From midnight UTC for about two hours we have seen our nodes getting out of sync from time to time, which caused delayed tip response on our mainnet backends. We are still investigating what was the root cause of this, but the working theory is that it was connected to the epoch switch with the hard fork.