Adding logging to your code

Back to: Reference | Logging


Having Granitic inject a logger

Your code can send log messages via the logging.Log interface if the struct that provides the implementation for your component has a field with exactly this name and type:

  Log logging.Logger

Granitic will automatically inject a configured and ready to use logger into your component.

Logging messages

Your code can send messages to be logged using one of the LogXXXX methods on logging.Log. Convenience methods are provided to log messages at each of the standard logging severity levels.

For example, to log a message at the INFO level, a method on your component might look like

func (mt *MyType) doSomething() {
  mt.Log.LogInfof("An informational message")
}

This message will only actually be written to stdout/a log file if the log level configuration of the logger allows this (see below).

Context methods

Each of the convenience methods on logging.Log have a variant to allow a context.Context to be provided. Using these methods will allow you to include information held in the context in the meta-data printed along with each log line

Formatting messages

The f suffix on each method indicates you can use the same verbs pattern as used by Go’s fmt package.

For example:

  LogInfof("Running job %d of %d", current, total)

Stack traces

When recovering from a panic it is often useful to record the stack trace of function/method calls leading to the panic. The logger methods LogErrorfWithTrace and LogFatalfWithTrace generate a stack trace and append it to the message to be logged.

Checking to see if a message will be logged

Sometimes it is relatively expensive to construct a message to be logged. In these cases it is good practise to see if the message would be logged. The can be achieved with the IsLevelEnabled method on the logger.

For example:

if i.logger.IsLevelEnabled(logging.Debug) {
  i.logger.LogDebugf("Loading configuration from: ")
  
  for _, fileName := range configPaths {
    i.logger.LogDebugf(fileName)
  }
}

Next: Log levels

Prev: Logging principles