Known issues

Intro

• Please report issues to support@computecanada.ca

Shared issues

1. The CC slurm configuration preferentially encourages whole-node jobs. Users should if possible request whole-nodes rather than per-core resources. See Job Scheduling - Whole Node Scheduling (Patrick Mann (talk) 20:15, 17 July 2017 (UTC))
2. Quotas on /project are all 1 TB. The Storage National team is working on a project/RAC based schema. Fortunately Lustre have announced group-based quotas but that will need installation. (Patrick Mann (talk) 20:12, 17 July 2017 (UTC))
3. SLURM epilog does not fully clean up processes from ended jobs, especially if the job did not exit normally. (Greg Newby) Fri Jul 14 19:32:48 UTC 2017)
4. Email from graham and cedar is still undergoing configuration. Therefore email job notifications from Slurm are failing. (Patrick Mann (talk) 17:17, 26 June 2017 (UTC))
   • Cedar email is working now (Patrick Mann (talk) 16:11, 6 July 2017 (UTC))
   • Graham email is working
5. The SLURM 'sinfo' command yields different resource-type detail on graham and cedar. (Greg Newby) 16:05, 23 June 2017 (UTC))
6. Local scratch on compute nodes has inconsistent naming. Cedar has /local and Graham has /localscratch.
7. The status page at http://status.computecanada.ca/ is not updated automatically yet, so does not necessarily show correct, current status.
8. "Nearline" capabilities are not yet available (see https://docs.computecanada.ca/wiki/National_Data_Cyberinfrastructure for a brief description of the intended functionality)
   • Update July 17: still not working. If you need your nearline RAC2017 quota then please ask CC support. (Patrick Mann (talk) 20:45, 17 July 2017 (UTC))
9. Operations will occasionally time out with a message like "Socket timed out on send/recv operation" or "Unable to contact slurm controller (connect failure)". As a temporary workaround, attempt to resubmit your jobs/commands, they should go through in a few seconds. (Nathan Wielenga) 08:50, 18 July 2017 (MDT))

Cedar only

• Environment variables such as $SCRATCH and $PROJECT are not yet set, although the filesystem are available. (Greg Newby) 16:10, 21 June 2017 (UTC))

Graham only

• Graham scheduling is not properly running small jobs. Working on the problem. (Patrick Mann (talk) 20:14, 17 July 2017 (UTC))
  • too many nodes in the full node partition, not enough in the by core partition - no idea who is modifying the slurm conf
  • Nodes have been added to the <72 by core partition, which should alleviate some of the bbacklog. (Nathan Wielenga 14:36, 19 Jul 2017 (MDT))
• big memory nodes need to be added to the scheduler
• no network topology information in the scheduler

Resource Allocation Competition