On Fri, Jul 26, 2024 at 06:18:00PM +0000, Antoni Przygienda wrote: > Well, you cannot design protocols around broken platforms since the > spec needs to hold up for 20 years, the platforms are fixed daily.
We do design protocols to operate in a given environment though, and IS-IS doesn't require a synchronized clock at all, just a roughly syntonized one. If we were to add such a requirement, I'd be raising serious concerns at this point. For a debugging aid, I don't care, I'm just pointing out what current reality looks like. I'll also point out that if a router acquires its clock primarily from NTP, it has no way of getting at the TAI-UTC offset from there. NTP only carries differential leap seconds signalling, i.e. +1/-1 "as it happens". It would need to consume a tzdb leap-seconds file from somewhere else, which if the stack doesn't implement PTP, it might not even have an option for. -equi _______________________________________________ Lsr mailing list -- [email protected] To unsubscribe send an email to [email protected]
