• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1675
  • Last Modified:

Recover Data from FreeBSD lost+found

Over the weekend, I had an email server RAID5 array fail. After bringing the array back online and running fsck, the entire /usr now how has 1 directory, lost+found. Inside the lost+found are hundreds of directories, of which very few contain anything. df - h still reveals /usr as having 24GB used. However most of this data does not appear to be visible. I would really like to recover these emails for my customers, but I do not know where to go from here, assuming it is even possible. I would appreciate any direction or advice.

Sample directory listing of /usr/lost+found

#05274997       #05648578       #05956413       #06287585       #06929083
#05274998       #05648580       #05956416       #06287586       #06929084

All of the directories look like this except for these 3:

Makefile
patch-ad
pkg-plist
0
FusionNetworks
Asked:
FusionNetworks
  • 3
1 Solution
 
gheistCommented:
file * */*

If /usr did not contain mail spool you can reinstall -

csup all sources to RELENG_6_3
buildworld + installworld base system

record output of pkg_info

do pkg_delete -a and pkg_add -r packages one by one

That should more or less recover content of /usr

0
 
gheistCommented:
You can try installing a system aside on some working but soon-to-recommision workstation while reinstalling server with 7.0
0
 
FusionNetworksAuthor Commented:
Sorry gheist, I may not have been perfectly clear on what I needed help with. But I greatly appreciate your responses. I was trying to recover 24GB of email directories and message in /usr/local/virtual when the entire /usr slice was corrupted.

I was able to resolve this. Here is what I did.
1. Installed FreeBSD on a second drive array in the server. Which could then access the old drive array.
2. Copied the corrupted slice from the original drive array to a file on the new drive array using DD. (dd if=/dev/SOURCE of=/dev/DEST bs-8192)
3. Mounted the image file.
       mdconfig -a -t vnode -f /IMAGEFILE -u 0
       mount /dev/md0 /mnt
4. Moved the lost+found directory from the newly out of the newly mounted image file
5. Umounted the image file (umount /mnt)
6. ran fsck on the unmounted image file (fsck -t ufs -fy /IMAGEFILE)

This created a new lost+found directory. Inside of which, among other folders was a 23GB folder (#04945920) which appears to be my old /usr/local with all of the subdirectories listed. Specifically the virtual folder that contains all of my customers maildirs and emails. Upon closer inspection all maildirs and messages are intact. Woohoo!

0
 
gheistCommented:
I clearly did not get the essence of your request.
Glad you fixed it.

dd conv=noerr,notrunc does go over bad blocks if you ever run into problem again

Now you noticed softupdates in action - ./local was unlinked because of excess activity below it. Let me recommend using separate system for fast changing data... (like /var)

0

Featured Post

Concerto's Cloud Advisory Services

Want to avoid the missteps to gaining all the benefits of the cloud? Learn more about the different assessment options from our Cloud Advisory team.

  • 3
Tackle projects and never again get stuck behind a technical roadblock.
Join Now