Reboot Required #

Meaning #

Node is expressing a need to be rebooted. Usually this is shown because of an unattended upgrade which included a new kernel version.

Impact #

Running with older kernel.

Diagnosis #

Log into a machine and check if latest kernel is used.

Mitigation #

Reboot the machine by following those steps:

1. Drain node with kubectl drain node NODE_NAME
2. Ensure all pods are moved with kubectl describe node NODE_NAME
3. Add silence to alertmanager silencing every alert with node=NODE_NAME for 1h.
4. ssh into node and run reboot
5. wait for node to come up
6. Make node schedulable kubectl uncordon NODE_NAME
7. (optional) remove silence from alertmanager.