...
qbsub submits a command for the Supervisor to schedule. If the --range is specified, a job with an agenda will be submitted.
See Jobs, instances, agenda items, and job slots for a description of the taxonomy of a Qube! job.
Usage
qbsub [options] command
where command is a command to be executed on the remote host.
...
Option | Description |
---|---|
--name string | Job name |
--priority int | Priority value |
--requirements string | Job requirements expression - Job Requirements |
--reservations string | Job reservations expression - Job Reservations |
--restrictions string | Job restrictions expression - worker_restrictions |
--pid int | Parent job to submitted job |
--pgrp int | Process group ID |
--cluster string | Cluster specification |
--kind string | Job kind |
--account string | Arbitrary accounting data string |
--user string | User account to use Note: needs admin or impersonation privilege |
--domain string | Qube! domain to use Note: needs admin or impersonation privilege |
--cpus int | Number of subjobs to run |
--max_cpus int (or "*") | Maximum number of instances to use ("*" means "no limit") for Smart-share auto-expansion (new in 6.6) |
--processors int | Number of processors to reserve per host |
--status string | Initial job state (blocked or pending) |
--hosts string,… | List of comma delimited hosts to run job on |
--omithosts string,… | List of comma delimited hosts to not run job on |
--groups string,… | List of comma delimited host groups to run job on |
--omitgroups string | List of comma delimited host groups to not run job on |
--hostorder string | Order in which the hosts are preferred - hostorder syntax |
--flags string,… | Flags (see below for description of all available flags) |
Email job owner when job is complete. | |
--dependency string | Wait until trigger string condition has been met before starting. |
--preexec string | Execute command string on the Supervisor when the job starts running. |
--postexec string | Execute command string on the Supervisor when the job completes. |
--waitfor int | Wait until job ID int is done before starting. |
--timelimit int | Automatically kill the job if it runs longer than int seconds. |
--agendatimelimit int | Automatically kill an agenda item if it runs longer than int seconds. |
--waiton | Qbsub will block until the submitted job completes. |
--range string | Execute over range string QB_FRAME_NUMBER |
--padding int | Range pad value. |
--reverse | Reverse the frame range |
--binary | Use binary for frame range ordering |
--partitions int | Number of partitions to use (default: # of frames) |
--chunk int | Number of frames to use per partition (default: 1) |
--p_agenda string | Specify which frames to use as p-agenda (default: 1st, last, and middle frames) |
--p_priority int | Set priority for p-agenda (default: site-wide value, or 1) |
--p_cpus int | Number of subjobs to use to process p-agenda (default: length of p-agenda list) |
--export string | Export job to *.qja file. |
--import string | Import job from *.qja file. |
--address string, … | List of string email addresses when using --email option |
--type string --prototype string | Use string job type. |
--data string | Raw package data string (for use only with --type option). |
--retrysubjob int | Retry each subjob int times. |
--retrywork int | Retry each work agenda item int times. |
--retryworkdelay int | Insert artificial delay of int seconds before each automatic retry of work item |
--preflights string,... | instance-level preflights (comma-separated) |
--postflights string,... | instance-level postflights (comma-separated) |
--agenda_preflights string,... | agenda-level preflights (comma-separated) |
--agenda_postflights string,... | agenda-level posflights (comma-separated) |
--xml | Output in XML format |
--help | Command help |
--version | Command version |
Anchor | ||||
---|---|---|---|---|
|
Include+ | ||||||||
---|---|---|---|---|---|---|---|---|
|
Notes
Examples
Submit "ls -l" job named "myjob1" to run on 2 cpus:
...
Execute "myprog /home/data/myfile.data", but convert the path at runtime to a suitable path on the worker if necessary (as dictated by the worker_path_map defined on the worker):
See also
How to restrict a host to only one instance of a given kind of job, but still allow other jobs