Error handling

As mentioned earlier, each application service in a distributed application is dependent on other services. As developers, we should always expect the worst and have appropriate error handling in place. One of the most important best practices is to fail fast. Code the service in such a way that unrecoverable errors are discovered as early as possible and, if such an error is detected, have the service fail immediately. But don't forget to log meaningful information to STDERR or STDOUT, which can be used by developers or system operators later to track malfunctions of the system. Also, return a helpful error to the caller, indicating as precisely as possible why the call failed.

One sample of fail fast is to always check the input values provided by the caller. Are the values in the expected ranges and complete? If not, then do not try to continue processing, but immediately abort the operation.