We are a bit light on error messages so far about akka-edge-rs HOT 2 CLOSED

Yeah, always difficult to judge the level when logging from a library. Some guidelines:

ERROR is for critical problems, such as bugs that should not occur. It should be possible to run a production system without ERRORs, and if they happen they should be good candidates for alerts.

WARN is a problem, but the system will be able to recover by itself, e.g. retry.

INFO is just informational about ordinary events, such as successful startup of important things, but not too much.

DEBUG and TRACE is for troubleshooting, understanding details of what is going on. Normally disabled in production. It should still be possible to turn on DEBUG in production without causing severe performance degradation because of logging. TRACE can be more chatty.

from akka-edge-rs.

ERROR is for critical problems, such as bugs that should not occur. It should be possible to run a production system without ERRORs, and if they happen they should be good candidates for alerts.

This should be addressed by panics. If there's an irrecoverable error then a panic will occur and a stack trace written to stderr (depending on the panic handler, which can be changed).

WARN is a problem, but the system will be able to recover by itself, e.g. retry.

i.e. operator audience. I think this will become the major focus of addressing this issue.

INFO is just informational about ordinary events, such as successful startup of important things, but not too much.

i.e. operator audience. We've got some of this, but I'll look further.

DEBUG and TRACE is for troubleshooting, understanding details of what is going on. Normally disabled in production. It should still be possible to turn on DEBUG in production without causing severe performance degradation because of logging. TRACE can be more chatty.

i.e. developer audience. This one is trickier because we don't know what trouble we're up for ahead of it occurring. :-) We've got some of this but I'll revisit.

from akka-edge-rs.