Recommended discovery mechanism for production cluster on ec2


What is the recommended discovery mechanism for production clusters on ec2?

Zen: My main concerns is the possibility of dedicated master nodes changing
their ip addresses (could happen on ec2). Does zen uses the configured
unicast hosts just as gossip seed, or is it a static list? If it is just a
gossip seed, does zen persist the gossip back to disk in case a process
restart is needed and the original ip addresses are out of date? If it does
not persist gossiped ip address to disk, do I need a script that injects
fresh seeds each time elasticsearch service starts?

ec2: My main concern is the resiliency of the cluster if ec2 api returns
inconsistent results causing some kind of split brain scenario. Has such a
thing been reported?

zookeeper: This plugin has been reported to work better than zen discovery
regarding to split brain. Is this observation still relevant in v1.4? I am
willing to risk an unofficial plugin if it makes the cluster more stable.


You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
To view this discussion on the web visit
For more options, visit