April 30, 201412 yr A brief rundown up till now: - found a disk was red balled with ridiculous amounts of read/write errors (numbers were 10+ digits long) - ran SMART and file system checks, and and scanned for bad sectors. All came back clean. - unassigned red balled drive in process to reinstate drive. - Unraid unassigned ALL drives - reassigned drives in correct locations - started array with 'trust parity' just so it wouldn't rewrite the parity drive in case I needed it intact - so far everything seems to be working fine - as a note nothing has been written to the array since back before the drive red balled Now, I'm running a parity check with nocorrect. Within the first 30 seconds I got 68 sync errors. I'm about halfway through a 4TB check now and that number remains unchanged. As far as I understand it 1 sync error would mean 1 block and 1 block is 4K big, so that means a max of 272K data is 'bad' or at least untrusted since the parity doesn't match up which appears pretty miniscule. Anyway I can find out if the files are bad or if the parity was just off? Also any way I can find out which files these errors occurred on? I know the sectors and I know the drive so I just need to find out what files reside there. Any other advice on how to proceed from here? Thanks! syslog_04-30-14.zip
May 1, 201412 yr Stop the array and un-assign the disk. Start the array and examine the contents of the disk. These contents should have less corruption...
May 1, 201412 yr Author Not sure I'm following you. Won't removing that disk then restarting the array require a parity rebuild? Also what am I looking for once I get the drive isolated. I've already run it through a few tests and no errors have been found. Do you have something else in mind?
May 1, 201412 yr The disk was disabled because a write to it failed. Starting the array with disk missing will not required any subsequent rebuild. It will allow access to the version of the data disk that is reflected in parity. There are currently 2 versions of the disk distinguished by the parity errors. Which version would you prefer to keep? The emulated version is probably the correct one.
May 1, 201412 yr Author OK That makes sense that the data defined by the parity is the correct version. My parity check never found any more errors... just the 68 (blocks?) it found right off the bat. Any way to rebuild just those 68 blocks or do i have to rebuild the entire 3TB?
May 5, 201412 yr Author Trying to get unraid to the point of rebiulding the suspected drive but no having much luck. Following the wiki I unassign the drive and reboot, When it reboots the drive is back in it's assigned spot.
May 5, 201412 yr Author I tried shutting down an unplugging the disk to try to get the one disk to unassign. Unfortunately I unplugged the wrong disk - could have sworn I installed them in order! Anyway so now it wants to rebuild the wrong drive. I thought I could just start the array and say parity is valid with all disks installed (since it still should be) but I can't find that option anywhere. How do I force it to start?
May 5, 201412 yr Author OK got it doing what it should be. I copied over the config directory with a backup to get the array status back to normal. then got the array to unassign the suspect disk and it is now rebuilding that drive. I copied all the contents of that drive before this rebuild so I could compare later, I'll be curious to see what files were corrupt.
Archived
This topic is now archived and is closed to further replies.