Madelyn Olson

> Yes. Didn't target the second one particularly. But having strict ordering and identical response helps with a simple string match and no additional parsing. This is why I think...

We should automate running with coverage too so they're attached to PR :)

> A successful PING/PONG is required as a sender for a given node to be marked as extensions_supported and then extensions message will be sent to it. This could cause...

> @hpatro please continue with this PR. I will be the "designated" reviewer :-) Reminds me of the fact we should add code owner when we get this all sorted.

> I think we should try our best to make it happen. It is an important and fully compatible fix. Yeah, I was pushing for this because I think it's...

> @madolson Regarding your concern of displaying IP address incorrectly for a period of time, I think we should decouple it from this issue. And I think that issue was...

> Exactly, this issue exists for a new node being added and not related to the upgrade scenario. So, it could happen for any cluster running with hostnames feature. Today,...

> Won't that node be added via gossip as well and in the gossip information, hostname won't be present. Am I missing something? 🙈 That was one of the two...

> The cluster bus is an eventual consistency system, and the information broadcast via the cluster bus including "hostname" is inherently dynamic/non-constant. Although generally this is true, in practice today...

> First and foremost, I am not sure about the importance/value of this property of "a node will never observe a node that doesn't have a hostname". This already can...