Between 16:15 and 16:40 UTC some machine executor jobs may have been run twice. During that time some jobs may have unexpectedly failed. Customers would not have been charged for any repeated job. We have already resolved the underlying issue and all machine jobs are running as intended.