partners placeholder

Moab Workload Manager Change Log

Change Log

Moab Workload Manager 4.5.0
• Moab-4.5.0p5

Moab Workload Manager 4.5.0p5

- accepts newline delimited nodes.dat files
- active policy usage fix
- added RM description field
- added job reject/job preemption event records
- added "absolute fs" support
- added account tracking to XT job mon tool
- added AM performance reporting
- added auto version
- added blockcredlist to identity manager
- added bounds on PBS event processing
- added bw/latency reporting via checkjob
- added class default gres pointer check
- added class RMAList enhancements
- added commented out support for log rolling
- added Cray X1E subtype
- added data staging helps for LSF
- added -d support to msub
- added event aggregation
- added forced job requeue on XT job reject
- added gevent 'off' handling
- added gmetric job requirements
- added Gold fixes and improvement in license errors
- added historical backlog checkpointing
- added IDCFG resetcredurl attribute
- added info service facilities
- added invocation logging
- added IPMI tools.
- added job hold routing to TORQUE
- added jobprologs on global default class
- added minor optimizations
- added native exec failure logging
- added NETUSAGE for warewulf
- added node pools
- added node reboot event
- added opsys RM extension
- added --orange
- added original loglevel handling
- added parameter config/list grid changes
- added partition backlog info
- added partition level resource sync
- added pbspro RM extension support
- added peer to peer partition message passing
- added policy estart time logging
- added preemptor diagnostics to checkjob
- added QoS level min preempt time
- added QoS rsv access list
- added reboot support
- added redundant ID manager check to hourly refresh
- added reporting of bad client messages
- added RESTARTINTERVAL
- added RMCFG[] STAGINGTHRESHOLD
- added SCHEDCFG[] RECOVERYACTION
- added showq -R <RSVID> and modify showres
- added SLURM par-to-class mapping
- added some enhancements to node deletion
- added some fixes and enhancements to job checkpointing
- added some fixes for decompression
- added some fixes for uc and rsv
- added some loadleveler fixes
- added some more build info
- added some netload stuff
- added startcount priority factor
- added summary and exclusive mshow flags to odmaster info query
- added support for 0 task jobs
- added support for dependencies
- added support for --forward
- added support for LSF job dependencies
- added support for mjobctl -m PAL=<pal>
- added support for net usage rolling
- added support for soft limit on jobmaxoverrun (sends email)
- added support for xfactor trigger threshold
- added support PBS nice specification
- added suspend/resume enhancements
- added threshold queuetime for triggers
- added update code for suspend/resume handling
- added update to detect zombie jobs and place them on hold with a message
- changed IGNNODESET to ENFORCENODESET for reservations
- changed moab checkpoint permissions to 0600
- changed mustrremove
- checknode now displays nodeaccess policy
- cleaned-up dup nodes
- cleaned-up job limit message
- cleaned-up job reject handling
- cleaned-up showstats -v backlog
- enabled authorized peers to view/modify local parameters
- enabled BLOCKCREDLIST
- enabled BProc resource manager subtype
- enabled data-staging with INSTANTSTAGE resource managers
- enabled defaults for storage RM (should set timeout to something other than 0 now...)
- enabled double count cred usage change
- enabled dynamic RM deletion
- enabled future allocation requests
- enabled internal queue job aggregation
- enabled job async start time
- enabled larger fairshare diagnostics buffer
- enabled LSF data-staging
- enabled mjobctl -m eeduration to reset job's effective queue time
- enabled mjobctl -w cred=val
- enabled NC + TPN spec via msub (no TC)
- enabled node pools with MAXJOB
- enabled QoS level preemptmintime
- enabled SR absolute offsets
- enhanced id manager to support cred reset
- enhanced internal charging
- enhanced mdiag -C
- enhanced netload tracking
- enhanced rsv delay message
- enhanced SLURM ranges
- enhanced stattestcount
- fixed native workload query to continue loading jobs after failed job detected
- fixed node state tracking for rsv creation
- fixed personal reservations
- fixed resourcelimitpolicy parsing
- fixed RM extension based single task node allocation policy
- fixed rsv checkpointing buffer size management
- fixed rsv id recycle handling on failure
- fixed rsv non-terminated node list bug
- fixed showstate for loarge jobs
- fixed sid truncation
- fixed some 64-bit issues
- fixed start accuracy stats
- fixed syscfg error message
- fixed task spec in setres
- fixed TORQUE job state issue
- fixed various memory leaks
- implemented CLASSCFG[] SYSPRIO=TRUE
- implemented prolog and epilog
- improved data staging failure recovery
- improved data-staging scripts to overcome SSH banners
- improved forwarded failure messages
- improved job aggregation method
- improved LSF job dependency support
- improved mode handling, logging, and accounting of RM suspend/resume actions
- improved SLURM integration
- improved start accuracy display
- made SSH the default cluster query method