I made a foolish mistake this morning before work.  I tried to rebuild a failed drive on my cache drive instead of the spare I had available.  To make matters worse I had copied a bunch of files to it (cache drive) that were going to go to another drive to consolidate directories that were split among different drives.  I stopped the rebuild before it got a long ways (couple of gigabytes) into it but it was enough to rebuild the directory structure of the failed drive at least enough that unRAID shows the same amount of free space anyway.  It is currently rebuilding on the correct drive now and I hope that will be done when I get home from work.

 

So what should I do first to try to restore the cache drive contents?  Would like to get some of it back if possible anyway.  Should I follow the instructions here?  That was for version 4.2 and a data drive.  Or would I be better with one of the other links here.

I'd say they are gone... The rebuild process will overwrite the entire file system with the calculated contents of your failed disk.

 

Sorry...

I'd say they are gone... The rebuild process will overwrite the entire file system with the calculated contents of your failed disk.

 

Sorry...

Since he stopped the process, the disk likely has a partially valid partition and file system, but very little if any valid files from the rebuild. The majority of space still contains his files that were on the disk originally, just no valid file system to point to them. I'm not really familiar with file system recovery on reiserfs (thank goodness) but if there is a way to scan the whole disk for valid files, ignoring both partition info and file system, he should be able to recover some things. If the file recovery process requires a valid partition table, that might need to be adjusted first, depending on whether it got changed by the unraid rebuild. It might still be ok.

 

The real question is, how much time is it worth to be investing in recovering and validating any recovered files, vs just recreating the files from other sources if possible. That is the $10,000 question that only the OP can answer.

I'd say they are gone... The rebuild process will overwrite the entire file system with the calculated contents of your failed disk.

 

Sorry...

Since he stopped the process, the disk likely has a partially valid partition and file system, but very little if any valid files from the rebuild. The majority of space still contains his files that were on the disk originally, just no valid file system to point to them. I'm not really familiar with file system recovery on reiserfs (thank goodness) but if there is a way to scan the whole disk for valid files, ignoring both partition info and file system, he should be able to recover some things. If the file recovery process requires a valid partition table, that might need to be adjusted first, depending on whether it got changed by the unraid rebuild. It might still be ok.

 

The real question is, how much time is it worth to be investing in recovering and validating any recovered files, vs just recreating the files from other sources if possible. That is the $10,000 question that only the OP can answer.

That's what I was figuring.  If I'd stopped it sooner I might have still had some of the directories available.  I was hoping there might be a copy of the directory structure elsewhere on the 2TB disk that I could access but I wasn't holding out much hope.  They were recorded videos from Cable and Satellite so might be able to get some of it back but some were from as far back as 2004 and rather obscure as in I have yet to see them repeated.  I might be able to get a list of the files from my SageTV server that recorded them.  I told unRAID not to export the user share so that SageTV thinks the directory is off line and not to delete the files from it's database when it can't find them.  I think I'll export the list to XML tonight and then re-export the share and let SageTV delete them from it's database so that I can get them recorded again if they repeat.  Then I will make an image of the drive and use the tools from the links I posted and see if I can get any files back and more importantly figure out what the file names and date and time of the files that might be recovered are.

 

Since the directories on the disk now are corrupted should I format the disk first to make an empty directory then try to recover files or will that make no difference or make it worse?  Any thoughts?

rieserfsck -scan-whole-partion will recover anything recoverable.

rieserfsck -scan-whole-partion will recover anything recoverable.

Thanks.  I'll try that once I'm able to image the drive and try it from the image.

I didn't know what I was doing with reiserfsck so wanted to ask how to proceed before attempting anything.  Turns out I didn't need --scan-whole-disk as --rebuild-tree alone was sufficient.  I thought that --scan-whole-disk was a "mode" not an "option" so it naturally didn't work when used alone.  So since I had made an image of the disk first and was applying this to the image I just tried --rebuild-tree and it worked.  It wasn't until after I started the --rebuild-tree that I found out it was an option to use along with --rebuild-tree or --fix-fixable (not sure which but didn't need it).  I managed to get back most of my files and only have 5 without names - yah!  When done it had free space of 299GB out of 2TB and based on what I remember copying it should have been between 150 and 200GB free when the copy was done.  So I only lost ~100GB of data - yah!

 

Thanks very much for the help.

Wow... Never realized the disk check was that effective.. That's a excellent one for the old memory bank.