This fault appears to have the exact same cause as the same issue 3 days ago, and the fault has come from the same server. After running well for a few days, it has repeated the same issue.
We will conduct a deeper investigation and decide if we need to permanently take this server out of circulation.
A minority of races from around 17:00 GMT were impacted but it was caught quickly this time and resolved before the next batch of races at 17:30. Our apologies to anyone impacted by this, such technical problems involving networks and hardware happen sometimes and we can't always plan for them.