When a node is recovering an index from snapshot it does not return 503

Issue Summary

When a node recovers an index from a snapshot it does not return 503 Maintenance as it does when you do a full reindex, instead it returns a 200 and it still allows user traffic.

This means that users can access a node without a working index, which results in incidents.

Jira should block access to a node that does not have a working index.

Steps to Reproduce

1. Perform and index recovery on a node
2. Confirm you can still access that node while recovery is underway

Expected Results

A node should block while it does not have a working index, the same as it does it you perform a full reindex.

Actual Results

The node is available and users can access it, but without an index they cannot view issues, reports, etc.

Workaround

1. Manually remove the node from your load balancer while it is recovering the index
2. Watch the logs for progress
3. Once complete, manually re-add the node into the load balancer

NB: This affects all versions of Jira