This is noisy especially for day long proving periods unlike testing. But it is transient and stops erroring once we’re through a proving period

This is a curio usability issue and not a PDP protocol issue filoz is going to fix.

My guess would be we are consulting the wrong on chain method in the case that we are restarting at a weird time. We are probably trying to get setup for the next challenge window when we are less than challenge finality epochs away from it.

In such a case we really should be trying until we get through, we can’t stop. I suppose we could go directly to the next challenge window