In elasticsearch with replica of 1 when a node goes down one other node is
still receiving the writes. What happens when the failed node comes up and
has inconsistent data? Could reads return inconsistent data for some of the
queries?
When a failed nodes comes up, the cluster will start a recovery of the
indexes on the node, and for each index, a node with obsolete shards will
receive the latest shards. During recovery, the shards are not open for
read.
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.