Writing to an ece run elastic from elastic-spark connector fails

[rkalluri (rajesh kumar kalluri)]

I have about 600M rows or 200GB of data being written from Hive to elastic via elastic-spark connectors.

Things start out fine and slow down progressively and around mid point I get the following exception on the spark client side.

Caused by: org.elasticsearch.hadoop.rest.EsHadoopNoNodesLeftException: Connection error (check network and/or pr
oxy settings)- all nodes failed; tried tried [[http://77eba63e96b349a88b3ce7560af43d4a.1.2.3.4.ip.es.io:9200]]

My elastic cluster is a 4 node cluster with 8 GB RAM each.

I have not setup load balancer in front of the proxy, but specified all the proxies via the es.nodes=http://1.2.3.4:9200, http:2.3.4.5:9200 etc

Any tips on how to troubleshoot?

[Yuri (Yuri Tceretian)]

I would suggest you to check cluster's metrics and logs. Logs and metrics are shipped to cluster logging-and-metrics. On the cluster details page you can find links that navigate to Kibana assigned to the logging cluster with pre-defined filters.

[rkalluri (rajesh kumar kalluri)]

The cluster working fine after wards, what are we looking for in the metrics cluster anything specific.