Logging Best practices

General approach to sys logging

• Use a wrapper over logging library

• Use cyclic files or UDP://syslog

• Use always the most vervose timestamps (first element in line)

• Use prefix related to filename/module (second element)

• Use as many IDS as available in a current execution scope (third element)

• Use just one line (as far as possible)

• Use [idParam = JSON] to represent data.

• Use just three levels: INFO/USER, WARN/ERROR, DEBUG

INFO:

• Incoming request + data

• Request end + summary result

ERROR / WARN:

• Weird states, possible bugs

• Exceptions||if(err)

• Exceptions scope (all related objects)

• Service accessibility (DB connection error…​)

DEBUG:

• In/Out for every public method

• Params for every public method

• Return/Callback + data for every public method

• Performance metrics (TPS)

• Sys monitoring (TOP / REDIS): may be periodic