A colleague has been using a 10TB RAID 0 array to edit a film. He suddenly realized it's 98% full. The array has started seriously acting up, but cooperates grudgingly if you shut down all processes that might index this array and interact only with terminal commands. Working on the theory that the 98% fullness was causing the drive slowness and issues, we copied around 300GB from the array to a separate drive and then removed those files from the RAID array. The array does NOT respond well to the rm commands, and the system logs suggest that one of the drives might be failing. The RAID array is still about 95% full.
Yes, I'm aware that 'RAID is not a backup' and I'm also aware that a single drive failure usually means disaster on RAID 0 because of the striping, but the failing drive is still functioning. I'm wondering specifically can the dd command be used to copy the ailing drive exactly to another SATA drive such that RAID might continue to function? I've mentioned a few times that we should consider wiping the RAID and restoring from a backup from December and my colleague breaks into a cold sweat and starts stammering about lost work. If we could copy & replace the one drive reporting trouble, this might allow us to compare the latest RAID file structure with the December backup possibly rescue any new changes.
I'm unaware of the particulars of RAID storage -- and I realize that disk formats use different block sizes and different drives may have different cylinder or sector counts. If drive cloning is a possible solution, any details about how to proceed, especially regarding dd usage, would be much appreciated. In particular, I want to avoid any interaction with the drive while it's being copied. I'm also wondering if on can dd from a spinning disk and then copy that data to an SSD? Does that work?
NOTE: I did not set this system up. Someone else set it up, using the enclosure as JBOD, and, rather than using the manufacturer's drivers and RAID manager, instead used software RAID.
