Open llyons opened 2 weeks ago
I went ahead and disabled the PDB in this case and will test again.
It seems like when it determined that it could not do kuber04 worker node because of the pdb, I would have thought it would eventually time out, uncordon the kuber04 worker and then move on to the kuber03 control plane.
Am I able to configure this kind of behavior? it would be preferable then leaving the worker node with scheduling disabled.
We have pulled down v1.15.1 of kured and installed on k3s v1.29.2. the cluster is a based on Alma Linux 9 machines.
We have it installed on control planes and workers. This is an unusual cluster in that there are 3 control planes in HA mode and 1 worker. machine 1,2,3 are control planes and 4 is worker
The configuration of the kured command is
i did the recommended test with sudo touch /var/run/reboot-required on a control plane node and a worker.
The pods are all running
The logs from the 2 machines that we did the reboot-required show this.
abal-kuber03
abal-kuber04
it looks like it did disable scheduling on the lone worker node kuber04 and that was the state I found it in this morning.
What are we missing?