I currently am running a 4 node instance of Elasticsearch, and on one node we have an instance of Elasticsearch and Logstash running. This node has a master/data/ingest role, but that in combination with the Logstash instance is proving to be CPU intensive. We however are adding a 5 node to the cluster so to alleviate some of the cpu load on the Elasticsearch/Logstash node we changing it to be a dedicated master role. My question is what would be the correct process to change the role of a node to limit damage and downtime. I understand you can change roles in the elasticsearch.yml , but if I change the role to not be a data node anymore would this effect the index shards currently on that node? In turn would that effect the rest of the cluster? I should add that our indices are running with number of replicas set to 1.
warkolm (Mark Walkom) #2
You will need to run an exclude allocation filter to remove shards from the node, shut it down, change the role, then restart it.
system (system) #3
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.