I created an index with number of shard : 3 & number of replica : 0 and inserted some documents in it. After some time one of my node is down and cluster health goes to RED. Now I again insert some documents in index.
Expected outcome
All documents should be inserted in index. Elastic should insert document in remaining primary shard.
Actual outcome
some documents inserted in index while some failed due to unassigned shard.
I Created 3 node Elastic cluster with discovery.zen.minimum_master_nodes: 2.
We are designing ES Cluster. How to design the cluster so that complete cluster will not go down due to one node. It is acceptable to have data un-available which is present on the node which is down, till it will not get recovered.
Replicas are hurting our performance, as the same documents will get indexed twice. So we don't want replicas to be present. Replica also calls for more memory and hardware.
So can we achieve the same without replica.
Actually in our case we do not require High availability, We only require to scale the index by increasing shards. As by increasing shards and having shards on different machines, we can increase our indexing and searching power.
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.