If you have dedicated client nodes (called "coordinating only nodes" nowadays), you can point Logstash to those nodes. If you do not have dedicated coordinating only nodes, use the data nodes.
Whatever you do, if you have dedicated master nodes, do not send those master nodes any requests, ie. do not point Logstash to those nodes. Use the data nodes or coordinating only nodes.
The nodes that you choose to connect Logstash to should not impact the availability of ingestion statistics.
Thanks for your response. It clarifies my question.
Wanted to know one more thing, if I send my search requests to a specific Data Node, the statistics captured and displayed through Kibana would not be impacted?
What statistics are you talking about? The dashboards and visualizations you have created yourself? Unless you put a lot of pressure on your data nodes, those data nodes will be able to still respond to search and aggregation requests fine and your dashboards should work fine.
Don't send all requests to one specific data node though - load balance between the available nodes by round robining the requests.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.