We've changed some parameters regarding the increasing number of 502 errors we see on the instance. We would kindly ask you to report us such 502 errors in the next few days with the exact time you see them.

We got no feedback on any 502 errors in the last 11 days so it looks like the problem is finally solved 🎉

The only problem is, we changed two things. We might change one back for a short period of time in the next few days to check which of both lead to the problem. So if you experience 502 errors it will only be for a short time but please still report them.

Show thread
Follow

We now know what the cause of the 502 errors was. The backends of the mastodon application bind only to ipv4 by default. We used the name localhost on the nginx reverse proxy to refer to the backend. This, as usual first resolves to ::1 the local ipv6 address. As there is no backend listening under this address we need to fallback to ipv4 and we think this was to slow in some cases leading to the user facing 502 error.

· · Web · 2 · 8 · 34

@ordnung what was the solution? enable ipv6 for docker or change localhost to 127.0.0.1?

@gcrkrause we don't use docker. And the solution was to change it to 127.0.0.1, so yes.

@ordnung Nice find! That's not the sort of failure mode you see every day.

Sign in to participate in the conversation
chaos.social

chaos.social – a Fediverse instance for & by the Chaos community