Ensure fail gracefully: No action required if a node fails

[jamesmontalvo3]

(assuming the node has redundancies)

[jamesmontalvo3]

Initial testing, shutting down servers:

• Kill replica server: no loss of functionality. Perhaps a bit slower either due to fewer resources or due to MediaWiki continually checking to see if it can use the nonexistent server.
• Kill master server: error message "experiencing technical difficulties"
• Kill elasticsearch server: Searching fails, at least initially. I only did restarts of the servers. Further testing required with full shutdown.
• Kill parsoid server: not tested. Having other issues with parsoid
• Kill app server: Service unavailable 503 error