Kibana shard timeout setting

I am using elasticsearch 2.3 and Kibana 4
In Kibana dashboard, if I add searches, sometimes, kibana complains "2 shards failed"
I noticed in Kibana settings the default setting of elasticsearch.shardTimeout is 0, so I changed it to 5 minutes.
I want to know that if I change the shard timeout to a long period, will it affect the overall performance of Elasticsearch?

Did that solve the problem of the failed shards notification in Kibana?

It will only cause connections from the Kibana client to Elasticsearch to stay open longer.

Yes that solves the problem of failing shards.

I want to know more about the reason behind it, why the default timeout setting is 0? Will it affect the overall performance of data nodes if I set the shard timeout too long?

In performance tuning Elasticsearch, typically users aim for speeding up either the indexing performance or the search performance. When the cluster needs to be able to handle a lot of bulk requests for incoming documents without losing any data or uptime, that sometimes means there is a sacrifice in search performance. Sometimes it is intentional, where users are willing to deal with extra time waiting for search results as long as they have peace of mind that the cluster is managing the load of incoming data to their expectations. Other times it could be unintentional and perhaps the cluster hardware is not sized appropriately, or configuration options are not set correctly.

The documentation has How-Tos about tuning Elasticsearch for indexing performance and search performance. Unfortunately it looks like this How-Tos are only relevant for v5.x:

If you want to know more about how to choose the best hardware, cluster size, and bulk indexing options for your use case, you would certainly get better discussion in the Elasticsearch category of the forums. :slight_smile:

1 Like

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