fix: The default log levels for different NODE_ENVs are surprising

[theneva]

Hi!

I set the environment variable NODE_ENV to production in my Docker image to avoid installing dev dependencies and having libraries like envalid work as intended.

As a side effect of this setting, Unleash sets log4js's log level to ERROR:

unleash/lib/logger.js

Lines 5 to 13 in 3a983d2

	function getDefaultLogProvider() {
	let level;
	if (process.env.NODE_ENV === 'production') {
	level = log4js.levels.ERROR.levelStr;
	} else if (process.env.NODE_ENV === 'test') {
	level = log4js.levels.FATAL.levelStr;
	} else {
	level = log4js.levels.DEBUG.levelStr;
	}

I find this behaviour quite surprising. I'm pretty much always interested in INFO and up even in a production environment.

I also find it strange that test environments only log FATAL messages, even though the log provider validation doesn't check for an implementation of fatal logging:

unleash/lib/logger.js

Lines 37 to 45 in 3a983d2

	function validateLogProvider(provider) {
	validate(typeof provider === 'function', 'Provider needs to be a function');
	const logger = provider('unleash:logger');
	validate(typeof logger.debug === 'function', 'Logger must implement debug');
	validate(typeof logger.info === 'function', 'Logger must implement info');
	validate(typeof logger.warn === 'function', 'Logger must implement warn');
	validate(typeof logger.error === 'function', 'Logger must implement error');
	}

Am I missing a way to override the log level in a production environment without providing my own logger implementation? If not, exposing a config option and/or an env var (for example, LOG_LEVEL) would be great!

[theneva]

As a side note, we set NODE_ENV to "production" in the official Docker image, which kills all logging you might expect to see from the app when reading the docs Unleash/unleash-docker#36 (comment)

(Maybe it's better to RUN NODE_ENV=production npm ci instead of doing ENV NODE_ENV=production…)