...
This script will submit a job with a 30 second callback , agenda-based timeout that runs a bunch of "sleep" commands. Each agenda item will do exactly nothing for a number of seconds equal to the frame number. Timeouts have a resolution of about 15 seconds, so you should start seeing agenda items timeout at around frame 30-50, and then timeout consistently beyond that (or until the job is auto-wrangled away, assuming auto-wrangling is enabled).
| Code Block | ||||
|---|---|---|---|---|
| ||||
import qb # a job can have several callbacks, so they are stored in a list callbacks = [] # each callback is a Qube Callback object. Callback objects (like like # all Qube objects in the Python API) are dictionaries at the end end # of the day, so we'll define a blank dict. cb = {} cb['language'] = 'qube' # requires the "qube" supervisor_language_flag be enabled cb['triggers'] = 'timeout-work-self-*' # this is the trigger for which we'll be watching - a timeout event on any of our work (agenda) items cb['code'] = 'fail-work-self' # this is the thing we will do when the trigger fires - fail the work item that triggered the event # append all individually created callbacks to the callback list callbacks.append(cb) # now create a basic job object j = {} j['name'] = 'agenda timeout test - 30 second timeout (+/- 15 seconds)' j['agendatimeout'] = 30 # number of seconds before timeout is triggered - this can be edited after submission in Qube 6.6+ j['prototype'] = 'cmdrange' j['package'] = {'cmdline':'qbping -sleep QB_FRAME_NUMBER','range':'1-100'} # 'qbping -sleep X' is a cross-platform method of sleeping j['callbacks'] = callbacks # use our previously created callbacks list j['agenda'] = qb.genframes(j['package']['range']) # don't forget to generate frames. # finally, submit the job. submitted = qb.submit(j) for s in submitted: print ('submitted job %d' % s.get("id")) |
| Note |
|---|
If "qbping" is not in your system PATH env var, then you may have to specify the full path to it. qbping lives in $QBDIR/bin. |