Previous incidents

We traced the errors to requests that intermittently stalled in our backend's hosting environment (Vercel Edge) and then hit the platform's 25-second time limit. Our database and services stayed responsive throughout.

We have moved the backend to dedicated, always-on servers in Frankfurt, close to our database, and optimized it for production load. After testing it under load, we completed the switch at 21:50 CEST, and all traffic is now served by the new setup. Response times are back to no...

Website is back up.

Our hosting provider Hetzner had a failure in their DNS servers, the internet’s “phone book” that turns names like api.sanity.io into addresses. Our website couldn’t look up the services it depends on, so every page waited about 10 seconds and the load balancer took the servers offline, which showed visitors an error page. Nothing in our code changed. We switched our servers to Cloudflare’s and Google’s DNS servers instead of Hetzner’s. Pages load normally again (under 0....