Add health check which doesn't go upstream

[j-m-harris]

This addresses a problem we saw with having a TCP 'socket open' health check for amqproxy (e.g. in a k8s environment). If debug was enabled then we would see socket EOF errors, as amqproxy tries to read from the client.

This health check is specifically for amqproxy, and not testing the upstream connection (we could use an AMQP heartbeat frame for that).

• Client sends AMQP9999 only.
• amqproxy returns HEALTHOK and closes the connection.

[kickster97]

Thanks for the contribution, looks interesting!
We have considered implementing something more generic, like an http health check with a prometheus endpoint to get better monitoring.

[j-m-harris]

Thanks for the contribution, looks interesting! We have considered implementing something more generic, like an http health check with a prometheus endpoint to get better monitoring.

Fair enough, that sounds good!

[carlhoerberg]

Could also just debug log EOFErrors during the client negotiation phase so that they are easily ignored.