After a member of a RAID stripe set dropped out on the system that received the subject message (apparently running 100% across all available cores and hyperthreads isn't such a great idea when running non-mirrored RAID stripe sets, especially on a Microsoft "Patch Tuesday"), a reboot put the disk back on-line. In the name of caution, I then ran a verify cycle in Windows with Intel's Rapid Storage Technology management tool (no verify error, but media block errors), dropped the member out of RAID again and ran disk scans with Western Digital's Data Lifeguard without ever reentering Windows (no errors - edit: twice, once in AHCI and once in IDE mode), and then re-added it to the RAID set and upon entering Windows reran another Intel RST verify (no errors of any kind).

So I told BOINC to run, and BOOM! - dumped 26 results with "Computational Errors" back to the server. (Lesson learned: In the future, "Reset Project".) And everything BOINC subsequently downloaded went BOOM!, too. So I uninstalled BOINC, formatted the drive (the hard way, not "quick"...no errors generated), copied BOINC data directory from a known good backup to keep the computer ID synched, reinstalled BOINC, started it, it downloaded 20 tasks, and...displayed the subject message.

Which brings me to my question: Seven of those 20 tasks in queue are running (and all 7 at the same percentage completed...synchronized I/O, which is good way to find out if that RAID set is going to dork up again, I suppose). If I permit those tasks to complete "in spite of" the "daily quota" message, will they be accepted?

Or should I abort them all? And perhaps drop the # of simultaneously CEP2 tasks for that system to two for "awhile" and run some other project on the idle cores?