Error Handling

A MessageFailure indicates an error handling an HTTP message. These include:

For Beginners

When you start from a "clean slate" with http4s, one of the things you're likely to notice, is that http4s is swallowing your exceptions - let's see if we can prove it. Assuming you've gotten the hello world example started, let's introduce another route, which is going to error out.

import cats.effect._
import cats.syntax.all._
import org.http4s._, org.http4s.dsl.io._, org.http4s.implicits._
import org.http4s.ember.server.EmberServerBuilder
import org.http4s.server.Router
import org.http4s.ember.client.EmberClientBuilder
import org.http4s.server.middleware.ErrorAction
import org.http4s.server.middleware.ErrorHandling
import cats.data.OptionT
import com.comcast.ip4s._

val errorRoute: HttpRoutes[IO] = HttpRoutes.of[IO] {
      case GET -> Root / "error" =>
        throw new Exception("Hey don't swallow me")
}

val server = EmberServerBuilder
  .default[IO]
  .withPort{port"8081"}
  .withHost(host"localhost")
  .withHttpApp(errorRoute.orNotFound)
  .build

Once you've started this server, hit up the "/error" route in browser. What you'll find, is that it returns a 500 response code... and no messages, and the console that hosts the http4s dev server is also showing you ... nothing.

Now, in general swallowing exceptions in software engineering is widely considered bad practise - now we have no idea our program is going wrong! Presumably in the case of http4s, having your public facing webserver "secure by default" trumps that consideration.

http4s provides an answer to this seeming paradox, in the form of middleware. As this section is written by a beginner, we're going to assume you want to know your code is throwing exceptions and take the beginners path of least resistance to surfacing them.

We're going to make the path to instantiating the server look like this, instead.

import org.http4s.server.middleware.ErrorAction
import org.http4s.server.middleware.ErrorHandling

def errorHandler(t: Throwable, msg: => String) : OptionT[IO, Unit] =
  OptionT.liftF(
      IO.println(msg) >>
      IO.println(t) >>
      IO(t.printStackTrace())
  )


val withErrorLogging = ErrorHandling.Recover.total(
  ErrorAction.log(
    errorRoute,
    messageFailureLogAction = errorHandler,
    serviceErrorLogAction = errorHandler
  )
)

val thisServerPrintsErrors = EmberServerBuilder
  .default[IO]
  .withPort(port"8081")
  .withHost(host"localhost")
  .withHttpApp(withErrorLogging.orNotFound)
  .build

And now, you'll get told which endpoint is failing, and get the stacktrace printed to the console. Leveling up your error handling experience through fancy logging frameworks, tracing et al is left as an excercise for the (no longer beginner) reader...

Logging

If a MessageFailure is not handled by your HTTP service, it reaches the backend in the form of a failed task, where it is transformed into an HTTP response. To guard against XSS attacks, care is taken in each of these renderings to not reflect information back from the request. Diagnostic information about syntax errors or missing fields, including a full stack trace, is logged to the org.http4s.server.message-failures category at DEBUG level.

Customizing Error Handling

TODO