Limitations for Quota management

Follow the recommendations on the limitations of Cloudera AI Quota management.

CPU and memory quotas are not enforced for workloads running with GPUs.
GPU utilisation is not tracked in the grafana Quota Management Dashboard.
GPU quota is enforced for the total number of GPUs.
In Cloudera AI Workbench, GPU-type quota is only supported with the Yunikorn scheduler.
If the resource pool gets deleted, the underlying Cloudera AI Workbench and workloads will also be deleted. However, the stale entries will still be available in the Cloudera AI UI, reflecting that the workbench still exists but the underlying applications on the cluster are deleted.

Any modifications to the team quota will take effect immediately.

note

For Cloudera AI on premises 1.5.5 SP1 and lower releases, if the quota for users or teams is modified, it will be reflected when the next workload (session or job) is submitted.
Enabling Quota management for existing Cloudera AI non-quota enabled Workbench is not supported.