Hi Timo, "Timo Wilken" <[email protected]> skribis:
> Last night, this hit the exact moment of the CET/CEST change and seems to have > confused the Shepherd a lot, to the point where it kept spamming the same > message to /var/log/messages, and used up 80% of my system's total memory, > according to htop. (This led to various things getting oom-killed, and the > system grinding to a halt until I had a look at it in the afternoon.) > > First, at 01:00, the Shepherd spawned about 30 processes for this one timer: > > --8<---------------cut here > 2025-03-30 01:00:00 localhost shepherd[1]: Timer 'guix-gc' spawned process > 1691. > 2025-03-30 01:00:00 localhost shepherd[1]: Timer 'nextcloud-cron' spawned > process 1692. > 2025-03-30 01:00:00 localhost shepherd[1]: Registering new logger for > nextcloud-cron. > 2025-03-30 01:00:00 localhost shepherd[1]: Timer 'gpsplot' spawned process > 1693. > 2025-03-30 01:00:00 localhost shepherd[1]: Registering new logger for gpsplot. > 2025-03-30 01:00:00 localhost shepherd[1]: Timer > 'dynamic-dns-ipv4.api.mythic-beasts.com' spawned process 1694. > 2025-03-30 01:00:00 localhost shepherd[1]: Registering new logger for > dynamic-dns-ipv4.api.mythic-beasts.com. > 2025-03-30 01:00:00 localhost shepherd[1]: Timer 'guix-gc' spawned process > 1698. > 2025-03-30 01:00:00 localhost shepherd[1]: Registering new logger for guix-gc. I saw the exact same problem on berlin (aka. ci.guix.gnu.org): https://issues.guix.gnu.org/77401 but not on the other servers I co-administer; I suspect the difference is that these other servers did not have a timer running at 01:00. So I guess that the fix for <https://issues.guix.gnu.org/75622> was bogus somehow? To be continued… Ludo’.
