When an indexer is restarted during a normal rolling restart, does it shutdown via the `offline` method? Or does it do a normal shutdown process?
The documentation indicates that during a "searchable" rolling restart, each peer is actually offlined and not simply shut down. But for a normal rolling restart, how does the indexer shutdown?
How does this work from the cluster masters perspective? I've been unable to find rest calls against the restart or offline endpoints during a rolling restart, but maybe the CM isn't actually making those calls (or they aren't being logged for some reason)?