Details
-
Page
-
Resolution: Fixed
-
Minor
-
None
-
None
-
41: Odds and Sods, 42: GA items and Connect
-
1
Description
If your Kubernetes cluster has X dedicated worker nodes and your Couchbase cluster specifies a size of X with anti-affinity enabled then each pod will be deployed to a separate node and no nodes will be free.
If a pod upgrade is triggered (e.g. by a Couchbase upgrade) then there are no free nodes now to run the new pod to swap in. You need X+1 nodes at least temporarily.