Visit Platform at http://www.platform.com

parameterInfo Struct Reference

The parameterInfo structure contains the following fields:. More...


Data Fields

char * defaultQueues
char * defaultHostSpec
int mbatchdInterval
int sbatchdInterval
int jobAcceptInterval
int maxDispRetries
int maxSbdRetries
int preemptPeriod
int cleanPeriod
int maxNumJobs
float historyHours
int pgSuspendIt
char * defaultProject
int retryIntvl
int nqsQueuesFlags
int nqsRequestsFlags
int maxPreExecRetry
int localMaxPreExecRetry
int eventWatchTime
float runTimeFactor
float waitTimeFactor
float runJobFactor
int eEventCheckIntvl
int rusageUpdateRate
int rusageUpdatePercent
int condCheckTime
int maxSbdConnections
int rschedInterval
int maxSchedStay
int freshPeriod
int preemptFor
int adminSuspend
int userReservation
float cpuTimeFactor
int fyStart
int maxJobArraySize
time_t exceptReplayPeriod
int jobTerminateInterval
int disableUAcctMap
int enforceFSProj
int enforceProjCheck
int jobRunTimes
int dbDefaultIntval
int dbHjobCountIntval
int dbQjobCountIntval
int dbUjobCountIntval
int dbJobResUsageIntval
int dbLoadIntval
int dbJobInfoIntval
int jobDepLastSub
int maxJobNameDep
char * dbSelectLoad
int jobSynJgrp
char * pjobSpoolDir
int maxUserPriority
int jobPriorityValue
int jobPriorityTime
int enableAutoAdjust
int autoAdjustAtNumPend
float autoAdjustAtPercent
int sharedResourceUpdFactor
int scheRawLoad
char * jobAttaDir
int maxJobMsgNum
int maxJobAttaSize
int mbdRefreshTime
int updJobRusageInterval
char * sysMapAcct
int preExecDelay
int updEventUpdateInterval
int resourceReservePerSlot
int maxJobId
char * preemptResourceList
int preemptionWaitTime
int maxAcctArchiveNum
int acctArchiveInDays
int acctArchiveInSize
float committedRunTimeFactor
int enableHistRunTime
int mcbOlmReclaimTimeDelay
int chunkJobDuration
int sessionInterval
int publishReasonJobNum
int publishReasonInterval
int publishReason4AllJobInterval
int mcUpdPendingReasonInterval
int mcUpdPendingReasonPkgSize
int noPreemptRunTime
int noPreemptFinishTime
char * acctArchiveAt
int absoluteRunLimit
int lsbExitRateDuration
int lsbTriggerDuration
int maxJobinfoQueryPeriod
int jobSubRetryInterval
int pendingJobThreshold
int maxConcurrentJobQuery
int minSwitchPeriod
int condensePendingReasons
int slotBasedParallelSched
int disableUserJobMovement
int detectIdleJobAfter
int useSymbolPriority
int JobPriorityRound
char * priorityMapping
int maxInfoDirs
int minMbdRefreshTime
int enableStopAskingLicenses2LS
int expiredTime
char * mbdQueryCPUs
char * defaultApp
int enableStream
char * streamFile
int streamSize
int syncUpHostStatusWithLIM
char * defaultSLA
int slaTimer
int mbdEgoTtl
int mbdEgoConnTimeout
int mbdEgoReadTimeout
int mbdUseEgoMXJ
int mbdEgoReclaimByQueue
int defaultSLAvelocity
char * exitRateTypes
float globalJobExitRate
int enableJobExitRatePerSlot
int enableMetric
int schMetricsSample
float maxApsValue
int newjobRefresh
int preemptJobType
char * defaultJgrp
int jobRunlimitRatio
int jobIncludePostproc
int jobPostprocTimeout
int sschedUpdateSummaryInterval
int sschedUpdateSummaryByTask
int sschedRequeueLimit
int sschedRetryLimit
int sschedMaxTasks
int sschedMaxRuntime
char * sschedAcctDir
int jgrpAutoDel
int maxJobPreempt
int maxJobRequeue
int noPreemptRunTimePercent
int noPreemptFinishTimePercent
int slotReserveQueueLimit
int maxJobPercentagePerSession
int useSuspSlots
int maxStreamFileNum
int privilegedUserForceBkill
int mcSchedulingEnhance
int mcUpdateInterval
int intersectCandidateHosts
int enforceOneUGLimit
int logRuntimeESTExceeded
char * computeUnitTypes
float fairAdjustFactor
int simAbsoluteTime
int extendJobException

Detailed Description

The parameterInfo structure contains the following fields:.

Field Documentation

DEFAULT_QUEUE: A blank_separated list of queue names for automatic queue selection.

DEFAULT_HOST_SPEC: The host name or host model name used as the system default for scaling CPULIMIT and RUNLIMIT.

MBD_SLEEP_TIME: The interval in seconds at which the mbatchd dispatches jobs.

SBD_SLEEP_TIME: The interval in seconds at which the sbatchd suspends or resumes jobs.

JOB_ACCEPT_INTERVAL: The interval at which a host accepts two successive jobs.

(In units of SBD_SLEEP_TIME.)

MAX_RETRY: The maximum number of retries for dispatching a job.

MAX_SBD_FAIL: The maximum number of retries for reaching an sbatchd.

PREEM_PERIOD: The interval in seconds for preempting jobs running on the same host.

CLEAN_PERIOD: The interval in seconds during which finished jobs are kept in core.

MAX_JOB_NUM: The maximum number of finished jobs that are logged in the current event file.

HIST_HOURS: The number of hours of resource consumption history used for fair share scheduling and scheduling within a host partition.

PG_SUSP_IT: The interval a host must be idle before resuming a job suspended for excessive paging.

The default project assigned to jobs.

Job submission retry interval.

For Cray NQS compatiblilty only.

Used by LSF to get the NQS queue information

nqsRequestsFlags

The maximum number of times to attempt the preexecution command of a job from a remote cluster ( MultiCluster only).

Maximum number of pre-exec retry times for local cluster.

Event watching Interval in seconds.

Run time weighting factor for fairshare scheduling.

Used for calcultion of the fairshare scheduling formula.

Job slots weighting factor for fairshare scheduling.

Default check interval.

sbatchd report every sbd_sleep_time

sbatchd updates jobs jRusage in mbatchd if more than 10% changes

Time period to check for reconfig.

The maximum number of connections between master and slave batch daemons.

The interval for rescheduling jobs.

Max time mbatchd stays in scheduling routine, after which take a breather.

During which load remains fresh.

The preemption behavior, GROUP_MAX, GROUP_JLP, USER_JLP, HOST_JLU,MINI_JOB, LEAST_RUN_TIME.

Flags whether users can resume their jobs when suspended by the LSF administrator.

Flags to enable/disable normal user to create advance reservation.

CPU time weighting factor for fairshare scheduling.

The starting month for a fiscal year.

The maximum number of jobs in a job array.

Replay period for exceptions, in seconds.

The interval to terminate a job.

User level account mapping for remote jobs is disabled.

If set to TRUE, Project name for a job will be considerred when doing fairshare scheduling, i.e., as if user has submitted jobs using -G.

Enforces the check to see if the invoker of bsub is in the specifed group when the -P option is used.

Run time for a job.

Event table Job default interval.

Event table Job Host Count.

Event table Job Queue Count.

Event table Job User Count.

Event table Job Resource Interval.

Event table Resource Load Interval.

Event table Job Info.

Used with job dependency scheduling.

Used with job dependency scheduling, deprecated.

Select resources to be logged.

Job synchronizes its group status.

The batch jobs' temporary output directory.

Maximal job priority defined for all users.

Job priority is increased by the system dynamically based on waiting time.

Waiting time to increase Job priority by the system dynamically.

Enable internal statistical adjustment.

Start to autoadjust when the user has this number of pending jobs.

If this number of jobs has been visited skip the user.

Static shared resource update interval for the cluster actor.

Schedule job based on raw load info.

The batch jobs' external storage for attached data.

Maximum message number for each job.

Maximum attached data size to be transferred for each message.

The life time of a child MBD to serve queries in the MT way.

The interval of the execution cluster updating the job's resource usage.

The account to which all windows workgroup users are to be mapped.

Dispatch delay internal.

Update duplicate event interval.

Resources are reserved for parallel jobs on a per-slot basis.

Maximum job id --- read from the lsb.params.

Define a list of preemptable resource names.

The preemption wait time.

Maximum number of rollover lsb.acct files kept by mbatchd.

mbatchd Archive Interval

mbatchd Archive threshold

Committed run time weighting factor.

Enable the use of historical run time in the calculation of fairshare scheduling priority, Disable the use of historical run time in the calculation of fairshare scheduling priority.

Open lease reclaim time.

Enable chunk job dispatch for jobs with CPU limit or run limits.

The interval for scheduling jobs by scheduler daemon.

The number of jobs per user per queue whose pending reason is published at the PEND_REASON_UPDATE_INTERVAL interval.

The interval for publishing job pending reason by scheduler daemon.

Interval(in seconds) of pending reason publish for all jobs.

MC pending reason update interval (0 means no updates).

MC pending reason update package size (0 means no limit).

No preemption if the run time is greater than the value defined in here.

No preemption if the finish time is less than the value defined in here.

mbatchd Archive Time

Absolute run limit for job.

The job exit rate duration.

The duration to trigger eadmin.

Maximum time for job information query commands (for example,with bjobs) to wait.

Job submission retrial interval for client.

System wide max pending jobs.

Max number of concurrent query.

Min event switch time period.

Condense pending reasons enabled.

Schedule Parallel jobs based on slots instead of CPUs.

Disable user job movement operations, like btop/bbot.

Detect and report idle jobs only after specified minutes.

Use symbolic when specifing priority of symphony jobs.

Priority rounding for symphony jobs.

The mapping of the symbolic priority for symphony jobs.

Maximum number of subdirectories under LSB_SHAREDIR/cluster/logdir/info.

The minimum period of a child MBD to serve queries in the MT way.

Stop asking license to LS not due to lack license.

Expire time for finished job which will not taken into account when calculating queue fairshare priority.

MBD child query processes will only run on the following CPUs.

The default application profile assigned to jobs.

Enable or disable data streaming.

File to which lsbatch data is streamed.

File size in MB to which lsbatch data is streamed.

Sync up host status with master LIM is enabled.

Project schedulign default SLA.

EGO Enabled SLA scheduling timer period.

EGO Enabled SLA scheduling time to live.

EGO Enabled SLA scheduling connection timeout.

EGO Enabled SLA scheduling read timeout.

EGO Enabled SLA scheduling use MXJ flag.

EGO Enabled SLA scheduling reclaim by queue.

EGO Enabled SLA scheduling default velocity.

Type of host exit rate exception handling types: EXIT_RATE_TYPE.

Type of host exit rate exception handling types: GLOBAL_EXIT_RATE.

Type of host exit rate exception handling types ENABLE_EXIT_RATE_PER_SLOT.

Performance metrics monitor is enabled flag.

Performance metrics monitor sample period flag.

Used to bound: (1) factors, (2) weights, and (3) APS values.

Child mbatchd gets updated information about new jobs from the parent mbatchd.

Job type to preempt, PREEMPT_JOBTYPE_BACKFILL, PREEMPT_JOBTYPE_EXCLUSIVE.

The default job group assigned to jobs.

Max ratio between run limit and runtime estimation.

Enable the post-execution processing of the job to be included as part of the job flag.

Timeout of post-execution processing.

The interval, in seconds, for updating the session scheduler status summary.

The number of completed tasks for updating the session scheduler status summary.

The maximum number of times a task can be requeued via requeue exit values.

The maximum number of times a task can be retried after a dispatch error.

The maximum number of tasks that can be submitted in one session.

The maximum run time of a single task.

The output directory for task accounting files.

If TRUE enable the job group automatic deletion functionality (default is FALSE).

Maximum number of job preempted times.

Maximum number of job re-queue times.

No preempt run time percent.

No preempt finish time percent.

The reservation request being within JL/U.

Job accept limit percentage.

The low priority job will use the slots freed by preempted jobs.

Maximum number of the backup stream.utc files.

If enforced only admin can use bkill -r option.

It controls the remote queue selection flow.

It controls update interval of the counters and other original data in MC implementation.

Jobs run on only on hosts belonging to the intersection of the queue the job was submitted to, advance reservation hosts, and any hosts specified by bsub -m at the time of submission.

Enforces the limitations of a single specified user group.

Enable or disable logging runtime estimation exceeded event.

Compute unit types.

Fairshare adjustment weighting factor.

abs runtime and cputime for LSF simulator

switch for job exception enhancement



     
Date Modified: 9 Sep 2009
Platform Support: support@platform.com
Platform Information Development: doc@platform.com

Platform Computing. Accelerating Intelligence(TM).
Copyright © 1994-2009 Platform Computing Corporation. All rights reserved.