Document the connection-hang incident and its cause

A full write-up of the intermittent five-minute connection hangs, so the next occurrence is recognised in minutes rather than days.

Covers the symptom and how to tell it apart from a slow server, the evidence gathered at each layer, the host limit that caused it, why every health check stayed green throughout, the fix and how it was verified, and the diagnostic commands to confirm it directly on a machine.

It also records what the investigation got wrong on the way — including a hardware migration completed while chasing it, which changed nothing — and the checks that would have caught it sooner.