Runai DaemonSet Rollout Stuck

Meaning¶

Runai daemonset has 0 available pods on a relevant node.

Impact¶

No fractional gpu workloads support.

Severity¶

Critical

Diagnosis¶

run

kubectl get daemonset -n runai-backend

Identify the one or more daemonsets that have no running pods on some of the nodes.

Mitigation¶

Run kubectl describe daemonset X -n runai on the relevant deamonset(s) to try and figure out why it cannot create pods.

If you cannot correct the issue, contact Run:ai support.