[Issue]: Rolling Upgrades don't account for new unassigned replicas and stalls #1409
Closed as not planned
Description
Type of issue
None
What documentation page is affected
https://www.elastic.co/docs/release-notes/cloud-on-k8s
What happened?
When performing a rolling upgrade (ex: 8.17.1 to 8.18.1) with ECK operator 2.16.1, I have observed a number of "stalled" upgrades related to newly created shards not being able to allocate replicas because not enough nodes have been upgraded.
Per the response to elastic/elasticsearch#127845, this is intended from the Elasticsearch side and should be safe to proceed. However, it appears that the ECK operator doesn't apply this same logic and will "stall" the upgrade until a node is manually restarted to force an upgrade.
Additional info
No response