HACKER Q&A
📣 algocli

How do you approach debugging production issues?


Logs, traces, metrics, reproduce locally? What's your mental framework when something breaks at 2am?


  👤 spaceprison Accepted Answer ✓
What stopped working

Were there any recent changes

How does it do the thing that stopped

What does that thing need in order to work normally

How does that thing tell us it’s working

How do the things that it needs to work work

Logs, traces, metrics to prove each; repeat

—-

Local repro is fine but if it’s not doing prod stuff with prod data it may not do much for you