a firmware issue does not necessarily show up all the time and causes flood of complaints, With packet coming and the box software routing it (as bridge) and user use the phone at the same time, with quality off service (QOS) enforce, all of them happens in parallel and are dynamic and there are matrix of scenarios that software has to handle properly (the so called race condition may pretty hard to reproduce and hit as you need to came cross this transient state exactly, but once hit, and if there is a bug, it sets the software into unknown state.
As I said the box is used as a bridge, that keeps working and can route network activities, so you do not need to use ping to know ping is still working as as ping's icmp sits in between layer 2 and 3 on tcp/ip stack, if it breaks, the birdge mode cannot be still working.
A simple power recycle of the box solves the issue w/o any cable readjustment also demonstrate it is the software inside the box that got partially brain dead.