I use 2 RAID0s in my system, backed up by a RAID5. On Friday evening, one disk in each of my RAID0s started to go bad with Unrecoverable Read Errors. The pending failures occurred within just a couple of hours of each other. What are the odds against that happening, eh? In addition to that, one of the 3 disks in my RAID5 started triggering "benign" Unexpected Sense Errors at the same time. At this point I was thinking that maybe I should go play the Lottery, since my chances of winning the big jackpot were about the same as my chances of having 3 disks develop problems almost simultaneously. The errors on the disks in the RAID0s kicked both arrays offline, and upon reboot the drives showed as bad in my controller's software Console.
My critical stuff was safe on the RAID5, but there were a few files here and there that I'd kept on the RAID0s that were nice to have and that were not part of my regular backups. Well, I found that all was not lost for those files.
The Console has an option to let me manually change a drive's status from Unconfigured Bad to Unconfigured Good, which I did for each bad drive. Then I told the controller to scan for foreign configurations, and it found the previously working configuration. I imported the old configuration, rebooted, and both RAID0s came back on line with a surprising amount of data still intact! So if you ever have a RAID0 go pear-shaped, and you have to get critical data off of the failed array, forcing the bad drive back online and making it part of a saved configuration may possibly let you retrieve that data.
I don't know if this information is obvious to our hardware gurus, but I haven't seen it discussed before and it was eye-opening for me. I've always thought that once a RAID0 fails, all data is lost forever. At least in this situation, I can change that to *maybe* all data is lost forever. I also don't know the chances of losing the RAID5 as well as the RAID0s by importing the old configuration. Had that happened, I would have only been able to recover my in-work projects, which I keep backed up to an external drive as well.
From a hardware perspective, I have a 1000-watt power supply and a 1200-watt UPS, so insufficient power isn't a likely candidate here. I am considering the possibility, however, that the controller itself is bad and should be replaced. Regardless, this has turned out better than I expected all around.