We have 3 node cluster each node having configurations - 4 CPU (Intel(R) Xeon(R) CPU @ 2.30GHz) and 15 GB RAM. We have indexed 150M documents on this cluster. We have seen that without firing any query the CPU usage on all 3 x 4 = 12 CPU is 100%. Why would it be so? Is there a way to monitor whats the load on this nodes leading to it?
Also the RAM is caching a lot of data, could that slow down the process. See RAM usage - ]
root@elastic-server-instance-3:~# free -m
total used free shared buffers cached
Mem: 15040 14762 277 0 152 8725
-/+ buffers/cache: 5884 9155
Swap: 0 0 0