EXCEPTION HANDLING · Workflow idea
When an operation fails, check whether the issue is a known exception with its own appropriate next step. Route recognized cases separately, retry unexpected failures within a defined limit, and make unresolved problems visible to an owner instead of letting a repeated failure continue unnoticed.
01 / The starting point
An unsuccessful operation that needs either a known-exception path or a limited retry
“Recognize expected exceptions, retry other failures only a bounded number of times, and surface any remaining issue for a person to resolve.”
02 / The work
Helps teams make routine failure recovery more deliberate and keep expected exceptions from being mistaken for transient problems.
Capture enough information to distinguish an expected exception from an unfamiliar or temporary problem.
Send a known case to its intended alternate handling path rather than repeatedly attempting the same operation.
Give other failures only the approved number of additional attempts and stop when that limit is reached.
Show the original issue, attempts made, and affected work to an owner who can decide what to do next.
03 / The result
A resolved operation, a separately handled known exception, or a visible unresolved failure awaiting review.
Operations · Example outcome
A resolved operation, a separately handled known exception, or a visible unresolved failure awaiting review.
Repeated attempts can duplicate an action, increase load, or conceal a permanent problem. Identify known exceptions precisely, set a safe retry limit, avoid repeating actions that are not safe to repeat, and require an owner to review failures that remain unresolved.
04 / Make it yours
A few choices shape the work around your team.
Define which errors are expected and what safe, separate handling should happen for each one.
Choose a maximum number of attempts and verify which operations can safely be repeated.
Name who receives issues that exceed the limit and how the team records their resolution.
Confirm recognized cases take their alternate route, temporary failures stop after recovery, and persistent ones become visible.