The 500 errors were caused by a bug in recently updated code of our load balancer configuration. Specifically, the auto-redirect for www subdomains was broken. We have deployed a fix to our production servers, and everything is back to normal.

We are sorry about this! In addition to the bug we fixed, we also fixed a misconfigured alerting rule, which prevented us from noticing this issue immediately. We will take further steps in regards to integration-testing our infrastructure continuously in order to prevent accidental misconfigurations even better.