On Sun, Feb 8, 2015 at 4:04 PM, constantine <costas.magn...@gmail.com> wrote: > Thank you very much for your help. I do not have any recovery backup > and I need these data :(
Sorry to not be more sympathetic, but by definition data that isn't backed up is not important data. On top of it, the setup you have is fragile, not least of which is that you're using a relatively new filesystem and a 3.19rc kernel from December... This is for testing purposes, not production usage. So I'd say, you're doing it wrong and this is the sort of thing that can happen. But mainly the issue here is significantly misbehaving hardware well before the current drive failure and nothing was done about it sooner, i.e. Oh my god I need to make a backup now while I still can. > Before my problems begun I was running btrfs-scrub in a weekly basis > and I only got 17 uncorrectable errors for this array, concerning > files that I do not care of, so I ignored them. I clearly should not. The primary error is the lack of a backup while using supposedly important data. But 17 uncorrectable errors on a raid1 is a major warning sign something isn't right. A properly functioning raid1 volume doesn't have uncorrectable errors. > SCT Error Recovery Control command not supported These drives lack SCTERC support, which means they're consumer drives that aren't intended to be used in raid. > SCT Error Recovery Control: > Read: 70 (7.0 seconds) > Write: 70 (7.0 seconds) Two drives have appropriate SCTERC settings for raid. > # for letter in i d c g a b; do cat /sys/block/sd$letter/device/timeout; done > 30 > 30 > 30 > 30 > 30 > 30 The SCSI command timer is set incorrectly for the drives that do not have SCTERC support. So pretty much everything that could be done wrong was done wrong. The lack of SCTERC plus 30 second time out means there's a good chance that read failures aren't being reported by the drive to the kernel, so the kernel (Btrfs specifically) doesn't actually fix the problems; instead the command timer is reached, and the link to the device is reset instead. > The only way I can think of making a backup with my current available > hardware is removing my one WD Red 6TB from the array and copying > every file on this removed disk. > Can I remove the /dev/sdb without letting any of the data enter the > soon-to-fail /dev/sdc1? No. What I would do is make sure no further changes happen to this volume, consider it read only. Get the supposedly important information off the volume first. Then the next step is to *add* a good drive of any size; then delete sdc1. Only then can you correctly remove another drive. Even if Btrfs allowed it, I think it's a bad idea to try and remove drives from a degraded array. This could be possible if it weren't for sdc1 producing so many errors. Since it is, I don't see a way to do it, and if you try I think I'll make things much worse. If you try doing this, I'm not going to help anymore, put it that way. It's a big problem when a stranger cares more about your data than you do! Backup the data. That's first. Go buy a drive. There can be no possible excuses if this is truly important data. -- Chris Murphy -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html