ES 6.0 timeout on cluster

(Rakotobe) #1


I've 3 differents server with 1 instance of ES 6.0 on each and another server with nodejs, to query on.

On each server I just changed : : [ LIST_ES_IP ]
discovery.zen.minimum_master_nodes: 2

My problem is that after some times (not define), I've timeout error from nodejs server. But if I call

curl -XGET 'IP:9200/_cluster/health?pretty'

On this same server, I can see that ES works fine.

If I remove one server from cluster (comment previous 2 config lines), and query only on it, all works good & I never have timeout.

Did I need to change another config to make this cluster works ?
Did you have ideas about why I've timeout only on cluster mode ?

Thanks in advance,

(Rakotobe) #2

Apparently it's cause on elasticsearch-js client because I reactivate cluster, but define host to


and it works for 3 hours now.

Before I've

[ "IP1:9200", "IP2:9200", "IP3:9200" ]

I try with

[ {host: "IP", port: 9200}, {...} ]

But time out too..

So no way to have rollback if one server failed ?

(andy_zhou) #3

this i think it is IP:9300 or not use port..

(David Pilato) #4

It depends. If it's in the context of the JS client, 9200 is ok. If it's elasticsearch.yml, then it's indeed 9300.

(Rakotobe) #5

Yes previous configuration is on JS (node) client via elasticsearch-js

I've same problem with a python client (elasticsearch-py). If I try to configurate with multiple host, it throw timeout.

On elasticsearch.yml I just not write port

(andy_zhou) #6

is about this?

discovery.zen.publish_timeout: 180s
discovery.zen.fd.ping_interval: 60s
discovery.zen.fd.ping_timeout: 180s
discovery.zen.fd.ping_retries: 10
discovery.zen.commit_timeout: 180s

(Rakotobe) #7

Hum it's works better from 1hour for now.

I'll continue to make some tests comming days.

Thanks a lot if it's solved by this config.

(Rakotobe) #8

Nope already timeout. Really don't know why..

I transfer my old ES2.1 data on my ES6.0 cluster (3 200 000 products), and now I've timeout even if I connect to only one node.

(andy_zhou) #9

how many bulk onces?
show the cluster status, show the nodes status. show the logs . about master and data nodes.

(system) #10

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.