Problem Definition

During a node startup on Confluence Data Center, it will run an index validation to see if it needs to be recovered (from a snapshot or from another node) or fully rebuilt (if recovering it isn't successful).

Sometimes, for maintenance purposes, you would like to restart a Confluence node with a known outdated (or corrupted) index without triggering an index recovery nor the index rebuild.

While preventing the node from running an index recovery is possible through a JVM property (see confluence.cluster.index.recovery.num.attempts in Recognized System Properties), preventing the node from starting a reindex isn't.

This is also helpful when we need to restart a node while another one is running a reindex.

Suggested Solution

Make a configuration available so we can prevent a node from running a full reindex on startup.
It could be a JVM property similar to confluence.cluster.index.recovery.num.attempts.