In case some tests block the whole system, it would be useful to have some emergency tools available to instructors:

  • kill this test
  • kill all tests running / waiting for this course
  • restart this JVM (?)
Submitted by Mathijs de Weerdt on 16 December 2014 at 10:24

On 16 December 2014 at 11:11 Elmer van Chastelet commented:

Are there still cases where executions block the whole system?
In the previous case, the JVM hung completely, and could not kill threads anymore, so the only option left was to restart the JVM.
We won’t support restarting the backend through the weblab frontend, unless this is really the only workable option left. In case the backend hangs, the (existing) watchdog will eventually restart the backend.

On 16 December 2014 at 11:31 Mathijs de Weerdt commented:

No; we haven’t found any new cases that block the system so far. For the next exercise (deadline January) we will support the use of WebLab again, but my assistants were afraid to get into the same situation as before and therefore asked for some fall-back procedure.

On 23 June 2017 at 13:41 Elmer van Chastelet tagged algoritmiek-2017

On 23 June 2017 at 13:45 Elmer van Chastelet commented:

Instead of a kill-all-jobs emergency tool, have a mode in which the job execution can be disabled for a single assignment with appropriate message in the console.

Log in to post comments