Single client node failure renders cluster unresponsive

I'm running ES 2.4 cluster with 8 data nodes and 3 client nodes. I'm running a python script which uses elasticsearch standard drivers to connect to this cluster. I give all the three client nodes during initialization of the client. The problem I'm facing is that sometimes client0 becomes overloaded with excessive CPU and Memory usage (not sure why). When this happens the whole system accessing this cluster (through clients ofcourse) becomes unresponsive. In other words, the python client isn't doing round-robin or load balancing between three client nodes provided. I don't know why is this happening. Moreover, I need a way to ascertain if this (load balancing) actually does happen. How can I test this behavior with a simple python client for my 3 cluster?

How many master-eligible nodes do you have? How are you initialising the client?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.