We see the behavior of elected master using high CPU and JVM always, and goes off when it reaches a higher number ~75% removing and restarting node. I'm suspicious whether it is due to transport client sending all the requests to the same master node though the sniffing is enabled.
network.host is 0.0.0.0 in all nodes. It is the IPs of 3 master nodes that are given for initial connectivity.
The JVM/CPU on all other nodes is well below 20%/70%. Elected master keeps reaching high CPU/JVM and goes off
Yes. We would be migrating to Rest Client, and is part of our road-map. But, it is not something that would happen soon and our current clusters are in 6.x with Transport Client. We are getting this issue now. Trying to see if we are missing anything as we see the JVM and CPU keeps going up on elected master=> LOoks like all connections from Transport Client is sent only to the same.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.