Update. Once the RAID rebuilds we are going to run a job to delete
old workunits and results from the array. These would normally be deleted
by the file deleter but the latter is still backed up from before the database
migration. The project will remain off through all of this.
The data server failed around 20:30 UTC for the third time in as many days.
The project will be down as we continue to diagnose the problem.