There’s a leap second coming up. Here’s a story of woe from the previous one, in 2012, from a sysop at fastmail, who had several redundant systems in several worldwide datacentres, failing apparently at random.
via the discussion at
https://news.ycombinator.com/item?id=9567761
which tells us that Amazon are going to skew the extra second in over a period of 24 hours - not totally unlike Google’s approach, although I don’t think they’ve said what their skew window is. Either way, times will be adrift by a half-second just fore and aft of the leap.
From the article:
“with very little work, even while not totally awake, the entire impact on our users was an ~15 minute outage for the few users who were primary on imap3 (less than 5% of our userbase) – plus delays of 5-10 minutes on incoming email to 16.7% of users each time a compute server crashed. While we didn’t do quite as well as Google, we did OK!”
Edit: FTR, there’s a link to
https://www.cl.cam.ac.uk/~mgk25/time/utc-sls/
which describes UTC-SLS, a client-side slew in the final 1000 seconds before a leap, which allows time to be correct on every hour and half-hour but still monotonic. The document notes that Google’s approach is server-side: the NTP servers never announce a leap second and instead distribute a slewed version of time. Whereas Amazon slew over 24 hours, Google have not said exactly how they do it.