My Elasticsearch consists of 2masters(1 is active), 1client, 3HOT and 3WARM data nodes,m and Curator is running at 12PM for allocation / delete works.
I am observing high CPU usage on all WARM nodes.
02, 03: Master
04, 05, 06: HOT
07, 08, 09: WARM
I am attaching _node/hot_threads result on one of WARM node (07 data node) with $top result which captured at 10AM (No Curator runs yet today. It will run 2hours later); 08,09 nodes are the same way busy.
As I mentioned, Curator works are not started yet, but hot_threads results show "flush" to busy.
Why WARM nodes are still busy to "flush" even though Curator run finished about 20hours ago? I expected WARM nodes are doing nothing after Curator work(allocation) is done, then they are busy only for "search" requests come. Is my understanding wrong?
Please help me to understand why WARM nodes are busy yet, and if there are any other commands to check the node status, please advise.