANTJOBDEPTH
Specifies how the simulator will submit new jobs into the idle queue. NORMAL mode causes jobs to be submitted at the time recorded in the workload trace file, CONSTANTJOBDEPTH and CONSTANTPSDEPTH attempt to maintain an idle queue of SIMINITIALQUEUEDEPTH jobs and procseconds respectively. REPLAY will force jobs to execute at the exactly the time specified in the simulation job trace file. This mode is most often used to generate detailed profile statistics for analysis in Moab Cluster Manager (see Simulation Overview).
SIMJOBSUBMISSIONPOLICY NORMAL

Moab will submit jobs with the relative time distribution specified in the workload trace file.

   
SIMNODECONFIGURATION
UNIFORM or NORMAL
NORMAL
Specifies whether or not Moab will filter nodes based on resource configuration while running a simulation.
SIMNODECONFIGURATION  UNIFORM
   
SIMNODECOUNT
<INTEGER>
0 (no limit)
Specifies the maximum number of nodes Moab will load from the simulation resource file.
SIMNODECOUNT  256
   
SIMPURGEBLOCKEDJOBS
<BOOLEAN>
TRUE
Specifies whether Moab should remove jobs which can never execute (see Simulation Overview).
SIMPURGEBLOCKEDJOBS  FALSE
   
SIMRESOURCETRACEFILE
<STRING>
Traces/resource.trace
Specifies the file from which Moab will obtain node information when running in simulation mode. Moab will attempt to locate the file relative to <MOABHOMEDIR> unless specified as an absolute path. See Simulation Overview and Resource Trace Format.
SIMRESOURCETRACEFILE traces/nodes.1

Moab will obtain resource traces when running in simulation mode from the <MOABHOMEDIR>/traces/nodes.1 file.

   
SIMRMRANDOMDELAY
<INTEGER>
0
Specifies the random delay added to the RM command base delay accumulated when making any resource manager call in simulation mode.
SIMRMRANDOMDELAY  5
Moab will add a random delay of between 0 and 5 seconds to the simulated time delay of all RM calls.
   
SIMSTARTTIME
[HH[:MM[:SS]]][_MO[/DD[/YY]]]
---
Specifies the time when the simulation starts.
SIMSTARTTIME 00:00:00_01/01/00
Moab will set its clock to January 1, 2000 at 12:00:00 in the morning before starting the simulation
   
SIMSTOPITERATION
<INTEGER>
-1 (don't stop)
Specifies on which scheduling iteration a Moab simulation will stop and wait for a command to resume scheduling. See Simulation Overview for more information.
SIMSTOPITERATION 10
Moab should pause after iteration 10 of simulated scheduling and wait for administrator commands.
   
SIMSTOPTIME
[HH[:MM[:SS]]][_MO[/DD[/YY]]]
---
Specifies the time when the simulation should pause.
SIMSTOPTIME 00:00:00_01/01/04
Moab will stop scheduling when its internal simulation time reaches January 1, 2004.
   
SIMTIMEPOLICY
REAL or NONE
---
Determines simulation time management policy (see SIMTIMERATIO).
SIMTIMEPOLICY REAL
Moab simulation time will advance in line with real time.
   
SIMTIMERATIO
<INTEGER>
0 (no time ratio)
Determines wall time speedup. Simulated Moab time will advance <SIMTIMERATIO> * faster than real wall time (see SIMTIMEPOLICY).
SIMTIMERATIO  10
SIMTIMEPOLICY REAL
Moab simulation time will advance 10 times faster than real world wall time. For example, in 1 hour, Moab will process 10 hours of simulated workload.
   
SIMWORKLOADTRACEFILE
<STRING>
Traces/workload.trace
Specifies the file from which Moab will obtain job information when running in simulation mode. Moab will attempt to locate the file relative to <MOABHOMEDIR> unless specified as an absolute path. See Simulation Overview and Workload Accounting Records.
SIMWORKLOADTRACEFILE traces/jobs.2

Moab will obtain job traces when running in simulation mode from the <MOABHOMEDIR>/traces/jobs.2 file.

   
SPOOLDIR
<STRING>
---
Specifies the directory for temporary spool files created by Moab while submitting a job to the RM.
SPOOLDIR /tmp/moab/spool
   
SPVIOLATIONWEIGHT
<INTEGER>
0
Specifies the weight to be applied to a job which violates soft usage limit policies (see Service Priority Component Overview).
SPVIOLATIONWEIGHT 5000
   
SRCFG[X]
One or more of the following <ATTR>=<VALUE> pairs
ACCESS, ACCOUNTLIST, CHARGEACCOUNT, CHARGEUSER, CLASSLIST, CLUSTERLIST, COMMENT, DAYS, DEPTH, DISABLE, ENDTIME, FLAGS, GROUPLIST, HOSTLIST, JOBATTRLIST, MAXTIME, NODEFEATURES, OWNER, PARTITION, PERIOD, PRIORITY, QOSLIST, RESOURCES, ROLLBACKOFFSET, RSVACCESSLIST, RSVGROUP, STARTTIME, TASKCOUNT, TIMELIMIT, TPN, TRIGGER, or USERLIST

Note: HOSTLIST and ACL list values must be comma delimited. For example: HOSTLIST=nodeA,nodeB

---
Specifies attributes of a standing reservation. See Managing Reservations for details.
SRCFG[fast] STARTTIME=9:00:00 ENDTIME=15:00:00
SRCFG[fast] HOSTLIST=node0[1-4]$
SRCFG[fast] QOSLIST=high,low
Moab will create a standing reservation running from 9:00 AM to 3:00 PM on nodes 1 through 4 accessible by jobs with QOS high or low.

   
STARTCOUNTCAP
<INTEGER>
0
Specifies the max weighted value allowed from the startcount subfactor when determining a job's priority (see Priority Factors for more information).
STARTCOUNTWEIGHT 5000
STARTCOUNTCAP    30000
   
STARTCOUNTWEIGHT
<INTEGER>
0
Specifies the weight to be applied to a job's startcount when determining a job's priority (see Priority Factors for more information).
STARTCOUNTWEIGHT 5000
   
STATDIR
<STRING>
stats
Specifies the directory in which Moab statistics will be maintained.
STATDIR /var/adm/moab/stats
   
STATPROCMAX
<INTEGER>
1

Specifies the maximum number of processors requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATPROCMAX     256
STATPROCSTEPCOUNT 4 
STATPROCSTEPSIZE  4

Each matrix output will display data in rows for jobs requesting between 4 and 256 processors.

Note A NONE in services will still allow users to run showq and checkjob on their own jobs.


STATPROCMIN
<INTEGER>
1

Specifies the minimum number of processors requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATPROCMIN       4
STATPROCSTEPCOUNT 4 
STATPROCSTEPSIZE  4

Each matrix output will display data in rows for jobs requesting between 4 and 256 processors.

Note A NONE in services will still allow users to run showq and checkjob on their own jobs.


   
STATPROCSTEPCOUNT
<INTEGER>
5

Specifies the number of rows of processors requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATPROCMIN       4
STATPROCSTEPCOUNT 4
STATPROCSTEPSIZE  4
Each matrix output will display data in rows for jobs requesting between 4 and 256 processors.
   
STATPROCSTEPSIZE
<INTEGER>
4

Specifies the processor count multiplier for rows of processors requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATPROCMIN       4
STATPROCSTEPCOUNT 4
STATPROCSTEPSIZE  4
Each matrix output will display data in rows for jobs requesting between 4 and 256 processors.
   
STATTIMEMAX
[[DD:]HH:]MM:]SS
00:15:00

Specifies the maximum amount of time requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATTIMEMAX       02:08:00
STATTIMESTEPCOUNT 4
STATTIMESTEPSIZE  4
Each matrix output will display data in columns for jobs requesting between 2 and 128 minutes.
   
STATTIMEMIN
[[DD:]HH:]MM:]SS
00:15:00

Specifies the minimum amount of time requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATTIMEMIN       00:02:00
STATTIMESTEPCOUNT 4
STATTIMESTEPSIZE  4
Each matrix output will display data in columns for jobs requesting between 2 and 128 minutes.
   
STATTIMESTEPCOUNT
<INTEGER>
6

Specifies the number of columns of time requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATTIMEMIN       00:02:00
STATTIMESTEPCOUNT 4
STATTIMESTEPSIZE  4
Each matrix output will display data in columns for jobs requesting between 2 and 128 minutes.
   
STATTIMESTEPSIZE
<INTEGER>
4

Specifies the time multiplier for columns of time requested by jobs to be displayed in matrix outputs (as displayed by the showstats -f command).

Note It is recommended that you not change any parameters via 'mschedctl -m' or 'changeparam' while Moab is running. Changing any of the parameters invalidates all past data and will start the collection over.
STATTIMEMIN       00:02:00
STATTIMESTEPCOUNT 4
STATTIMESTEPSIZE  4
Each matrix output will display data in columns for jobs requesting between 2 and 128 minutes.
   
STRICTPROTOCOLCHECK
<BOOLEAN>
FALSE
Specifies how Moab reacts to differences in XML protocols when communicating with other Moab peers. If set to TRUE, Moab will reject any communication that does not strictly conform to the expected protocol. If set to FALSE (the default), Moab will not reject XML that has extra or unknown attributes.
STRICTPROTOCOLCHECK TRUE
Moab will reject any XML communication that does not strictly conform to the expected protocol definition.
   
SUBMITFILTER
<STRING>
---
Specifies the directory of a given submit filter script.
SUBMITFILTER /home/submitfilter/filter.pl
   
SUBMITHOSTS
space delimited list of host names
---
If specified, SUBMITHOSTS specifies an explicit list of hosts where jobs can be submitted.
SUBMITHOSTS hostA hostB
   
SUSPENDRESOURCES[<PARID>]
<RESOURCE>[,...]

Where <RESOURCE> is one of the following:
NODE, PROC, MEM, SWAP, DISK
---
List of resources to dedicate while a job is suspended (available in Moab version 4.5.1 and higher).
SUSPENDRESOURCES[base]  MEM,SWAP,DISK

While a job is suspended in partition base, the memory, swap and disk for that job will remain dedicated to the job.

   
SYSCFG
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:
PRIORITY, FSTARGET, QLIST, QDEF, PLIST, PDEF, FLAGS, or a fairness policy specification.
---
Specifies system-wide default attributes.  See the Attribute/Flag Overview for more information.
SYSCFG PLIST=Partition1 QDEF=highprio
by default, all jobs will have access to partition Partition1 and will use the QOS highprio.
   
SWAPWEIGHT
<INTEGER>
0
Specifies the priority weight assigned to the virtual memory request of a job.
SWAPWEIGHT 10
   
SYSTEMMAXPROCPERJOB
<INTEGER>
-1 (NO LIMIT)
Specifies the maximum number of processors that can be requested by any single job.
SYSTEMMAXPROCPERJOB 256
Moab will reject jobs requesting more than 256 processors.
   
SYSTEMMAXPROCSECONDPERJOB
<INTEGER>
-1 (NO LIMIT)
Specifies the maximum number of proc-seconds that can be requested by any single job.
SYSTEMMAXJOBPROCSECOND 86400

Moab will reject jobs requesting more than 86400 procs seconds. i.e., 64 processors * 30 minutes will be rejected, while a 2 processor * 12 hour job will be allowed to run.

   
SYSTEMMAXJOBWALLTIME
[[[DD:]HH:]MM:]SS
-1 (NO LIMIT)
Specifies the maximum amount of wallclock time that can be requested by any single job.
SYSTEMMAXJOBWALLTIME 1:00:00:00

Moab will reject jobs requesting more than one day of walltime.

   
TARGETQUEUETIMEWEIGHT
<INTEGER>
0
Specifies the weight assigned to the time remaining until the queuetime is reached.
TARGETQUEUETIMEWEIGHT 10
   
TARGETWEIGHT
<INTEGER>
1
Specifies the weight to be applied to a job's queuetime and expansion factor target components (see Job Prioritization).
TARGETWEIGHT 1000
   
TARGETXFACTORWEIGHT
<INTEGER>
0
Specifies the weight assigned to the distance to the target expansion factor.
TARGETXFACTORWEIGHT  10
   
TASKDISTRIBUTIONPOLICY
One of DEFAULT, PACK, RR (round-robin), or LOCAL
---
Specifies how job tasks should be mapped to allocated resources. See Task Distribution Overview for more information.
TASKDISTRIBUTIONPOLICY DEFAULT
Moab should use standard task distribution algorithms.
   
THREADPOOLSIZE
<INTEGER>
2 (MAX: 25)
Specifies how many threads to have available for threaded operations.
THREADPOOLSIZE 10
   
TOOLSDIR
<STRING>
Tools
Specifies the directory in which Moab tools will be maintained (commonly used in conjunction with Native Resource Managers, and Triggers).
TOOLSDIR /var/adm/moab/tools
   
TRAPFUNCTION
<STRING>
---
Specifies the functions to be trapped.
TRAPFUNCTION UpdateNodeUtilization|GetNodeSResTime
   
TRAPJOB
<STRING>
---
Specifies the jobs to be trapped.
TRAPJOB pros23.0023.0
   
TRAPNODE
<STRING>
---
Specifies the nodes to be trapped.
TRAPNODE node001|node004|node005
   
TRAPRES
<STRING>
---
Specifies the reservations to be trapped.
TRAPRES interactive.0.1
   
TRIGCHECKTIME
<INTEGER> (milliseconds)
2000
Each scheduling iteration, Moab will have a period of time where it handles commands and other UI requests. This time period is controlled by RMPOLLINTERVAL. During this time period, known as the UI phase, Moab will periodically evaluate triggers. Usually this only takes a fraction of a second, but if the number of triggers are large it could take up substantially more time (up to several seconds). While Moab is evaluating triggers, it doesn't respond to UI commands. This makes Moab feel sluggish and unresponsive. To remedy this, use the parameter "TrigCheckTime." This parameter tells Moab to only spend up to X milliseconds processing triggers during the UI phase. After X milliseconds has gone by, Moab will pause the evaluating of triggers, handle any pending UI events, and then restart the trigger evaluations where it last left off.
TRIGCHECKTIME 4000
   
TRIGEVALLIMIT
<INTEGER>
1
Each scheduling iteration, Moab will have a period of time where it handles commands and other UI requests. This time period is controlled by RMPOLLINTERVAL. During this time period, known as the UI phase, Moab will periodically evaluate triggers. The number of times Moab evaluates all triggers in the system is controlled by the "TrigEvalLimit" parameter. By default, this is set to 1. This means that Moab will evaluate all triggers at most once during the UI phase. Moab will not leave the UI phase and start other scheduling tasks until ALL triggers are evaluated at least one time. If TrigEvalLimit is set to 5, then Moab will wait until all triggers are evaluated five times.
TRIGEVALLIMIT   3
   
UJOBWEIGHT
<INTEGER>
0
Weight assigned by jobs per user. -1 will reduce priority by number of active jobs owned by user.
UJOBWEIGHT 10
   
UMASK
<INTEGER>
0022 (octal) (produces 0644 permssions)
Specifies the file permission mask to use when creating new fairshare, stats, and event files. See the umask man page for more details.
UMASK 0127

Create statistics and event files which are 'read-write' by owner and 'read' by group only.

   
UNSUPPORTEDDEPENDENCIES
Comma delimited string
---
Specifies dependencies that are not supported and should not be accepted by job submissions. A maximum of 30 dependencies is supported.
moab.cfg:
UNSUPPORTEDDEPENDENCIES before,beforeok,beforenotok,on

Example:
> msub -l depend=before:105 cmd.sh

ERROR: cannot submit job - error in extension string
   
UPROCWEIGHT
<INTEGER>
0
Weight assigned by processors per user. -1 will reduce priority by number of active procs owned by user.
UPROCWEIGHT 10
   
USAGECONSUMEDWEIGHT
<INTEGER>
0
Specifies the weight assigned to per job processor second consumption.
USAGECONSUMEDWEIGHT 10
   
USAGEEXECUTIONTIMEWEIGHT
<INTEGER>
0
Specifies the priority weight assigned to the total job execution time (measured in seconds since job start). See Preemption Overview.
USAGEEXECUTIONTIMEWEIGHT 10
   
USAGEPERCENTWEIGHT
<INTEGER>
0
Specifies the weight assigned to total requested resources consumed.
USAGEPERCENTWEIGHT 5
   
USAGEREMAININGWEIGHT
<INTEGER>
0
Specifies the weight assigned to remaining usage.
USAGEREMAININGWEIGHT 10
   
USAGEWEIGHT
<INTEGER>
1
Specifies the weight assigned to the percent and total job usage subfactors.
USAGEWEIGHT 100
   
USEANYPARTITIONPRIO
<BOOLEAN>
FALSE
The FSTREE data from the first feasible FSTREE will be used when determining a job's start priority, rather than having no FSTREE data considered.
USEANYPARTITIONPRIO TRUE
   
USECPRSVNODELIST
<BOOLEAN>
TRUE
Specifies whether Moab should use the checkpointed reservation node list when rebuilding reservations on startup. If this is not used then Moab will use the reservation's specified host expression during rebuilding.
USECPRSVNODELIST FALSE
   
USEDATABASE
INTERNAL
-
Specifies whether Moab should store profile statistics, checkpoint information, and event information in an integrated database. See Layout of Scheduler Components with Integrated Database Enabled for more information.
USEDATABASE INTERNAL
   
USEMACHINESPEED
<BOOLEAN>
TRUE
Specifies whether or not job wallclock limits should be scaled by the machine speed of the node(s) they are running on.
USEMACHINESPEED TRUE
Job <X> specifying a wallclock limit of 1:00:00 would be given only 40 minutes to run if started on a node with a machine speed of 1.5.
   
USEMOABCTIME
<BOOLEAN>
FALSE
When Moab finds new jobs on the resource manager, it creates a job inside of Moab for each job in the resource manager. By default, when Moab creates a new job, it uses the time the job was submitted to the resource manager to calculate how long the job has been in the queue (Moab processing time - job creation in resource manager), which is then used in determining the job's priority.

In a system where more jobs are submitted to a resource manager than Moab can handle in one iteration, there is the possibility of jobs running out of order. For example, two jobs are both submitted at time 5. The first submitted job is processed first at time 6. So the first job's effective queue duration would be 1 (6-5). On the next iteration, the second job is processed at time 8. So the second job's effective queue duration would be 3 (8-5), indicating that it has been in the queue longer than the other job. Since the later job has a higher effective queue duration it will get a higher priority and could be scheduled to run before earlier submitted jobs.

Setting USEMOABCTIME to TRUE tells Moab to use the creation time of the job in Moab rather than the creation time in the resource manager. This corrects the possible problem of having later submitted jobs having higher priorities and starting before earlier submitted jobs.

USEMOABCTIME TRUE
   
USEMOABJOBID
<BOOLEAN>
FALSE
Specifies whether to use the Moab job ID, or the resource manager's job ID.
USEMOABJOBID TRUE
   
USENEWXMLVARIABLES
<BOOLEAN>
FALSE
Reports variables as child elements in node XML.
USENEWXMLVARIABLES TRUE
   
USERCFG[<USERID>]
List of zero or more space delimited <ATTR>=<VALUE> pairs where <ATTR> is one of the following:

General Credential Flags, CDEF, DEFAULT.TPN, DEFAULT.WCLIMIT, EMAILADDRESS, ENABLEPROFILING, FSCAP, FSTARGET, JOBFLAGS, MAX.WCLIMIT, QLIST, QDEF, NOEMAIL, OVERRUN, PDEF, PLIST, PREF, PRIORITY, TRIGGER, or a usage limit. VIEWPOINTPROXY is an option available to those using Viewpoint as the front-end user interface. More information on VIEWPOINTPROXY is available in the Viewpoint documentation.
---
Specifies user specific attributes.  For general user attribute information, See the Credential Overview. For a description of legal flag values, see flag overview.
USERCFG[john] MAXJOB=50 QDEF=highprio
USERCFG[john] [email protected]
Up to 50 jobs submitted under the user ID john will be allowed to execute simultaneously and will be assigned the QOS highprio.
   
USERPRIOCAP
<INTEGER>
-
Specifies the priority cap to be applied to the user specified job priority factor. Under Moab, only negative user priorities may be specified. See Credential (Service) Factor.
USERPRIOWEIGHT 10
USERPRIOCAP    -10000
   
USERPRIOWEIGHT
<INTEGER>
0
Specifies the weight to be applied to the user specified job priority. Under Moab, only negative user priorities may be specified. If this weight is set, users may reduce the priority of some of their jobs to allow other jobs to run earlier. See Credential (Service) Factor.
USERPRIOWEIGHT 10
   
USERWEIGHT
<INTEGER>
1
Specifies the weight to be applied to the user priority of each job. See Credential (CRED) Factor.
USERWEIGHT 10
   
USESYSLOG
<BOOLEAN>[<FACILITY>]
FALSE:daemon
Specifies whether or not the scheduler will report key events to the system syslog facility. If the FACILITY is specified, Moab will report events to this syslog facility. See Logging Facilities for more information.
USESYSLOG TRUE:local3

Moab will report key events, commands, and failures to syslog using the local3 facility.

   
USESYSTEMQUEUETIME
<BOOLEAN>
FALSE
Specifies whether or not job prioritization should be based on the time the job has been eligible to run, i.e., idle and meets all fairness policies (TRUE) or the time the job has been idle (FALSE). See Priority Factors for more info. Note: This parameter has been superseded by the JOBPRIOACCRUALPOLICY parameter.
USESYSTEMQUEUETIME FALSE

The queuetime and expansion factor components of a job's priority will be calculated based on the length of time the job has been in the idle state.

   
USEUSERHASH
<BOOLEAN>
FALSE
Enables searching of the user buffer using the user hash key instead of doing sequential searches of the user buffer.
USEUSERHASH TRUE
   
VCPROFILE
Comma delimited list of one or more of the following:
ACL, DEFAULTRSVPROFILE, DESCRIPTION, NODESETLIST, OPRSVPROFILE, QUERYDATA, REQENDPAD, REQSETATTR or REQSTARTPAD
---
Defines virtual cluster attributes (see VPC Overview).
VCPROFILE[pkgA] REQENDPAD=0:10:00

All pkgA VPC's will allocate an additional 10 minutes of time at the end to account for provisioning overhead.

   
VMCALCULATELOADBYVMSUM
<BOOLEAN>
False
When false, vmmigrate using overcommits uses the CPU load from the node to determine if VM's need to be migrated off the hypervisor. When true, overcommit vmmigrates calculates the total node load using the total sum reported by each VM on the hypervisor.
VMCALCULATELOADBYVMSUM  TRUE
   
VMCREATETHROTTLE
<INTEGER>
---
Sets the maximum allowable 'VM create' jobs at any given time.
VMCREATETHROTTLE  25
Only 25 VM creation jobs are allowed in the system at any given time.
   
VMMIGRATETHROTTLE
<INTEGER>
---
Sets the maximum allowable 'VM migrate' jobs at any given time.
VMMIGRATETHROTTLE  20
Only 20 VM migrate jobs are allowed in the system at any given time.
   
VMMIGRATIONPOLICY
<STRING>; values include RSV, GREEN, and OVERCOMMIT
NONE
  • RSV - If RSV flag is set, vacates VMs on a node for a reservation to run.
  • GREEN - If GREEN flag is set, Moab consolidates VMs to allow nodes to go idle.
  • OVERCOMMIT - The OVERCOMMIT flag must be set for the VMOCTHRESHOLD parameter to function.
VMMIGRATIONPOLICY GREEN,OVERCOMMIT
   
VMMINOPDELAY
[HH[:MM[:SS]
--
The minimum time between automatic VM node operations, such as creating, modifying, and destroying VMs. May prevent thrashing.
VMMINOPDELAY 30
   
VMOCTHRESHOLD
MEM:<0-1>,PROCS:<0-1>,DISK:<0-1>,SWAP:<0-1>
 
Percentage threshold at which Moab begins to migrate virtual machines to other nodes. VMMIGRATIONPOLICY must be set to OVERCOMMIT for this to occur.
VMOCTHRESHOLD PROC:0.7,MEM:0.9
When a node surpasses .7 (70%) load of CPU or .9 (90%) of memory, Moab begins to migrate virtual machines to other nodes.
   
VMPROVISIONSTATUSREADYVALUE
<INTEGER>
---
Checks a VM for a special value or values (which Moab gets from the resource manager) and, based on the value, tells Moab that a VM was created..
VMProvisionStatusReadyValue 2
VMProvisionStatusReadyValue 1-4,6,16
   
VMSARESTATIC
<BOOLEAN>
FALSE
When set to true, informs Moab that it can schedule under the assumption that no VMs will be migrated and no new VMs will be created, and disables Moab from scheduling any VM creations or migrations.
VMSARESTATIC   TRUE
   
VMSTORAGEMOUNTDIR
<PATH>
---
The specified path is used as the default location for storage mounts in all newly created VMs (created via the mvmctl command). This parameter defines the default storage mount directory if one is not specified.
VMSTORAGEMOUTDIR   /var/spool
Moab uses /var/spool as a storage mount directory if a storage directory is not submitted (but additional storage is requested) at VM creation.
   
VMTRACKING
<BOOLEAN>
FALSE
When set to TRUE, VMTracking jobs are used to represent VMs in the queue.
VMTRACKING TRUE
   
VPCFLAGS
Comma delimited list of one or more of the following:
CREDS, JOBS, NODES, or STATS
---
Virtual private cluster flags that specify the attributes to restrict to the users in a virtual private cluster.
VPCFLAGS=CREDS,JOBS,NODES
Moab will restrict the viewing of credentials, jobs and nodes to only within a virtual private cluster.
   
WALLTIMECAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum total pre-weighted absolute contribution to job priority which can be contributed by the walltime component. This value is specified as an absolute priority value, not as a percent.
WALLTIMECAP 10000
Moab will bound a job's pre-weighted walltime priority component within the range +/- 10000.
   
WALLTIMEWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to the amount of walltime requested by a job (in seconds) (see Resource (RES) Factor).
RESWEIGHT      10
WALLTIMEWEIGHT 100
Increase the priority of longer duration jobs.
   
WCACCURACYCAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum total pre-weighted absolute contribution to job priority which can be contributed by the wallclock accuracy component. This value is specified as an absolute priority value, not as a percent.
WCACCURACYCAP 10000
Moab will bound a job's pre-weighted wallclock accuracy priority component within the range +/- 10000.
   
WCACCURACYWEIGHT
<INTEGER>
0
Specifies the priority weight to be applied to the job's historical user wallclock accuracy (range 0.0 to 1.0) (see Fairshare (FS) Factor).
FSWEIGHT         10
WCACCURACYWEIGHT 100
Favor jobs with good wallclock accuracies by giving them a priority increase.
   
WCVIOLATIONACTION
one of CANCEL or PREEMPT
CANCEL
Specifies the action to take when a job exceeds its wallclock limit. If set to CANCEL, the job will be terminated. If set to PREEMPT, the action defined by PREEMPTPOLICY parameter will be taken. See JOBMAXOVERRUN or Resource Usage Limit Overview.
WCVIOLATIONACTION   PREEMPT
PREEMPTPOLICY       REQUEUE

Moab will requeue jobs which exceed their wallclock limit.

   
WIKIEVENTS
<BOOLEAN>
FALSE
When set to true, Moab events are set to native wiki format (ATTR=VALUE pairs) to facilitate easier readability .
WIKIEVENTS TRUE

Moab events will generate output in the format of the following sample:

09:26:40 1288279600:5 job 58 JOBEND 58 REQUESTEDNC=1 REQUESTEDTC=3 UNAME=wightman GNAME=wightman 
WCLIMIT=60 STATE=Completed RCLASS=[batch:1] SUBMITTIME=1288279493 RMEMCMP=>= RDISKCMP=>= 
RFEATURES=[NONE] SYSTEMQUEUETIME=1288279493 TASKS=1 FLAGS=RESTARTABLE PARTITION=pbs DPROCS=1 
ENDDATE=2140000000 TASKMAP=proxy,GLOBAL SRM=pbs EXITCODE=0   SID=2357 NODEALLOCATIONPOLICY=SHARED 
EFFECTIVEQUEUEDURATION=107
   
XFACTORCAP
<DOUBLE>
0 (NO CAP)
Specifies the maximum total pre-weighted absolute contribution to job priority which can be contributed by the expansion factor component. This value is specified as an absolute priority value, not as a percent.
XFACTORCAP 10000
Moab will bound a job's pre-weighted XFactor priority component within the range +/- 10000.
   
XFACTORWEIGHT
<INTEGER>
0
Specifies the weight to be applied to a job's minimum expansion factor before it is added to the job's cumulative priority.
XFACTORWEIGHT 1000
Moab will multiply a job's XFactor value by 1000 and then add this value to its total priority.
   
XFMINWCLIMIT
[[[DD:]HH:]MM:]SS
-1 (NO LIMIT)
Specifies the minimum job wallclock limit that will be considered in job expansion factor priority calculations.
XFMINWCLIMIT 0:01:00
Jobs requesting less than one minute of wallclock time will be treated as if their wallclock limit was set to one minute when determining expansion factor for priority calculations.