Elektrine lite

← Feed

@ordnung@chaos.social

Post #1282010

2026-03-07 23:26 UTC

Looks like the problem is solved. Systemd-resolved was in a strange stage where it still resolved requests via dig as usual but for example a curl or mtr timed out on the DNS stage. Restarting systemd-resolved solved the problem. #mastoadmin

Replies (7)

  • @zornem@chaos.social 2026-03-07 23:28

    @ordnung thank you!

    Open ##1282013

  • @Jay16K@chaos.social 2026-03-07 23:29

    @ordnung sorry, not sorry But thanks for keeping it running ❤️

    Open ##1282015

  • @manawyrm@chaos.social 2026-03-07 23:33

    @ordnung I've seen this exact issue in prod sooo, sooo many times and I still can't figure out what's wrong with the thing. Ended up replacing it entirely for prod machines, but I'd still really like to know what's happening there. There's some issue that looks the same, which sometimes (but not reproducibly) occurs when IPv6 addresses change while resolved is running, but I'm not sure if it's the same issue as the sudden/random hangs... *sigh* :( *hugops*, thanks for your time and effort!

    Open ##1282019

  • @sandzwerg@chaos.social 2026-03-07 23:39

    @ordnung thank you ❤️

    Open ##1282029

  • @jay9@chaos.social 2026-03-07 23:47

    @ordnung thank you! <3

    Open ##1282033

  • @RonjaBiernat@chaos.social 2026-03-08 05:24

    @ordnung thank you!

    Open ##1282036

  • @jssfr@zombofant.net 2026-03-08 15:22

    @ordnung One thing I found out at some point is that curl sets some DNS options which dig does not set by default. That has already caused a lot of headaches because those options factor into the cache key (at least in some implementations), so while changing records in a zone, one might see a difference between curl and dig and be very confused. I wonder whether those same DNS options caused this issue.

    Open ##1282054