3 physical Elasticsearch nodes in a cluster. Each node 8 core 64gb ram ubuntu svr 20.04 lts with 2TB M2 storage. 30gb Heap. 600 shards, 1 replica. I will eventually index 10TB on this cluster so 20GB per shard ~12TB. My strategy is to start on these 3 servers and expand the cluster as docs grow. Why on earth is my heap so high on 600 shards with 0 docs indexed yet? So I need to immediately expand h/w for 3 separate master nodes? What am I missing?
Which version of Elasticsearch are you using?
Thank you. The latest version 7.10.2
Hi @chrislinx ,
the heap used may contain lots of garbage too, which is essentially free space. You can try to use
jcmd <pid> GC.run, hopefully it then drops to the minimal level required for the node.
thank you very much for the heads up!