Had a weird one today. Last one website of the several of on this server suddenly started returning 503 (service unavailable) errors. There was nothing in the PHP error log or Apache error log. All server configs are already thoroughly optimized for performance. Other websites on the same server were functioning normally.
I didn’t notice this at the time, but my uptime monitor didn’t report an outage. When I used redirect-checker.com to check the status code, it returned 200, which should have been a clue, also.
Next time, before doing all sorts of arcane troubleshooting:
1. Try with a different browser
2. Is there a CDN? Try bypassing it.
3. Are you using a VPN? Try selecting a different endpoint (VPN server) if it will let you, or turning it off.
I use the NordVPN plugin in Firefox, and quic.cloud is my CDN. It turned out I now couldn’t access my site or quic.cloud’s website with the VPN enabled without getting a 503 error, when nothing reelevant changed on my end in a long time, end both were accessible before last night. Disabling the CDN immediately restored access to my website. Turning off the VPN made both work again, regardelss if the CDN was active or not.
Refreshing my NordVPN connection, so it chose a different VPN server, solved the problems completely. My guess is that someone using the same VPN server launched an attack so quic.cloud blocked that one server.
One other note: I described the problem to Gemini to see if it had any troubleshooting tips and it walked me through a lot of fundamental server configuration changes, then began forgetting what it had told me to do and referring to its own suggested changes, some of which were breaking changes, as “your original settings”. Always make sure to manually write down every configuration change while troubleshooting: the path of the configuration file, the previous value or state, and what changes you made, so you can completely revert everything if it turns out to be unrelated to the problem.
