Lectures

We break down why a bunch of println calls is still not logging, introduce the DEBUG/ INFO/ WARN/ ERROR levels, the concept of a minimum level ( minLevel), and build a minimal logger: enum class LogLevel, a single line format timestamp [ LEVEL ] message, filtering by importance (via ordinal or an explicit priority), wrapper methods debug/ info/ warn/ error, and common mistakes when introducing logging.

We break down how to turn “scattered lines” into readable logs: introduce context as key=value pairs, store it in Map<String, String>, format it via joinToString(), add context to the logger API ( Logger), and link events of a single operation using requestId. At the end: hands-on practice on a command handler and a list of typical mistakes.

We break down how to log exceptions so that troubleshooting is fast and accurate: why printing only e . message is not enough, why you need a stack trace, and how to conveniently get it via stackTraceToString(). We learn how to extend Logger to handle Throwable, how to properly “wrap” errors while preserving the cause, how to choose a single responsibility boundary for logging, and how not to duplicate the same stack trace at every layer. Finally, we cover nuances of reading a stack trace and common mistakes.

We cover basic observability in a console application: how metrics differ from logs, how to measure operation duration with measureTimeMillis, how not to smear timers all over the codebase, how to build counters with MutableMap, agree on key names ( "command.add.total"), and integrate timings/counters into command handling. At the end, there is a mini metrics report and a list of common mistakes.

We break down how to make logs useful instead of noisy: levels DEBUG/ INFO/ WARN/ ERROR, switching verbosity via the "--verbose" flag, and choosing minLevel. We learn how to write lazy debug (via a lambda) so we don’t pay for building messages when the level is off, and we add safe key=value context with automatic secret masking via maskSecret().