Drive shows "invalid data" and shares are read only after hardware move.

[knalbone]

So here's the latest.  I moved all my drives and flash drive back to my old server. I let the data rebuild and all seemed well until I tried to write to the array. From windows I get access denied messages.  The syslog shows that Disk 4 is a read only file system. Can I somehow fix this with chmod? Should I try to rebuild by swapping disk 4 with another disk?  syslog attached.

syslog.zip

[trurl]

Check Disk Filesystems

[knalbone]

I found that article and thought it may be suggested.  Ran reiserfsck on md4 and it suggested --rebuilt-tree. I guess I'll give it a shot.

[knalbone]

One question.  What is going to happen when I restart the array after running reiserfsck --rebuild-tree?  Is parity going to rebuild based on the contents of disk 4?  Before running this, the contents of the disk looked fine as far as I could tell, they were just read only.  I'm just a bit concerned that there will be a lot of unintelligible files and parity will be rebuilt based on that.

 

Can I mount the disk to look at the files without starting the array?

[trurl]

One question.  What is going to happen when I restart the array after running reiserfsck --rebuild-tree?  Is parity going to rebuild based on the contents of disk 4?  Before running this, the contents of the disk looked fine as far as I could tell, they were just read only.  I'm just a bit concerned that there will be a lot of unintelligible files and parity will be rebuilt based on that.

 

Can I mount the disk to look at the files without starting the array?

First let me say that I have, fortunately, never had to use reiserfsck, so anything I say is based on reading forum threads, and you might be better of with another opinion.

 

When you use the md devices, as in the wiki, any changes made on the drive also update parity in realtime. It is also possible to use the sd device, which wouldn't update parity, but that is not usually recommended since it would invalidate parity.

 

I suppose it might be possible to do it on the sd device, and then if you weren't happy with the result, use parity to put it back like it was, but I don't think I have ever seen that recommended for some reason. Perhaps because it would be pointless since you would still have the file system problem.

 

Probably rebuild-tree isn't going to break anything that isn't already broken. I think what happens is you wind up with some files getting put into a lost+found folder, but those should be files that you couldn't have accessed anyway.

 

Wait and see what others say.

 

[knalbone]

Thanks. This is my first time with a reiserfsck so I guess I will just have to see what others may have to say.

 

I'm not sure what caused this when moving to my dell server, but it seems clear that it was the trigger.  I have purchased a new controller and cables to use if I ever have the guts to move back to that server.  The backplane is the only thing that isn't being changed.  Is there a good way that I can test all the connections on that backplane with some spare drives before moving my data drives over?

 

I honestly don't suspect the backplane is the issue as I have been running a ZFS VM on 4 of the 12 slots without issue for weeks. I do want to be sure it is ok before moving data disks back though.

[RobJ]

What he said.  Parity works at a lower level with the sectors themselves, and knows nothing about file systems, files, and folders.  So long as you use the md's and not the sd's, parity will be updated, along with the corrections that reiserfsck makes.  A file system is like a huge hidden file in itself, but consists of a B-tree with leaves storing the info about all of your files and folders, and making it searchable, retrievable.

 

If reiserfsck says you need the --rebuild-tree option, then it found corruption in the tree, in its branches and leaves.  If the damage wasn't visible, then that's good, means much of the tree should already be fine, with a better chance of a more complete recovery.

[knalbone]

Thanks RobJ, I'm crossing my fingers that everything will be ok.  Pass 0 has completed and I am currently on Pass 1.

[knalbone]

Thanks for all the help. The reiserfsck finished and there are only 2 files in lost+found.  Anything further I should do to ensure file integrity (parity check or otherwise).

 

In regards to my earlier questions.  Once I install a new controller and cables in the Dell server I was trying to move to, is there anything in particular I can do to test the drives/backplane before putting my data disks in.  I have another unraid key and some spare disks I could use for testing.  Just not sure what the best method would be.

[trurl]

Thanks for all the help. The reiserfsck finished and there are only 2 files in lost+found.  Anything further I should do to ensure file integrity (parity check or otherwise).

 

In regards to my earlier questions.  Once I install a new controller and cables in the Dell server I was trying to move to, is there anything in particular I can do to test the drives/backplane before putting my data disks in.  I have another unraid key and some spare disks I could use for testing.  Just not sure what the best method would be.

Run a preclear or some other disk test using each slot in the backplane. Maybe someone else will have some other ideas.

[dgaschk]

I have a set of cheap e-bay drives for testing all the slots together.

[knalbone]

I have a set of cheap e-bay drives for testing all the slots together.

 

Do you just run preclear on all the drives for your test?

[dgaschk]

I preclear in all slots for a new build and rebuild in each slot. Do some or all of this. With small disks it doesn't actually take that long.

[knalbone]

Thanks I will give that a shot. I've got a few small disks.  Shouldn't have too much trouble getting my hands on a few more.