Answer
There's no official recommendation from IFS to periodically restart the Kubernetes cluster server as routine maintenance — if a crash like this happens, journalctl (including the kubelite and containerd snap service units) and dmesg are the places to look for a root cause on the Linux side, verbose as they are, and checking pod status across all namespaces afterward confirms everything came back cleanly. Restarting periodically as part of general OS patching is common practice for servers generally, but it isn't something IFS documents as a requirement specifically for the middle tier.