I have deployed a 4 node Couchbase cluster using Docker images on the Google Container Engine with Kubernetes. I was able to access the Couchbase Console, look at the buckets, query etc. Now, after a couple of days, I go the Console URL and the Couchbase initial setup screen comes up! As though this is a fresh install. I can see that the nodes and pods are all still up and running. Had a similar problem on my Windows box with Docker cluster (No Kubernetes). I redeployed that again. Anyone else experienced this?
When you destroy and recreate container instances all the underlying state is lost.
If you want to preserve the state of your couchbase installation you'll need to use a docker data volume. Just create one and mount your couchbase data file directory.
On gcp, you'll additionally want to map a directory on the data volume to a persistent disk.