I've upgraded an ElasticSearch server running on Ubuntu 18 from version 7.5.1 to 7.9.3. I use ES, Kibana and Logstash on this server. The load was immediately much higher after the upgrade and this has persisted. The load is about double what it was before the upgrade.
When I looked into Management -> Stack Monitoring in Kibana I could see that the search/rate for the .kibana_4 index was ~1100/s whereas the next busiest index is only 5/s and after that 0.5/s. This is a small system, there's one node, no replicas and about 150GB of indices. These are time-based indices (one per day). There's 722 indices and 1393 shards. There's 12GB RAM on the server and 4 vCPUs. I've allocated a heap of 4GB to ES. There's only one person querying Kibana - me. Based on this seemingly high search rate I shut down Kibana and the load immediately fell to the level it was before the upgrade. Cluster health is appearing as green/A-OK whether Kibana is running or not. The size of the .kibana_4 index is only 10MB (60 documents). After the upgrade I experienced an error message in the log relating to the .kibana_4 index when starting Kibana . So I deleted the index and restarted Kibana, it had been stuck on the error for many minutes prior to doing this.
Anybody have any ideas?