Cluster.routing.allocation.enable vs planned node outages, eg. rolling node restarts

Trying to recapture BCP for doing a planned node outage eg. by rolling restarts.

Wouldn't it be good practice to set cluster.routing.allocation.enable transient to 'none' before taking a node offline and the then set it back to 'all' after node rejoins cluster?

Trying to pick my brain if there were some settings controlling the timeout before shard reallocations kicks in after a node has left the cluster... or maybe I'm confusing it with Cassandra clusters.

The recommended steps for a rolling restart are covered in the docs. These docs also mention the index.unassigned.node_left.delayed_timeout setting which controls the timeout you are asking about.

1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.