Hmm, this is not as simple. EUnit tests are failing in chttpd because per-node rexi servers are started up asynchronously after the mem3 app supervisor has already finished. So making cluster requests immediately after mem3 and rexi have started could fail.
In the tests, though it is assumed as soon as all the apps start requests should succeed immediately. I think we'd probably want to start server in the server_mon (and buffer_mon) in the init callback to avoid this asynchronicity in general. This would work because the supervisors (`*_sup`) come before monitors (`*_mon`). [ Full content available at: https://github.com/apache/couchdb/pull/1627 ] This message was relayed via gitbox.apache.org for [email protected]
