I saw similar questions but no answers (sorry if this is a duplicate still):
Am I correct in assuming that if I change a Master + Data node to just a Master only node that the shards that we're on that server will get replicated to other Data eligible nodes?
I don't think so. But I never tested that scenario.
What I'd probably do in that case is to manually exclude the IP of the master+data nodes for all indices then wait for relocation then switch off the node, change it to master only and restart.
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.