New node finds master, fails to connect (ES 0.90.5, simple unicast cluster)


(Zachary Lammers) #1

Trying to add a new node to my test cluster (ES 0.90.5), my master is
10.1.2.92. All communication between are allowed (can telnet 9200
and 9300 just fine). New node (b212) finds master (b211) just fine, and I
can see traffic between the two hosts when running tcpdump on both, yet
b212 fails to join the cluster (same happens with new node b213, same
error, same config below):

[2013-10-21 12:58:27,335][INFO ][node ] [b212]
version[0.90.5], pid[28881], build[c8714e8/2013-09-17T13:09:46Z]
[2013-10-21 12:58:27,336][INFO ][node ] [b212]
initializing ...
[2013-10-21 12:58:27,342][INFO ][plugins ] [b212] loaded
[], sites []
[2013-10-21 12:58:29,360][INFO ][node ] [b212]
initialized
[2013-10-21 12:58:29,360][INFO ][node ] [b212] starting
...
[2013-10-21 12:58:29,523][INFO ][transport ] [b212]
bound_address {inet[/0:0:0:0:0:0:0:0:9300]}, publish_address
{inet[/10.1.2.93:9300]}
[2013-10-21 12:58:53,574][WARN ][discovery.zen ] [b212] failed
to connect to master
[[b211][kOaHix9tQSWhV5gDmPmgpA][inet[/10.1.2.92:9300]]{data=false,
master=true}], retrying...
org.elasticsearch.transport.ConnectTransportException:
[b211][inet[/10.1.2.92:9300]] connect_timeout[30s]
at
org.elasticsearch.transport.netty.NettyTransport.connectToChannels(NettyTransport.java:675)
(...more...)
at java.lang.Thread.run(Thread.java:679)
Caused by: java.net.ConnectException: Connection timed out: /10.1.2.92:9300
at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)
(...more...)
at
org.elasticsearch.common.netty.util.internal.DeadLockProofWorker$1.run(DeadLockProofWorker.java:42)
... 3 more

My config for b212 and b213 is pretty basic at this time (master b211 has
same config, but with master=true; data=false):

cluster.name: beach
node.name: b212
node.master: false
node.data: true
index.number_of_shards: 5
index.number_of_replicas: 0
path.conf: /opt/beach/etc/default
path.data: /var/lib/elasticsearch
path.logs: /opt/beach/logs/default/
bootstrap.mlockall: true
discovery.zen.ping.multicast.enabled: false
discovery.zen.ping.unicast.hosts: ["10.1.2.92", "10.1.2.93", "10.1.2.94"]
indices.memory.index_buffer_size: 80%
index.refresh_interval: 30
index.translog.flush_threshold_ops: 10000
index.store.compress.stored: true

I can't for the life of me figure out what's wrong here, especially as I
can telnet to the open ports to/from each server to the other. I've had
VM-based multicast test clusters up no problem, but this is my first
unicast cluster, so I was wondering if my error was around that portion.

Thanks in advance,

-Zachary

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


(system) #2