Decommission druid historical nodes

Hi, I saw
and I wonder is there is plan to enable the option Maintenance mode on the druid web console, similar to the option “disable worker”

The point is to minimize the time to get the cluster stable after replace some data node (historical node) by allowing the segments to be fully replicated before do a shutdown graceful for the deprecated node to be retired. This looks really supportive on a cloud env where replaces nodes used to be planned but not as usual task as in pre-cloud era.

Or if there is any best practice to get this done.

Thanks in advance

Sounds like a good improvement for the Druid console, you can create the improvement request here: Issues · apache/druid · GitHub

1 Like

Hey @aureli.gomez welcome! So there is a dynamic coordinator configuration called decomissioningNodes that might be what you are looking for – although the last time I looked I don’t think that it’s particularly obvious on the console – it’s not like the icon for a Historical will go grey on the Services tab if it’s decommissioning, for example… but that would be cool! +1 on the issue raise for sure.