I 've used eck to set up an elasticsearch cluster at v7.2.0. I've successfully upgraded it to 7.3.0 and 7.3.1. However, when I attempted to upgrade it to 7.4.0 the change to the CRD was accepted but the operator did not upgrade the pods. Now the operator won't let me downgrade the CRD to match the pods, and it won't upgrade the pods because it thinks they're already at v7.4.0. Deleting a pod will only recreate it with the same version it's already at. I'd like to either rectify the operator's data to match the cluster state, or manually update the pods to match the CRD, and am looking for recommendations.
Operator not upgrading the pods is certainly not expected. Could you share the version of the operator that you are using, the spec you've applied and ECK logs? If you already didn't, taking a look at https://www.elastic.co/guide/en/cloud-on-k8s/current/k8s-troubleshooting.html could be helpful.