CODEWITHSUNDEEP
elasticsearch
Matt R. Wilson
Matt R. Wilson

Reputation: 7575

Trying to replicate across two remote servers

I have two Elastic Search on different VPSs with their own host names and I haven't been able to get them to replicate to each other. They are both version 0.90.2.

My settings are:

cluster.name: mycluster
name.name: "nodeA"
node.master: true
node.data: true
index.number_of_shards: 5
index.number_of_replicas: 1
discovery.zen.ping.multicast.enabled: false
discovery.zen.ping.unicast.hosts: ["nodeB.example.com"]

and:

cluster.name: mycluster
name.name: "nodeB"
node.master: false
node.data: true
index.number_of_shards: 5
index.number_of_replicas: 1
discovery.zen.ping.multicast.enabled: false
discovery.zen.ping.unicast.hosts: ["nodeA.example.com"]

When I attempt to start the nodeB instance the result is:

[INFO ][discovery.zen] [nodeB] 
failed to send join request to master 
[[nodeA][IZFgtrTPSISQR7VklS3www][inet[/*.*.*.*:9300]]{master=true}],
reason [org.elasticsearch.ElasticSearchTimeoutException:
Timeout waiting for task.]

So communication is there because nodeB knows the others name is 'nodeA', but what I am I missing or what can I check?

UPDATE:

Unfortunately this turned out to be a server issue on nodeB and had nothing to do with ES.

For the record, the above settings work fine and the unicast setting for nodeA is pointless as nodeB is not a master only nodeB needs to know about nodeA.

Upvotes: 0

Views: 1164

Answers (1)

shy
shy

Reputation: 1410

UPDATED:

Make the following in both config files:

discovery.zen.ping.unicast.hosts: ["master_node_ip"]

If provided settings above is the all of it. Then you should add cluster name to your settings.

cluster.name: my_cool_cluster_name

Upvotes: 1

Related Questions