Hard Out Of Memory Error. Bucket "default" on node "blah.blah" is full. All memory allocated to this bucket is used for metadata
We saw this the other day. The default bucket in this case was not full, and it wouldn't let us increase the bucket size, when the server(s) seemed to have plenty of available memory for it.
Once this happened, no servers could come in or out of the cluster (showed pending the whole time).
Restarting the service and the server didn't help.
We just use the default bucket, and re-installing was a quick as anything.
How have others resolved/worked around this?