bug#72166: Shepherd periodically goes unresponsive on one of my machines

[Jonathan Frederickson]

Aug 22, 2024 05:35:47 Ludovic Courtès <ludo@gnu.org>:

Hi,

You wrote:

The machine in question does not have a battery-backed RTC, so it
loses time when it loses power, but notice the time changes once ntpd
starts up.

So you’re hitting the Fibers bug described here:

  https://issues.guix.gnu.org/70848

There is still no fix upstream unfortunately, but I hope we’ll get there
in time for Shepherd 1.0.

The only workaround so far is to build shepherd against
‘guile-fibers-1.1’, which is what is done in Guix on AArch64 and RISC-V
since these platforms are likely to be used with single-board computers
lacking a battery-backed RTC.

Thanks,
Ludo’.

Ah! Yes, that seems likely.
  
  
I've also run into what looks like the same thing on a laptop that does have an RTC, but it also has an NTP daemon running on it. (I have also noticed that this is most common on that laptop after a suspend/resume cycle, so maybe that's triggering the bug as well?)
  
  
I'll check to see if I've recently had an NTP sync the next time this happens, because that would be a definitive confirmation that I'm running into that bug.