You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Need to create either a script that admins can run or that runs at fixed intervals that can run varying levels of diagnostics cluster wide. This will require us to manage workload lifecycle/pod placement as well as investigate further default DCGM health checks. We also have our own checking fabric health and gpu stress tests that we can add in ourselves.
Need to create either a script that admins can run or that runs at fixed intervals that can run varying levels of diagnostics cluster wide. This will require us to manage workload lifecycle/pod placement as well as investigate further default DCGM health checks. We also have our own checking fabric health and gpu stress tests that we can add in ourselves.
https://docs.nvidia.com/datacenter/dcgm/latest/user-guide/dcgm-diagnostics.html#
The text was updated successfully, but these errors were encountered: