Over the past couple of weeks we have had two incidents in which Couchbase takes all of the disk on a server and then begins to display a negative amount for the “Free RAM” and “Free Disk”. For example Couchbase will consume 30 GB of disk in hours and not give it back until we restart the couchbase service. Any ideas? We are on 3.0.1. Normally there is 30 GB free at all times. we have 7 buckets that are 2GB each. This is a two node cluster.