Server status, uptime and journal entries for my homelab and services — updated at best effort.
Badges by Healthchecks.io, uptime monitoring by updown.io, journal entries are managed using jrnl.
Server status, uptime and journal entries for my homelab and services — updated at best effort.
Badges by Healthchecks.io, uptime monitoring by updown.io, journal entries are managed using jrnl.
This website has been moved from the blog to www subdomain, and /YYYY/MM/title to /posts/title URLs for blog posts. Redirects are in place so the transition should be seamless.
Why move? Well… I felt the site has outgrown the blog subdomain, it’s more than a blog — there are also computer lists, static pages and this status page. I do like the aesthetics of naked domains, but not enough to disregard the DNS disadvantages — so www it is.
I’ve been thinking about getting rid of the dates in the post URLs for a while. It’s always annoyed me a tiny bit that /YYYY/ and /YYYY/MM were not valid indexes, and thus not browsable.
Custom built, using xcaddy:
./xcaddy build \
--with github.com/caddy-dns/luadns \
--with github.com/mholt/caddy-ratelimit
With nodata diabled; Systemd-resolver did IPv6 lookup, got public CNAME and resolved everything to the local reverse proxy. Causing local A overrides to not work as intended.
But with nodata enabled; an APEX domain override caused all non local-data domains to fail lookup…
111 repos migrated from Gitea to Forgejo. About time — I’ve had Forgejo up and running for about a year, but never got around to do the migration.
Now I guess there will be a lot of changing .git/config files and minor adjustments. And I do plan to organize better than I did before, but it remains to be seen how that goes.
Also disabled the systemd-networkd-wait-online.service as it was preventing the VM to start, as I understood it because the docker network is listed with no-carrier.
Starting 07:30 my local DNS, Knot Resolver, was unable to verify the peer certificate of Quad9:
The certificate is NOT trusted. The revocation or OCSP data are old and have been superseded.
Big @problem, as no DNS queries could be forwarded to the upstream server. At 10:35 normal operations assumed and the problem was @fixed.
Fallback servers have been added to the Knot Resolver config to prevent this kind of issue in the future.