Cluster nodes don't see each other


#1

Hi Guys,

we have installed elasticsearch (version 1.1.2 with plugin head) on a
server cluster consisting of 2 servers. The only configuration change was
defining of 'cluster.name' and 'node.name'. Both instances are running fine
but they don't see each other. Asking for the health or cluster state I see
only my own instance. I've increased the logging of discovery to 'ALL',
the relevant results are above.
Also data included in one node are not visible in the other one. Can someone say me what is wrong?

Many thanks,

Amelie

[2016-01-12 20:07:19,601][INFO ][node                     ] [cn2] version[1.1.2], pid[1990], build[e511f7b/2014-05-22T12:27:39Z]
[2016-01-12 20:07:19,602][INFO ][node                     ] [cn2] initializing ...
[2016-01-12 20:07:19,609][INFO ][plugins                  ] [cn2] loaded [], sites [head]
[2016-01-12 20:07:21,089][DEBUG][discovery.zen.ping.multicast] [cn2] 
using group [224.2.2.4], with port [54328], ttl [3], and address [null]
[2016-01-12 20:07:21,096][DEBUG][discovery.zen.ping.unicast] [cn2] using initial hosts [], with concurrent_connects [10]
[2016-01-12 20:07:21,102][DEBUG][discovery.zen            ] [cn2] using 
ping.timeout [3s], master_election.filter_client [true], 
master_election.filter_data [false]
[2016-01-12 20:07:21,103][DEBUG][discovery.zen.elect      ] [cn2] using minimum_master_nodes [-1]
[2016-01-12 20:07:21,105][DEBUG][discovery.zen.fd         ] [cn2] 
[master] uses ping_interval [1s], ping_timeout [30s], ping_retries [3]
[2016-01-12 20:07:21,126][DEBUG][discovery.zen.fd         ] [cn2] [node 
 ] uses ping_interval [1s], ping_timeout [30s], ping_retries [3]
[2016-01-12 20:07:22,164][INFO ][node                     ] [cn2] initialized
[2016-01-12 20:07:22,164][INFO ][node                     ] [cn2] starting ...
[2016-01-12 20:07:22,299][INFO ][transport                ] [cn2] 
bound_address {inet[/0.0.0.0:9300]}, publish_address {inet[/IP:9300]}
[2016-01-12 20:07:22,332][TRACE][discovery                ] [cn2] 
waiting for 30s for the initial state to be set by the discovery
[2016-01-12 20:07:22,340][TRACE][discovery.zen.ping.multicast] [cn2] [1] sending ping request
[2016-01-12 20:07:23,844][TRACE][discovery.zen.ping.multicast] [cn2] [1] sending ping request
[2016-01-12 20:07:25,348][TRACE][discovery.zen            ] [cn2] full ping responses: {none}
[2016-01-12 20:07:25,348][DEBUG][discovery.zen            ] [cn2] 
filtered ping responses: (filter_client[true], filter_data[false]) 
{none}
[2016-01-12 20:07:25,357][INFO ][cluster.service          ] [cn2] 
new_master [cn2][nawV8B0mT9eBk05Nan_OyQ][host][inet[/IP:9300]], reason: 
zen-disco-join (elected_as_master)
[2016-01-12 20:07:25,529][TRACE][discovery                ] [cn2] initial state set from discovery
[2016-01-12 20:07:25,530][INFO ][discovery                ] [cn2] test/nawV8B0mT9eBk05Nan_OyQ

(Mark Walkom) #2

Why such an old version? 1.1.X is no longer supported!


#3

Because we work with redhat 5.8, the newest version could not be installed due to changed signature.

Now I've installed the version 1.3.2 on both cluster nodes. But the behaviour is the same. The data added to one node are not visible in the other one, the health request delivers just information about the own node.

Do you have an idea where is the mistake?


(Mark Walkom) #4

Sounds like they haven't clustered, what does your elasticsearch.yml look like?


#5

First there were just two setting configured: cluster.name and node.name. It didn't work.

Now we added the configuration for
network.host
transport.tcp.port
http.port
and the nodes are connected now.

Nice to have a running system!


(system) #6