curious - the backlog looked normal, and rates never get close to the limit (see weekly charts) - th

curious - the backlog looked normal, and rates never get close to the limit (see weekly charts) - this is a constant process that's been running for over a week now with little variation.
Queue send failed: Queue is overloaded. Please back off.
was received from workers all over the world at around the same time (calling
.send
on a queue).

Almost feels like an error message that would come from a DO request queue overload. Would it be possible for your team to track down what causes this message? I don't see any clues in workerd. Perhaps there can be an internal backup in whatever internal queue magic sits behind
.send
and the queue storage?

Ideally from an abstraction pov,
.send
would only fail when exceeding the msg/s limits (globally) or some truly edge case node failure where it couldn't even save to any sort of recoverable internal storage.
image.png
image.png
image.png
Was this page helpful?