Retry Pattern: An Example

A retry policy answers three questions, and each one can be got wrong on its own.

1. Is this failure worth retrying? A timeout might pass next time. A rejected request will be rejected again just as fast.
2. How long should we wait? Long enough for the fault to clear, and not so long that the caller gives up first.
3. When do we stop? Every policy needs a limit, or a broken dependency is retried until something else breaks.

The code below runs a policy against a service that answers from a rule. The same run comes out the same way every time.