I was facing this issue of network resiliency. I had cluster of
multiple servers and multiple node, I cut down number of ES nodes
primarily because network failure resiliency issues. I was tracking
the discussion on
https://groups.google.com/group/elasticsearch/browse_thread/thread/7c033ea3b9a92fd3/5f14c9841516962a?lnk=gst&q=sujoy#5f14c9841516962a
I tweaked the discovery.zen.minimum_master_nodes parameter to get the
some level of stability but I was then defying the whole concept of
distributed cluster and failover.
Thanks a lot Shay for this fix
It gives me immense happiness and confidence when you so promptly
address our concerns.
Thanks and Regards
Jagdeep