The issue we stumbled on is, that the auto-follow pattern is still in place and will start replicating when a new index is created at the leader cluster, which we would have to stop following instantly again to properly upgrade the cluster.
So my question in the end is, is there a way to pause and disable autofollowing for the whole cluster while in an upgrade scenario?
Thanks @kley for your continued usage and feedback for cross-cluster replication. You are right that auto-follow patterns can be a problem here, and there does indeed need to be a mechanism to pause them during the rolling upgrades on the two clusters. I have opened an enhancement issue on our GitHub issue tracker so you can track progress on this feature there. I can't make any statements about the priority of implementing this feature, so there might be some time before it's implemented, but I agree it's an important one for this use-case.
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.