My question is: Does this need to be done for master nodes?
I reckon this is only needed for data nodes and NOT needed for master nodes. Can someone confirm? The main purpose of disabling and enabling is to prevent un-necessary movement of shards across nodes but master nodes don't contain any shards in first place.
Thanks for your answer. I am aware it's a cluster level setting. The first point of rolling upgrades asks to disable shard allocation and my question was whether this needs to be done for upgrading/restarting of master nodes? because a master node doesn't contain any shards in first place and so I was wondering if we can skip disabling and enabling of shard allocation when we are doing rolling upgrade on master nodes. Got my point?
IMHO, the disabling and enabling of shard allocation steps are moot for master nodes. Even though the setting is at cluster level, it needs to be done only when restarting/upgrading data nodes. This i what I understand and was looking to confirm if my understanding is correct or not.
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.