I have 7 disk array with a parity. Each are 2tb. One disk has a corrupt partition and is even preventing me from copying certain files off of it. I want to get a 4tb into the parity and use that 2tb to replace the other disk. I read about a swap-disable, where you put the 4tb into parity and then can use that to replace the other disk. Would this work in this circumstance? If it has to rebuild parity will it rebuilt it with the corrupt information if I do this?  Does the "corrupted" disk need to be red balled for this to work?

 

According to the wiki looks like it works

For swap-disable, you use your existing parity disk to replace the failed disk, and you install your new big disk as the parity disk:

Stop the array.

Power down the unit.

Replace the parity hard disk with a new bigger one.

Replace the failed hard disk with you old parity disk.

Power up the unit.

Start the array.

When you start the array, the system will first copy the parity information to the new parity disk, and then reconstruct the contents of the failed disk.

The wiki describes it correctly => ASSUMING you have good parity (have you done regular checks??) then it will work just fine ...  it will, as you've noted, copy the old parity disk to the new parity drive; then clear the rest of the new parity drive;  and then do a rebuild of the failed drive onto the old parity drive.

 

Your system is, of course, "running at risk" until the process completes ... i.e. any other disk failures during that time will result in lost data.

 

It is not clear if the problem disk is actually giving read/write errors or if it just contains corrupt information at the filesystem level?  It might be worth taking and posting a smartctl report for the disk in question to see if it has any sectors pending re-allocation.

 

If the disk contains corrupt information, then I do not think the rebuild process will correct this, and it will be better to try to recover the data before swapping out any drives

 

You may want to first consider putting the array into Maintenance mode, and then run

reiserfsck --fix-fixable /dev/disk??

where ?? corresponds to the disk in question. That will check the validity of the file system and correct minor errors. 

 

If that command completes without reporting a hardware issue then there is a good chance the disk is physically OK.  The reisrefsck will also tell you if the disk file system appears corrupt, and it needs to be rerun with additional options such as --rebuild-tree to recover the filesystem.  If you are going to do this more drastic level of recovery you may want to consider first taking an image of the disk in case it fails so that there is still a chance of recovering data.

  Does the "corrupted" disk need to be red balled for this to work?

YES, it will ONLY work if the disk to be replaced is already failed/disabled ("red" indicator)

According to the wiki looks like it works

For swap-disable, you use your existing parity disk to replace the failed disk, and you install your new big disk as the parity disk:

Stop the array.

Power down the unit.

Replace the parity hard disk with a new bigger one.

Replace the failed hard disk with you old parity disk.

Power up the unit.

Start the array.

When you start the array, the system will first copy the parity information to the new parity disk, and then reconstruct the contents of the failed disk.

It is not clear if the problem disk is actually giving read/write errors or if it just contains corrupt information at the filesystem level?  It might be worth taking and posting a smartctl report for the disk in question to see if it has any sectors pending re-allocation.

 

If the disk contains corrupt information, then I do not think the rebuild process will correct this, and it will be better to try to recover the data before swapping out any drives

 

You may want to first consider putting the array into Maintenance mode, and then run

reiserfsck --fix-fixable /dev/disk??

where ?? corresponds to the disk in question. That will check the validity of the file system and correct minor errors. 

 

If that command completes without reporting a hardware issue then there is a good chance the disk is physically OK.  The reisrefsck will also tell you if the disk file system appears corrupt, and it needs to be rerun with additional options such as --rebuild-tree to recover the filesystem.  If you are going to do this more drastic level of recovery you may want to consider first taking an image of the disk in case it fails so that there is still a chance of recovering data.

You should NOT use reiserfsck on the /dev/disk??  (there is no such device)

  It should be used on the /dev/mdX device.  Instructions are in the wiki under "check disk file-systems"

You should NOT use reiserfsck on the /dev/disk??  (there is no such device)

  It should be used on the /dev/mdX device.  Instructions are in the wiki under "check disk file-systems"

You are quite correct - I must have a temporary brain failure   I have done the correct thing enough times I should know it by now!  Since the /dev/disk?? device should not exist at least the OP would not go wrong.

 

Just as a check - was I correct in that rebuilding a disk does not correct file system errors and that reiserfsck is the way to go?

Rebuilding the disk partition will not correct it's contents (i.e., the file system). reiserfsck will fix the file system.

I already ran the reiserchk and it says you need to run it with the rebuild tree for that particular corrupt sector disk. I am afraid if i do this itll put all the data into a lost folder without the current structure which is why I wanted to do this parity swap so I can get the drive out. It was red balled back then but is currently green even though you can tell the corruption is their. If I swapped out the corrupted disk and recovered data the corruption should then be gone right or is it in my whole filesystem now and I have to use the rebuild tree on that particular disk? I dont have an extra 2tb around which is why I wanted to do this parity swap and put a 4tb in parity and use that 2tb current parity to replace the disk with corruption. All other disks checked fine.  If everything would work according to plan is their a way to force that disk to be red? Or if I have to do a reiserchk tree one should I copy what data I can do another disk in case it puts it into a lost folder least I would have most everything?

The rebuilt disk with have exactly the same contents of the current disk, including file system errors. Again, rebuilding will not correct file system issues. Only reiserfsck can correct the file system.

Okay thanks. I am going to copy stuff off to a 1.5tb I have in my desktop and which ever files doesnt copy off I hope I can get back with the tree build cmd.

 

The rebuilt disk with have exactly the same contents of the current disk, including file system errors. Again, rebuilding will not correct file system issues. Only reiserfsck can correct the file system.

which ever files doesnt copy off I hope I can get back with the tree build cmd.

Repeat after me. Unraid alone is NOT a backup.

which ever files doesnt copy off I hope I can get back with the tree build cmd.

Repeat after me. Unraid alone is NOT a backup.

 

+1  When something goes wrong, the "fix" is to repair what's wrong with the array (replace a failed disk, cable, controller, etc.) and then simply copy the data from your backups 

 

NO RAID array is a backup -- RAID is designed to provide fault-tolerance and/or extra performance.

Ya I know, kinda hard to backup 16tb though less I made a complete second server. Data not worth that much to me.  So I make do with what it allows. Someday I am hoping for 2 drive parity support from unraid.

Ya I know, kinda hard to backup 16tb though less I made a complete second server. Data not worth that much to me.  So I make do with what it allows. Someday I am hoping for 2 drive parity support from unraid.

 

Support for dual-drive failures is STILL not a backup.

 

... A simple external eSATA caddy and a few extra disks is a fairly simple way to backup your data.  With modern disk sizes it's not difficult at all.    With only 16TB you could do it with only 4 4TB drives.    I have over 40TB in my 2 UnRAID arrays ... and it's certainly ALL backed up.    None of it is really all that important ... but it took a LOT of time and effort to produce, and I certainly wouldn't want to ever do it again !!    A couple thousand dollars of "insurance" (i.e. backups) is certainly worthwhile when you consider how much I've spent on the systems and the data.