Closed toabi closed 4 months ago
I have only noticed one incompatibility when changing libraries to 1.29. But it's only the go-client (it must be VolumeResourceRequirements
now). Would be happy if you can test out operator on 1.29. We are just moving to 1.28 here.
Hi @FxKu , I ran an update from EKS 1.28 to 1.29 on our test-box with Operator version 1.10.1 and observed following problem. Leader-Pods are not evicted automatically from a cordoned and draining node. I need to manually switchover and then the pod gets evicted. I did not observe this behavior during the upgrade from 1.27 to 1.28.
@minimax75 sounds unrelated to 1.29. Maybe you're using a newer Spilo/Patroni version? I would update to 1.11 to be sure you include this fix: https://github.com/zalando/postgres-operator/pull/2515
installed today in my kubeadm provisioned Kubernetes Cluster with version 1.29.0 and everything was working as expected for my use case.
updated to 1.11 and the issue is gone, so looks like it is related to spilo-15 3.2-p1 and operator 1.10.1
I have updated the README. You can run the operator on K8s 1.29 and higher without any problems.
The table in the readme currently only lists v1.28 as max version.
Is there really anything blocking the use of the operator in a kubernetes 1.29 cluster?