Getting your elasticsearch cluster running is not so long. But now you have to properly monitor it.
It can be down even if the server is still up and even the “java elasticsearch process”.
Here is a nice monit conf grab from chef elasticsearch cookbook:
check host elasticsearch_connection with address 0.0.0.0
if failed url http://0.0.0.0:9200/ with timeout 15 seconds then alert
check host elasticsearch_cluster_health with address 0.0.0.0
if failed url http://0.0.0.0:9200/_cluster/health
and content == 'green'
with timeout 60 seconds
Will send a mail if the http api not reply and the last but not least will send a mail if the cluster is in another state than green (this is the beggining of the end sometimes ;) ).
You have to install this on each node.
In my case i run all my node with monit (ping me if you want a 30 seconds monit install on debian like).