The team investigated this and this is due to a failure on the server-side. We had an instance fail unexpectedly and this Egress job was running on it at the time. We will look into mitigating this better in the future.