Currently we can do the following:
- rolling-upgrade (equivalent to upgrading the plugin manually and then doing a rolling restart)
We want to be able to do this as well:
- rolling-upgrade with full reboot
Because of the current structure, we should implement this as a new cookbook to unblock currently-needed Elasticsearch reboots, but at a later point we should circle back and refactor the elasticsearch cookbooks in general. We can combine the above 4 cookbooks into a single one that accepts two additional flags: one to tell it whether to reboot or to just restart, and another flag like --upgrade-plugins that tells the cookbook to do the plugin upgrade as well.