How could I set up a ES cluster?

4
votes

I have a master node which ip is 192.168.1.101 and a non-master node which ip is 192.168.1.106. The two use the same version of ElasticSearch-1.2.0.

But after I started the master node and the non-master node, then I got the following info:

[2014-06-04 02:38:49,350][INFO ][discovery.zen ] [node2] failed to send join request to master [[node1][TxZ5wuhnT1awPC1gEjYPdw][flyers-MacBook-Air.local][inet[/192.168.1.101:9300]]{master=true}], reason [org.elasticsearch.ElasticsearchTimeoutException: Timeout waiting for task.]

Config of the master node:

cluster.name: mycluster
node.name: "node1"
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: ["192.168.1.101"]

Config of the non-master node:

cluster.name: mycluster
node.name: "node2"
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: ["192.168.1.101"]

I don't know why this exception happens. Please give me some tips. Thanks in advance.

3
elasticsearch

3 Answers

3
votes

After I set network.bind_host、network.publish_host、network.host to the IP that the node held，it worked. Very strange.

1
votes

I had the same issue until I found out that my ES node did not bind to eth0 as expected but to eth2 instead. Of course this could not work because the registration response from the master node could not be sent to the IP address of my other network. I was able to fix this behaviour by setting the following parameter in my elasticsearch.yml (on the server which was not able to join the cluster)

network.publish_host: "_eth0:ipv4_"
0
votes

I'd better change ["192.168.1.101"] to ["192.168.1.101", "192.168.1.106"] in both configurations.