[Update as of 12th October 2018, 6PM PST - The hotfix containing the fixes for the issues causing database refresh request failures in the last 3 days has been rolled out to Production. As mentioned below, please trigger rollback if your environment is in a failed state and then resubmit a new request to refresh you sandbox with production data. ]
Starting on Monday, October 8th 2018, there were technical issues that caused some Database Refresh requests to fail. We have investigated the failures and are releasing a Lifecycle Services hotfix that will address all the known issues by Oct 12, 5 PM PST.
What you can do
If you have a Database Refresh request that failed for any reason that is outside of your control (anything that it is not data related), we have added a new Rollback button to your target environment. Please roll back your environment to the last good known state and then resubmit a new Database Refresh request.
We understand that timely refreshes of data is imperative for your own diagnostic, testing, and training needs. We'll continue to actively monitor issues. If you still see issues after the hotfix, please raise a support request. We will update this blog post once the hotfix is released.