Can I cancel cluster jobs which have stalled whilst still showing as 'submitting' via the cluster job status page?
Jobs which have been successfully submitted to the cluster job manager may be canceled at any time via the cluster job status page. However, its not possible to cancel a job during the job submission phase via the Columbus user interface. If a job has stalled during submission it will not prevent subsequent jobs from running, nevertheless the job can be cleaned up via the psql command line interface of the columbus_webapp database.
1) connect the the columbus server via ssh
2) switch to the columbus user account
$ su - columbus
3) connect to the columbus_webapp database
$ psql columbus_webapp
4)Jobs can be selectively removed from the job status page using either their unique job ID, or by their status.
- to clean up a job based on the job status ID (example uses ID 100):
columbus_webapp=> delete from cluster_jobstatus where id=100;
- to clean up ALL jobs with a status listed as 'Submitting':
columbus_webapp=> delete from cluster_jobstatus where state='SU';
Please sign in to leave a comment.
Comments
0 comments