[anonygler]

"Once in a blue moon" -- you mean the thing that constantly happens? If you're not using logs, you're not practicing engineering. Metrics can't really diagnose problems.

It's also a lot easier to inspect a log stream that maps to an alert with a trace id than it is to assemble a pile of metrics for each user action.

[hot_gril]

I think the above comment is just saying that you shouldn't use logs to do the job of metrics. Like, if you have an alert that goes off when some HTTP server is sending lots of 5xx, that shouldn't rely on parsing logs.