With a nod to Edsger Dijkstra and aplogies to Eric A Meyer, auto retries and restarts are shiny rings that lead to pain and suffering!
Auto retries are frequently implemented as work-arounds for processes or requests that fail intermittently. Auto retries address the symptom rather than the cause. They chew up system capacity and hide problems. With every automatic retry you implement, you are adding invisible anchors to your system’s performance.
Resist the temptation to implement auto retries and auto restarts. Identify and address the root cause.
If you feel you *must* implement an auto retry, establish a performance counter that automatically alerts the right team when the auto retry count spikes or trends above normal levels.
-Donny
