I have a cluster of 3 nodes with 3 primary shards and 2 replica sets. If Node 1 goes down, will the cluster attempt to create whatever shards existed on Node 1 on the remaining nodes?
I am sizing for the index storage requirements of the cluster and want to prevent the scenario of lack of storage space in case a node goes down.
If you have 2 replicas configured, all shards will already exist on all nodes. Elasticsearch will not allocate two copies of the same shard to any node, so if one of your nodes fail, your index will go yellow as the replica can not be allocated.
Yes, in that case new replica shards will be allocated in order to maintain the replication factor, which will mean increased disk usage on remaining nodes.
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.