We've used log levels inconsistently; document a standard practice, like:
- CRITICAL: wake me up at 3am; example: the lambda can't connect to the database
- ERROR: don't wake me up, but definitely tell me about; I probably gotta fix it
- WARNING: something the software can recover from, I don't need to be alerted on my phone
- INFO: something interesting happened, perhaps a significant milestone, progress in an operation's lifecycle, to answer customer questions like "what happened to that file I uploaded"
- DEBUG: noise, probably should be turned on in Dev, QA systems, and off in Prod
We've used log levels inconsistently; document a standard practice, like: