We've had several bug reports here on Meta about server 500 errors, and we know that these can deter participation. That's bad at any time but especially when we are trying to grow. We want to do better.

So far, our developers with database access have been looking these up when there's a bug report and addressing the problem, but we'd like to broaden the pool of people who can help. Every error is logged, but not all logs get investigated. If you're a developer, SRE, or DevOps person and you'd like to help us fix 500s, read on. Our hope is that reviewing the set of errors, rather than looking at them one at a time, will yield some root causes that we can fix -- fix several errors at once, instead of fixing them one at a time as they come up.

We believe that, in most cases, the records of these errors in the database contain all the information needed to debug them. Our developers can pull these records for all the 500s we've gotten and share them (no PII).

Would anybody like to help us investigate these errors? Right now the people who can investigate them are on other critical paths; if there are folks who'd like to help, we'd like to help you help us.

I'm making this post, instead of just asking for a dump to be posted somewhere public, to find out if there's interest first. This way we can work with you to figure out the best way to package this information and update it with new reports.

posted almost 3 years ago

CC BY-SA 4.0

Monica Cellio‭ staff

64 377 2269 838