Odd behaviour / dropouts on rock solid FTTH (Sky/Openreach)

We’ve got a rock solid 1000/110 Sky FTTH line, using a BT ONT and a UniFi UDM-SE. I’ve got smokeping running on a host internally.

Pings to Cloudflare/Google/… sit at a rock steady 2.5ms no matter what, and the jitter is basically nothing.

A few times in the last weeks I’ve woken up to no internet. The internet has dropped off at 4.01AM. The day before, from about midday, pings have gone a bit wild (20ms with loads of jitter - not terrible, but atypical for the line). The only thing that fixes it is an ONT reboot. USG reboot or cable pull does nothing. Normal lights on the ONT prior to the reboot.

This feels like an open reach / sky maintenance thing - is this normal for such maintenance to leave things disconnected?

To me this feels like an ONT issue. When the optical network goes down to whatever extent, the ONT goes into a state for which restarting the optical network does not provide an event which triggers the ONT to leave that state. It may be that the optical network could provide the trigger, but it is not sent, or it may be that there is no event which can trigger the restart.

As I understand it, Sky uses IPoE rather than PPPoE.

It's possible that your path through Sky's network is "sticky" if you just restart your router, but gets freshly allocated if you reset the ONT. But that's just a guess.

It could certainly be worth trying a different router for a few days (especially the Sky-provided one if they did) and if the problem reproduces with that, it's much easier to escalate to Sky tech support.

Have you tried pulling the fibre out the ONT and replacing it?