Bug ID 1574821: BIG-IP Next Central Manager VM instances shutdown a in High availability cluster do not restart certain Pods

Last Modified: Feb 11, 2025

Affected Product(s):
BIG_IP_NEXT(CM) All(all modules)

Known Affected Versions:
20.2.0, 20.2.1

Opened: Apr 01, 2024

Severity: 3-Major

Symptoms

In a BIG-IP Next Central Manager HA deployment, if one of the nodes is shutdown, then certain pods such as statefulsets, mbiq-llm and mbiq-upgrade-manager-feature do not automatically reschedule on other active nodes and remain in Pending state. Also, it may take up to 5 minutes for pods to get rescheduled to other active nodes and during this time BIG-IP Next Central manager services may not be available.

Impact

Central Manager services are unavailable.

Conditions

As part of a multi-node BIG-IP Next Central Manager deployment (HA deployment), if one of the nodes might become unavailable due to explicit node deletion or power failure. When this occurs, one or more of the Pods scheduled on those nodes goes into a Pending state.

Workaround

The workaround is to manually delete the node from the cluster using command 'kubectl delete node <node-name>' and delete the PVCs of Pending pods. This will ensure the Pods get rescheduled fresh on other available nodes.

Fix Information

None

Behavior Change

Guides & references

K10134038: F5 Bug Tracker Filter Names and Tips