Let’s start with some technical details - I’m running Couchbase Server (1 instance without replication) on 1 dedicated machine with 12 CPU, 125GB RAM and enterprise-class SSD disk. Today I had some issues with Couchbase Server disk and SWAP usage. It took exactly 4 hours to fill almost empty 400GB disk. As a consequence Couchbase Server stopped working (all buckets turned into yellow in just seconds before the downtime). I had no time to take some screenshots but what I remember:
- RAM usage decreased when Couchbase started to spam the disk (persist queue?),
- SWAP increased to maximum (it’s obvious)
I’m looking for a reason or at least “something” to help me avoid situations like this in the future. I had no such problems in the past (4.0 CE, 4.1 CE, 4.1.1 CE) - maybe it’s related to 4.5 CE which I’m already using?