LSF 명령어

bacct

• Displays accounting statistics about finished jobs.

badmin

• The badmin command is the administrative tool for LSF.

bapp

• Displays info about application profile configuration.

battach

• Runs a shell process to connect to an existing job execution host or container.

battr

• Provides a set of subcommands to manage LSF host attributes for attribute affinity scheduling.

bbot

• Moves a pending job to the bottom of the queue relative to the last job in the queue.

bchkpnt

• Checkpoints one or more checkpointable jobs

bclusters

• Displays information about IBM Spectrum LSF multicluster capability

bconf

• Submits live reconfiguration requests, updating configuration settings in active memory without restarting daemons.

bctrld

• Administrative tool to control LSF daemons.

bdata

• Provides a set of subcommands to query and manage IBM Spectrum LSF Data Manager. If no subcommands are supplied, bdata displays the command usage.

bentags

• Queries or removes information about the energy policy tag from the mbatchd daemon, which is saved in the energy-aware scheduling database. Used with energy policy, or energy aware scheduling feature.

bgadd

• Creates job groups

bgdel

• Deletes job groups

bgmod

• Modifies job groups

bgpinfo

• Displays information about global fair share and global resources.

bhist

• Displays historical information about jobs

bhosts

• Displays hosts and their static and dynamic resources.

 

• output: LSF job을 실행할 수 있는 호스트들의 정보
  • HOST_NAME: 호스트 이름
  • STATUS: host 및 sbatchd의 상태. batch job은 ok 상태의 호스트들에게만 전송될 수 있음
    • ok: batch job 받기 가능
    • unavail: LIM은 실행중이나 sbatchd는 unreachable
    • closed: 원격 batch job 받기 불가능.
    • closed_Cu_excl: this host is a member of a compute unit that is running an exclusive compute unit job
  • JL/U: Max number of job slots the host can process on a per user basis (단일 사용자가 허용하는 최대 작업 슬롯 수)
    • - (dash): no limit
  • MAX: Max number of job slots available
    • - (dash): no limit
  • NJOBS: number of tasks for all jobs that are dispatched to the host (including running, suspended, chunk jobs). with -alloc 옵션, total is the sum of RUN, SSUSP, USUSP, RSV counters (시스템에 있는 작업 수)
  • RUN: number of tasks for all running jobs on the host (실행 중인 작업 수)
  • SSUSP: number of tasks for all system suspended jobs on the host (시스템에 의해 일시중단된 작업 수)
  • USUSP: number of tasks for all user suspended jobs on the host (사용자에 의해 일시중단된 작업 수)
  • RSV: number of tasks for all pending jobs w/ reserved slots on the host (예약된 작업 슬롯 수)

bhpart

• Displays information about host partitions

bimages

• Displays information on Docker container images

bjdepinfo

• Displays job dependencies.

bjgroup

• Displays info about job groups

bjobs

• Displays and filters info about LSF jobs.
• Specify one or more job IDs (and, optionally, an array index list) to display info about specific jobs (and job arrays).

bkill

• Sends signals to kill, suspend, or resume unfinished jobs

bladmin

• Administrative tool for IBM Spectrum LSF License Scheduler.

blaunch

• Launches parallel tasks on a set of hosts.

blcollect

• License information collection daemon for LSF License Scheduler.
• The blcollect daemon collects license usage information.

blcstat

• Displays dynamic update information from the blcollect daemon for LSF License Scheduler.

blhosts

• Displays the names of all the hosts that are running the LSF License Scheduler daemon (bld).

blimits

• Displays information about resource allocation limits of running jobs.

blinfo

• Displays static LSF License Scheduler configuration information

blkill

• Terminates an interactive (taskman) LSF License Scheduler task.

blparams

• Displays information about configurable LSF License Scheduler parameters that are defined in the files lsf.licensescheduler and lsf.conf

blstat

• Displays dynamic license information.

bltasks

• Displays LSF License Scheduler interactive task information.

blusers

• Displays license usage information for LSF License Scheduler.

bmgroup

• Displays info about host groups and compute units.

bmig

• Migrates checkpointable or rerunnable jobs.

bmod

• Modifies job submission options of a job.

bparams
Displays information about configurable system parameters in the lsb.params file.

bpeek
Displays the stdout and stderr output of an unfinished job.

bpost
Sends external status messages and attaches data files to a job.

bqueues

• Displays information about queues.

 

Default Output	설명
QUEUE_NAME	큐 이름 - lost_and_found: LSF 관리자가 시스템에서 큐 제거 시 LSF는 lost_and_found라는 큐를 생성하고 여기다 제거된 큐 안에 있던 job들을 넣음. 여기 안에 있는 job들은 bswitch 명령어를 통해 다른 큐로 들어가지 않는 이상 시작되지 않음
PRIO	The priority of the queue. The larger the value, the higher the priority.
STATUS	상태 - Open: queue can accept jobs - Closed: queue cannot accept jobs - Active: 이 큐 안에 있는 job들은 시작될 수 있음 - Inactive: 이 큐 안에 있는 job들은 시작될 수 없음 badmin 명령어로 큐의 상태 변환 가능
MAX	The maximum number of job slots that can be used by the jobs from the queue. A dash (-) indicates no limit.
JL/U	The maximum number of job slots each user can use for jobs in the queue. A dash (-) indicates no limit.
JL/P	The maximum number of job slots a processor can process from the queue. A dash (-) indicates no limit.
JL/H	The maximum number of job slots a host can allocate from this queue. A dash (-) indicates no limit.
NJOBS	The total number of slots for jobs in the queue.  This number includes slots for pending, running, and suspended jobs.  If the -alloc option is used, the total is the sum of the RUN, SSUSP, USUSP, and RSV counters.
PEND	The total number of tasks for all pending jobs in the queue.  If used with the -alloc option, total is zero.
RUN	The total number of tasks for all running jobs in the queue.  If the -alloc option is used, the total is allocated slots for the jobs in the queue.
SUSP	The total number of tasks for all suspended jobs in the queue.
PJOBS	The total number of pending jobs (including both PEND and PSUSP job) in this queue.

 

 

bread
Reads messages and attached data files from a job.

brequeue
Kills and re-queues a job.

bresize
Decreases or increases tasks that are allocated to a running resizable job, or cancels pending job resize allocation requests.

bresources
Displays information about resource reservation, resource limits, and guaranteed resource policies.

brestart
Restarts checkpointed jobs.

bresume
Resumes one or more suspended jobs.

brlainfo
Displays host topology information.

brsvadd
Adds an advance reservation.

brsvdel
Deletes an advance reservation.

brsvjob
Shows information about jobs submitted with the brsvsub command to a specific advance reservation.

brsvmod
Modifies an advance reservation.

brsvs
Displays advance reservations.

brsvsub
Creates a dynamically scheduled reservation and submits a job to fill the advance reservation when the resources required by the job are available.

brun

• Forces a job to run immediately.

bsla

• Displays info about service classes.
• Service classes are used in guaranteed resource policies and service-level agreement (SLA) scheduling.

bslots

• Displays slots available and backfill windows available for backfill jobs.

bstage
Stages data files for jobs with data requirements by copying files or creating symbolic links for them between the local staging cache and the job execution environment. You must run bstage only within the context of an LSF job (like blaunch). To access a file with the bstage command, you must have permission to read it.

bstatus

• Gets current external job status or sets new job status.

bstop

• Suspends unfinished jobs.

bsub

• Submits a job to LSF by running the specified command and its arguments.

bsubmit

• Submits jobs as another user.

bswitch
Switches unfinished jobs from one queue to another.

btop
Moves a pending job relative to the first job in the queue.

bugroup
Displays information about user groups.

busers

• Displays information about users and user groups.

bwait

• Pauses and waits for the job query condition to be satisfied.

ch
Changes the host where subsequent commands run.

gpolicyd
Displays LSF global policy daemon information.

lim

• Load information manager (LIM) daemon or service, monitoring host load.

lsacct
Displays accounting statistics on finished RES tasks in the LSF system.

lsacctmrg

• Merges LSF RES task log files.

lsadmin

• Administrative tool to control LIM and RES daemon operations in LSF.

lsclusters

• Displays configuration info about LSF clusters.

lsfinstall

• The LSF installation and configuration script.

lsfmon
Install or uninstall LSF Monitor in an existing cluster.

lsfrestart

• Restarts the LIM, RES, sbatchd, and mbatchd daemons on all hosts in the cluster

lsfshutdown

• Shuts down the LIM, RES, sbatchd, and mbatchd daemons on all hosts in the cluster.

lsfstartup

• Starts the LIM, RES, and sbatchd daemons on all hosts in the cluster.

lsgrun

• Runs a task on a group of hosts.

lshosts

• Displays hosts and their static resource info.

 

output	설명
HOST_NAME	호스트 이름
type	LSF multicluster capability에서는 remote cluster 안의 호스트의 host type이 local cluster에 정의되어있지 않으면 unknown이라고 출력
model	LSF multicluster capability에서는 remote cluster 안의 호스트의 host model이 local cluster에 정의되어있지 않으면 unknown이라고 출력
cpuf	relative CPU performance factor. the faster the CPU, the larger the CPU factor (unknown host type을 갖고 있는 호스트의 기본값 = 1.0)
ncpus	number of processors
maxmem	maximum amount of physical memory available for user processes
maxswp	total available swap space
server	erver/client 호스트인지 여부. LSF servers = Yes, LSF clients = No, Dynamic hosts = Dyn
RESOURCES	resources that are defined for the host

 

lsid

• Displays the LSF version number, the cluster name, and the management host name.

lsinfo

• Displays LSF configuration info.

lsload

• Displays load info for hosts (옵션에 따라 host-basis, resource-basis로 출력 가능)

 

• 기본값:
  • displays load information for all hosts in the local cluster. 
  • displays raw load indices
  • load info for resources is displayed according to CPU and paging load
• With the IBM® Spectrum LSF multicluster capability enabled, the lsload command also displays load info for all hosts in equivalent clusters.

HOST 기반 출력 field	설명
HOST_NAME	Standard host name that is used by LSF, typically an internet domain name with two components
status	호스트의 상태. 상태 앞의 minus sign (-) 있으면 RES가 호스트에서 실행중이지 않다는 의미 상태 종류: - ok: 상태 정상, remote job 수신 가능, LIM is unlocked, LIM과 RES 둘 다 실행중 - -ok: LIM은 실행중이나 RES는 unreachable - busy: The host is overloaded because some load indices exceed configured thresholds. Load index values that caused the host to be busy are preceded by an asterisk (*). - lockW: The host is locked by its run window. Run windows for a host are specified in the lsf.conf configuration file, and can be displayed by the lshosts command. A locked host does not accept LSF jobs from other hosts. - lockU: The host is locked by the LSF administrator or root. - unavail: 호스트가 down됐거나 LIM 실행중이지 않
r15s	The 15 second exponentially averaged CPU run queue length.
r1m	The 1 minute exponentially averaged CPU run queue length.
r15m	The 15 minute exponentially averaged CPU run queue length.
ut	The CPU utilization exponentially averaged over the last minute, 0 - 1.
io (기본값으로 출력 안됨)	If the -l option is specified, shows the disk I/O rate exponentially averaged over the last minute, in KB per second.
pg	The memory paging rate exponentially averaged over the last minute, in pages per second.
ls	The number of current login users
it	On UNIX, the idle time of the host (keyboard is not touched on all logged in sessions), in minutes.  On Windows, the it index is based on the time that a screen saver is active on a particular host.
tmp	The amount of free space in /tmp, in MB.
swp	The amount of available swap space (기본값: 단위 in KB) The amount can appear in MB depending on the actual system swap space. Use the LSF_UNIT_FOR_LIMITS parameter in the lsf.conf file to specify a larger unit for the limit (GB, TB, PB, or EB).
mem	The amount of available RAM (기본값: 단위 in KB) The amount can appear in MB depending on the actual system memory. Use the LSF_UNIT_FOR_LIMITS parameter in the lsf.conf file to specify a larger unit for the limit (GB, TB, PB, or EB).
external_index (기본값으로 external load indices는 출력 안됨)	If the -l option is specified, shows indices for all dynamic custom resources available on the host, including shared, string, and Boolean resources. If the -I load_index option is specified, shows only indices for specified non-shared (host-based) dynamic numeric custom resources.

 

Resource 기반 출력 field (-s 옵션 사용)	설명
RESOURCE	자원 이름
VALUE	자원의 인스턴스의 value
LOCATION	자원의 인스턴스와 관계된 호스트

 

lsloadadj

• Adjusts load indices on hosts.

lslogin
Remotely logs in to a lightly loaded host.

lsmake
Runs LSF make tasks in parallel.

lsmon
Displays load information for LSF hosts and periodically updates the display.

lspasswd
Registers Windows user passwords in LSF. Passwords must be 3 - 23 characters long.

lsplace
Displays hosts available to run tasks.

lsportcheck
Displays ports that LSF is currently using or the LSF ports that will be used before starting LSF.

lsrcp
Remotely copies files through LSF.

slsreghost (UNIX)
UNIX version of the lsreghost command registers UNIX LSF host names and IP addresses with LSF servers so that LSF servers can internally resolve these hosts without requiring a DNS server.

lsreghost (Windows)
Windows version of the lsreghost command registers Windows LSF host names and IP addresses with LSF servers so that LSF servers can internally resolve these hosts without requiring a DNS server.

lsrtasks
Displays or updates a remote task list.

lsrun
Runs an interactive task through LSF.

pam
Parallel Application Manager (job starter for MPI applications).

patchinstall
UNIX only. Manage patches in LSF cluster.

pversions (UNIX)
UNIX version of the command. Displays the version information for IBM Spectrum LSF installed on UNIX hosts.

pversions (Windows)
Windows version of the command. Displays the version information for IBM Spectrum LSF installed on a Windows host.

ssacct
Displays accounting statistics about finished IBM Spectrum LSF Session Scheduler jobs.

ssched
Submit tasks through the IBM Spectrum LSF Session Scheduler.

taskman
Checks out a license token and manages interactive UNIX applications.

tspeek
Displays the stdout and stderr output of an unfinished Terminal Services job.

tssub
Submits a Terminal Services job to LSF.

wgpasswd
Changes a user’s password for a Microsoft Windows workgroup.

wguser
Modifies user accounts for a Microsoft Windows workgroup.

 

참조

1. https://www.ibm.com/docs/sr/spectrum-lsf/10.1.0?topic=reference-command