SCore-D Console Command

abort all|job-id

Abort specified job(s). Aborted jobs can be restarted by restart console command. Or, when SCore-D is restarted with the -restart option, the aborted job(s) will be restarted from the beginning or the most recent checkpoint.

checkpoint all|job-id

Checkpoint specified job(s).

devadd devicename hostno path [args ...]

Add SCore-D device having the specified name on the specified host number. Device server process is created as a result of exec(2)uting the specified progarm (path) with its argument(s) (args), if any.

devrem devicename

Remove specified SCore-D device.

exit

Terminate the current console session.

help [command]

Printout help message (of the specified command, if any).

info all|cluster|host|job|scheduling|scored|device

Output current SCore-D information depending on the command argument specified.

all

Same as the following command sequence.

uname
info cluster
info host
info device
info scheduling
limit cluster
limit queue
limit job
ps

cluster

Output how cluster(s) are configured.

host

List load average (1 minute), memory usage, disk usage, and swap usage of all hosts. Memory, disk and swap usage are in free/total format.

job

Same as the ps command.

scheduling

Output queue status.

scored

Same as the uname command.

device

List currently added SCore-D devices.

exit

Terminate current console session.

gclock [on|off]

Control global clock calibration.

kill all|job-id

Kill specified job(s).

limit

Same as the following command sequence.

limit cluster
limit queue
limit job

limit cluster

Output all limit values associated with clusters.

limit cluster all|cluster-id memory|disk|jobs value

Limit resource usage associated with the specified cluster. If the resource type is memory or disk, the number is in megabytes. If the resource type is jobs, then the number of runnable jobs in a host (not node) is limited.

limit job

Output all limit values associated with jobs.

limit job all|job-id cpu|memory|disk value

Limit resource usage associated with the specified job. If the resource type is memory or disk, the number is in megabytes.

limit queue

Output all limit values associated with scheduling queues.

limit queue all|queue-id time|remain|memory|disk|group value

Limit resource usage associated with the specified scheduling queue. If the resource type is time or remain then the number can be postfixed by a D, H or M character representing the time unit, day, hour or minute, respectively. If the resource type is memory or disk, the number is in percent. If the resource type is group then the value must be valid Unix user group name.

limit queue all|queue-id min|max nhosts

Limit the number of maximum or minimum hosts of the specified queue.

limit user [all|UID|user-name njobs|nhosts|nprocs value]

Limit resource usage, such as number of jobs, number of hosts, or number of processes, of the specified (or all) user(s). If this limit is set, total number of jobs, hosts, or processes scheduled by SCore-D is restricted. If user and limit value are omitted, then the current limit values will be displayed.

message all|job-id any

Send a message specified in the any arguments to the specified job's FEP (scrun) process. The sent message will be output to standard error of the scrun process.

nice all|job-id priority

Dequeue specified job(s) and enqueue the job(s) to the queue specified by the priority.

ps

List currently registered jobs.

restart all|job-id

Restart aborted job(s).

resume all|job-id

Resume job(s) which is (are) previously suspended by the suspend console command.

set timeslice time-slice

Set the time slice duration (in milliseconds).

set queue all|queue-id activated|deactivated|enabled|disabled

Change the status of the specified queue(s).

set queue all|queue-id gang|exclusive

Set the scheduling mode of the specified queue(s).

shutdown [sec]

Shutdown SCore-D. If sec is specified, then SCore-D will shutdown in the specified seconds. All running jobs will be killed.

source filename

Read SCore-D console command from the specified file. The specified file must be readable from the SCore-D server host. The specified file must not contain source command.

suspend all|job-id

Suspend sepcified job(s).

uname

Print certain system information.

unlimit cluster all|cluster-id [memory|disk|jobs]

Unlimit resource usage associated with the specified cluster. If resource type is not present, then all resource limit is unset.

unlimit job all|job-id [cpu|memory|disk]

Unlimit resource usage associated with the specified job. If resource type is not present, then all resource limit is unset.

unlimit queue all|queue-id [time|remain|memory|disk|group]

Unlimit resource usage associated with the specified scheduling queue. If resource type is not present, then all resource limit is unset.

unlimit user [all|UID|user-name njobs|nhosts|nprocs|all]

Unlimit resource usage of the specified (or all) user(s).

uptime

Tell how long SCore-D has been running.