0

I am running elasticsearch 1.4.3 as a worker-node/client (master=false, data=false) on AWS and I'm trying to get it to join a cluster.

My elasticsearch.yml looks like this:

cluster-name: elastisearch
node.master:false
node.data: false
plugin.mandatory: cloud-aws
discovery.type: ec2
cloud.aws.region: us-east-1
discovery.ec2.groups: sg-934a92fa

When I start up elasticsearch sudo -E /usr/share/elasticsearch/bin/elasticsearch -d I get this error:

[2015-08-17 20:49:02,690][INFO ][discovery.ec2        ] [Marsha Rosenberg]
failed to send join request to master [[Mayday Parker]
[7rfILKuXSLegJIT2bU9bsw][inet[/10.110.1.143:9300]]{data=false,master=true}],
reason [RemoteTransportException: [Mayday Parker][inet[/172.17.0.4:9300]]
[discovery/zen/join]; nested: NotSerializableTransportException
[[org.elasticsearch.transport.ConnectTransportException]
[Marsha Rosenberg][inet[/10.110.1.184:9300]] connect_timeout[30s];
 connection timed out; inet[/10.110.1.184:9300; ]; ]

I did a check on 10.110.1.143:9300 using netcat and it's ok. I did a tcpdump and I see traffic on 9300 between 10.110.1.184 and 10.110.1.143. But I have no idea where the ip: 172.17.0.4 came from. I don't even have a VPC created with that CIDR, let alone instances with that IP.

I think the problem is related to this weird IP, otherwise everything looks normal. I've run this ten times or more, with different configurations to no avail. I've even restarted my master nodes to no aval.

Any ideas?

1 Answers1

1

I experienced this same issue because the default bind address of 0.0.0.0 was falling back to eth1 subnet instead of eth0. After changing my bind address to that of eth1 (on both of my two ES nodes), ES was able to start up in the correct subnet. If you have two network interfaces, I would suggest to check this (ifconfig + reviewing the ES log should suffice).

jambii
  • 11
  • 1