Thanks for the KIP. Overall, it looks good to me. I really like the
One minor comment regarding the `old-client-connections-count` metric. Is
it really necessary? The number of connected clients whose version is not
known (prior to KIP-511) is already reported but with an "unknown" software
name and an "unknown" software version, which is, I suppose, similar to what
you intend to expose with this new metric, isn't it?
On Thu, Apr 16, 2020 at 7:24 AM Boyang Chen <[EMAIL PROTECTED]>