Whoops, sorry, the transport address didn't change, I had just done it against our Prod cluster the second time. I changed the node names with Firebug since we just use the server names.
Aw. Well, there goes that theory. At least know there's a workaround. 
Well, it partially works. Shards per node and Unassigned Shard Count is 0 (just upgraded the last of the 4 nodes to 2.3.1.
Screenshot
Just to add that we faced the same issue on our 9 node cluster and the node_resolver setting fixed it completely.
Thanks.
1 Like
@tebriel, I do not think that that issue is related. Do you mind creating a separate Discuss post for it if it's still happening?
I have tracked down the issue that is causing this problem and I expect to have the fix pushed back to v2.3.2, v2.4.0, and of course v5, barring some unexpected issue.
@pickypg now that my node is back to green, it's reporting the proper number. Maybe just a transient issue. I dunno. I can't reproduce it right now, so I'll report it if/when it happens again.
Thanks
1 Like
Marvel 2.3.2 was released today, which should have finally put the nail in the coffin for this issue!
@pickypg confirmed. Just upgraded the cluster and I'm all good. Thanks again!