My question is why it is specifically listed that a pre-GA version requires a full cluster restart, in comparison to a normal rolling restart that can be done from 5.6 or a different 6.y
Is there some flag in pre-GA versions that limits me from doing a rolling restart?
This is just safety for us: maintaining backward compatibility is extremely costly in terms of engineering, so we only support GA versions, and make sure we have the freedom to break anything from the wire format to the disk format between pre-GA versions and GA versions.
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.