High failure rates for 26-27 December

We had an issue with worker machines starting in a misconfigured state that affected the processing of quite a few reports over the days just after Christmas.

We'll be re-running those days for the affected endpoints, and will address the underlying issue so that it doesn't happen again. (Or at least chooses to happen on a nice rainy weekday when I'm not on vacation.)

Sorry for the inconvenience.