I have noticed a few things:
- A node which has not fully booted is (perhaps correctly) shown in the “Servers Down” area. In my opinion, this is not good because it conflates a Pending Good state with an Active Bad state. It would be great if nodes were, from the point when the first Couchbase server process starts, until say, 3 minutes later, were considered automatically in a “starting” state by the console.
2 While I’m at the business of making suggestions, another “interim” state is when a node is in the middle of joining a cluster, perhaps these two items could be grouped, as “transient” states, like I’m showing in this picture:
I think a category for cluster nodes that are in Transient/Temporary states, that should be expected to become Green or Red after some maximal time has gone by, should be moved out of the bad states.
That makes it a lot easier to know if a cluster is healthy, at a glance.