Elasticsearch indexing not working and error message: node null not part of the cluster Cluster [elasticsearch], ignoring

For those who met the same error and came here from search engines, make sure the TransportClient is using the same cluster name as the cluster itself.

1. Verify cluster name.

Visit http://localhost:9200 to check the cluster name. The default name is elasticsearch. If you customised the cluster name using elasticsearch.yml file, make sure the config file is picked up.

2. Set culster.name when creating TransportClient.

   Settings settings = ImmutableSettings.settingsBuilder()        .put("cluster.name", clusterName)        .put("client.transport.ignore_cluster_name", false)        .put("node.client", true)        .put("client.transport.sniff", true)        .build();client = new TransportClient(settings).addTransportAddress(new  InetSocketTransportAddress(host, port)); 

3. Ignore cluster name check

You can ignore the check of cluster name by setting client.transport.ignore_cluster_name to true.

4. Debug if the error still exists

If the error still exists, launch your application in debug mode, and debug TransportClientNodesService.

I had changed name on my dev elasticsearch server for the sake of experimenting and forgot about it.

The error message on the client were not that helpful, TransportClientNodeService makes a compare with the remote name but doesn't actually write the remote name ("cluster-name")in the log.

It is possible to bypass the name check with the following Spring configuration:

The resolution for me was to either:

• Change name of the cluster.name in elasticsearch server.
• Ignore cluster name of the client.

I went for both, this is my Spring config, hope it helps:

spring:    ...    data:    elasticsearch:        # Defaults to cluster-name 'elasticsearch'        cluster-name:        cluster-nodes: 127.0.0.1:9300        properties:            # https://www.elastic.co/guide/en/elasticsearch/client/java-api/current/transport-client.html            client.transport.ignore_cluster_name: true