User limits on Anansi

We have added new limits to the number of CPU cores and GPU resources a single user can request across all jobs on the interactive/debug cluster Anansi. The aim is to improve queueing times and to avoid situations where a single user submits many jobs and uses up the resources of Anansi, effectively blocking other users from starting their session on OnDemand or batch jobs.

You can now request up to 16 CPU cores and up to 100% of a GPU (4 GPU shards) across all your jobs on Anansi. If your running jobs already attain this limit any new jobs will stay in the queue with any of the following Slurm Job Reason Codes: QOSMaxCpuPerUserLimit or QOSMaxGRESPerUser. These jobs will start once your running jobs finish and free up the necessary resources.

If a single job requests more than 16 CPU cores or more than 100% of a GPU (4 GPU shards) it will be refused by the queue with the message:

Job violates accounting/QOS policy (job submit limit, user’s size and/or time limits)

No changes to the way job submission works have been made, so you can submit jobs as you always have. Just keep in mind the new limits for Anansi.

Contact us at VUB-HPC Support if you have any comments or questions about these changes.