Druid Cluster: connection for Master server from data server

I’m using Druid-0.14.1

I have 3 nodes (Master, Data abd Query servers). Master for coordinator and overlord, Data for Middlemanager and Historical and Query server for broker.

I used bin/post-index-task --file quickstart/tutorial/wikipedia-index.json for load data and test my cluster.

In Data server I have errors on logs:


WARN [WorkerTaskManager-CompletedTasksCleaner] org.apache.druid.discovery.DruidLeaderClient - Request[http://localhost:8090/druid/indexer/v1/taskStatus] failed.

java.net.ConnectException: Connection refused: localhost/


WARN [main] org.apache.druid.discovery.DruidLeaderClient - Request[http://localhost:8081/druid/coordinator/v1/lookups/config/__default?detailed=true] failed.

Connection refused: localhost/

I set "druid.host" in all my config files but these ips ignored by these processes and use for coordinator and overlord.

`Now I need to know how to set IP for ```coordinator and overlord processes in Data Server.``

Thanks and regards.


To maybe know where the problem we should take a look at your config.

Could you send the config files from your nodes please ?

Is your zookeeper throwing some errors ?

I don’t have any errors on my master server and zookeeper.

I see logs for all three servers and my middlemanager and historical have errors.

These are my config files. I think my middlemanager and historical can’t find IP of Master server.

druid-Data-server-config.tar.gz (14.8 KB)

druid-Master-server-config.tar.gz (3.51 KB)

druid-Query-server-config.tar.gz (13.3 KB)

This is my output for testing cluster. Task finished with success status but never finish loading.

Please share the logs of all the servers.

On the master server I see druid.zk.service.host= in the common.runtime.properties. On the other nodes I see Can you check if this is correct ?



Yes This config is true. I installed Zookeeper on Master server and it is listening on all ports.

These are my log files.

IPs don’t set correctly on cluster nodes and they use instead of the correct IP.

I don’k know which config is missing.

broker.log (264 KB)

coordinator.log (164 KB)

historical.log (256 KB)

middleManager.log (188 KB)

overlord.log (133 KB)

The reason was config of zookeepers on nodes. :slight_smile: