When draining a node after having cordoned it as stated in the documentation we get the following error:
error when evicting pods/"instance-manager-xxx" -n "longhorn-system" (will retry after 5s): Cannot evict pod as it would violate the pod's disruption budget.
Having the allowed disruptions of the instance manager pods set to 0 seems to cause the issue:
How is it possible to drain a node containing Longhorn on it? Is there a value related to the pod disruption budget we can update in the Helm chart? We need this for in-place Kubernetes upgrades.
This anwser is close to what we need, but involves manual steps that are not possible in an Ansible automated way of upgrading Kubernetes: https://github.com/longhorn/longhorn/discussions/4102#discussioncomment-3118777
We use Longhorn version 1.5.3 installed using helm chart.