I recently scaled out a zookeeper cluster, and as part of the configuration change, we set
leaderServes=no
in order to increase throughput. This has worked out great, except for one problem: the leader no longer responds to (most) four letter words, replying to any inquiries with the message
This ZooKeeper instance is not currently serving requests
I get it dear leader, you're not doing the menial work anymore, but ... could the monitoring software still get your status and metrics?
I really hope there's a work around that isn't "use JMX". Any help much appreciated!