Error return value of `app.errorLog.Output` is not checked

I'm pretty new to golang so sorry for my question.
I have the following function:

func (app *application) serverError(w http.ResponseWriter, err error) {
  trace := fmt.Sprintf("%s\n%s", err.Error(), debug.Stack())
  app.errorLog.Output(2, trace)

  http.Error(w, http.StatusText(http.StatusInternalServerError), http.StatusInternalServerError)
}

I'm using golangci-lint linter, and when I launch the linter, it returns the following error:

cmd/web/helpers.go:15:22: Error return value of `app.errorLog.Output` is not checked (errcheck)
  app.errorLog.Output(2, trace)
                     ^

How can I fix that?

In go, a common pattern is to have a function that returns two values, the first of which is the desired result, and the second of which is type error.

Typically, if an implementation cannot provide a value because of some error e, it will return nil, e or return <zero value for type>, e.

It doesn't just have to be one such desired result value, however - sometimes there will be zero or more than one desired result values. Basic rule: if the last return value of a function is error typed, and the docs don't say otherwise, always check that last return value.

So, when you see such a pattern in a return signature - you should not discard the last result with a ignored return or an assign to '_', but should check that value to make sure it is non-nil before continuing.

The first of those anti-patterns is what the linter is warning you about. You can check the error argument thusly (I'm assuming that there are zero "desired result" values here):

 if err := app.errorLog.Output(2, trace); err != nil {
     // ... do something to handle error here or panic(err)
 }

This will satisfy the linter and make your code more robust!