My Ubuntu 12.04.1 running on virtualbox ran out of space while I was installing software on it and now it is not rebooting. I've got files I need to recover from it. I have the ubuntu64.vdi file on my Windows 7 (host) laptop and I was wondering if there is any way of recovering the files from the vdi file.

In parallel, I generated a fresh 12.04.1 machine on virtualbox in the same laptop, and I'll use that one instead of the borked one.

I can add the borked vdi to the list of disks in the working one. Then I used gnome nautilus to mount the borked disk.

I can find the files in there, but they are about 2 months old. I think this coincides with the last time that I did a virtualbox snapshot of the machine, but not the last time I got a working machine, which was yesterday. Is there any way of mounting the vdi as of yesterday, just before it crashed?


Here is what Virtual Media Manager tells me about my ubuntu64.vdi file: enter image description here


I copied my ubuntu.vdi to an external USB with plenty of disk space. I tried using Windows to clone it, but got an error:

enter image description here

I tried the same from the working VM via a shared folder of the USB drive, and it started the cloning. Took a good while, but finished successfully:

VBoxManage clonehd ubuntu64.vdi ubuntu64.raw --format RAW

Clone hard disk created in format 'RAW'. UUID: 90f99198-18e1-40ee-85fe-c083dcbb1ac3


After creating the raw file, I run testdisk with Deeper Search, I get a list like this: enter image description here

Ater Deeper Search has finished, I get a list like this:

enter image description here

But I don't know which one is which snapshot... Any ideas?

  • @Takkat, that sounds cool. How can I do that?
    – 719016
    Feb 24, 2014 at 12:25
  • @Takkat, I did it and mounted it using gnome nautilus, but I believe it mounted a two month old version of the machine.
    – 719016
    Feb 24, 2014 at 16:18
  • 2
    The problem here is not how to recover the disk but how to do it merged with the snapshot. If you try to clone your vdi it would have another uuid an the snapshot will not work with it and you would loose the latest data. I don't use snapshots in a regular basis because I have had corrupted disks in the past and I was able to recover the base disk but lost always the deltas. I hope someone here gives a good recipe to recover both the base and the delta. Feb 24, 2014 at 23:23
  • 2
    This is for next time: never store data on virtual machine, always mount an external drive... or make frequent backup/synch via rsync.
    – Dee
    Feb 27, 2014 at 17:59
  • 1
    Try to clone the most recent snapshot, not the original disk. You should use Vboxmanage clonehd 2208... (copy the most recent UUID here). See serverfault.com/a/506581/195528
    – Rmano
    Feb 28, 2014 at 19:24

5 Answers 5


Of the command line

In case we use Ubuntu as a host we can convert our .vdi disk to a raw disk with the following command:

VBoxManage clonehd <sourcename> <clone.raw> --format RAW

We can then open testdisk Install testdisk with this file to recover files on the virtual drive's image (also see below):

testdisk <clone.raw>

Of a virtual machine

In case the broken VDI drive image is still registered with Virtual Box we may be able to add this image as a drive to our new virtual machine:

enter image description here

Start your virtual Ubuntu to which we had added the broken .vdi but do not mount the drive.

  1. Create a directory on a mounted partition there to hold the harvested data (e.g. ~/rescued),

  2. Install testdisk Install testdisk on the virtual machine.

  3. Start testdisk from a terminal there:

    sudo testdisk
  4. Choose the appropriate drive to harvest data from (here shown for /sdb) and select [Proceed]

    enter image description here

  5. Choose the appropriate partition layout (usually Intel)

  6. Select [Analyse] - [Quick Search]

  7. Select the partition which contains the data to harvest (here only a single NTFS partition was present):

    enter image description here

  8. Press P to list files on that partiton.

  9. Browse to the directory containing the data. View the filenames to make sure you did choose the right ones.

  10. Select directories or files to harvest using : . All selected files/directories will be marked green:

    enter image description here

  11. Press upper case C to copy all selected directories/files to the rescue directory which will be selected in the next step.

  12. Choose a directory on your Ubuntu VM's mounted partition(s) to hold rescued data:

    enter image description here

  13. Press c to start copying the data from the broken .vdi to our rescue directory in your healthy VM.


  • In case we can not see our files with the [Quick Search] option in 6. we may have to perform a [Deeper Search]. This will also list previous deleted and overwritten partitions. So we need to make sure we select the appropriate partitions where our data reside.

  • If we had taken snapshots of the .vdi we may find these in ~/.VirtualBox/Machines/<Name of Machine>/Snapshots/{###}.vmdkbut whenever we had used the snapshot feature of Virtual Box any changes from a snapshot will be written to a differential file.

    To merge the differential file with the hard drive we need to delete all snapshots (which is probably not what we want as then we could just restore to previous snapshot rather than call for data recovery), or we need to clone the machine's drive using:

    VBoxManage clonehd <name_of_source> <clone>

    Then we bind the clone to our rescue VM as depicted above.

  • Writing to the broken .vdi should only be done after the data was safe as this may further damage the file system.

  • In case our rescue Ubuntu is too small to hold the rescued data we may create another new virtual drive to bind to the VM. This drive then has to be mounted in the VM to be able to save data from testdisk there.

  • thanks for these brilliantly detailed instructions. I followed them correctly and added the files to copy to my rescued folder. The copy was successful but the timestamp of the files was quite old: summer 2013 instead of two days ago. I can't remember if I did anything in particular 8 months ago, but at first glance, my files from two days ago weren't there. I am now using Deeper Search to browse through it again.
    – 719016
    Feb 25, 2014 at 9:36
  • @213441265152351 - also have a look at the snapshot folder... there may be a newer .vmdk (at least it was with my systems here). Hope you'll find your data :)
    – Takkat
    Feb 25, 2014 at 9:47
  • I did a Deeper Search, and after a couple of hours, browsed again the contents I wanted to recover: they are still pointing to a summer 2013 version of the virtual machine, earlier than just mounting the partition with gnome nautilus, which was giving me a january 2014 version, but still not able to find the files from two days ago.
    – 719016
    Feb 25, 2014 at 10:25
  • 1
    @213441265152351: well that's more complicated then. It may be that differencing images were not yet written appropriately to the drive when the "crash" occured. You can try to have a look at the disk from the Virtual Media Manager (see virtualbox.org/manual/ch05.html#diffimages) to see if that was the case, but I have to say that I will then not have a good idea on how to recover data from these images.
    – Takkat
    Feb 25, 2014 at 10:41
  • 2
    @Rmano: actually a cloned VDI will not include any snapshots but will be the last present state of the VM with differentials merged. Only in case we need to clone an earlier snapshot we may use a snapshot VDI to clone but then all changes later than the snapshot will not be included in the clone.
    – Takkat
    Feb 28, 2014 at 19:37

1) Try to recover the snapshot by the UUID, use following hack:

  • choose UUID of the state you want to recover.
  • Open file .xml, preferably in a tool (Internet Explorer works) that allows easy navigation of the XML file.
  • At the VERY BOTTOM, in the Storage Controllers section, find the UUID of the Current State and note the StorageController name=, type= and the AttachedDevice type=, port= and device= settings.
  • In the middle, find the snapshot you want.
  • In the Storage Controller section of the snapshot, find the AttachedDevice with the values (name/type/type/port/device) that matches the values retrieved above.
  • That image UUID identifies the Snapshot image for the hard disk. Remove the {} brackets, and

enter the command C:> VBoxManage clonehd <UUID> CloneDisk.vdi

Hope it helps, i did found the solution here: (https://forums.virtualbox.org/viewtopic.php?t=791)

2) if the steps in 1) does not work, heres another steps you can try. How to clonehd the Current State of a disk:

  • Open the Virtual Media Manager (Menu: File>Virtual Media Manager).
  • Navigate to the hard disk you want to clone and fully expand the tree (use * key on numeric keypad).
  • Find the {hexUUID}.vdi file that is attached to the VM you want to clone the Current State of. It will be (one of) the leaf files at the end of a branch. When you select it, the ‘Attached to:’ text at the bottom of the window will say: 'Attached to: ', without any (snapshot name) in brackets after it.
  • Click the filename to the right of the 'Location:' displayed at the bottom of the window.
  • Ctrl+C / Copy to copy the text to the clipboard, which will be something like: (path){b6441469-5ccb-418c-8fdd-73c5e1a17314}.vdi
  • Ctrl+V / Paste and extract the hexUUID value (removing the {} brackets) and run clonehd to clone it!

Thi solution is from here: (http://www.i-m-code.com/blog/blog/2012/02/04/virtualbox-clone-snapshots/) Both speaks mostly about the same.


Found my corrupt vdi file to be recoverable using 7-Zip , hope this simple method too helps someone.

  • It works for me!!! Thanks.
    – Domi.Zhang
    Jun 28, 2020 at 3:31
  • I have 7-Zip, but I could not find and fix feature. How did you do it?
    – AlexSC
    Mar 11, 2021 at 17:46

I had a Linux system get corrupted at / and fsck could not fix it. I used the following simple approach:

  1. Add a new replacement VDI to the machine.
  2. Download a new linux ISO.
  3. Boot and install a new linux on the disk you added in (1).
  4. Mount the disk on your new system and copy the files you need. Usually, this means copying the /home/user/ folder. Remember to copy the hidden files (settings files).

Worked well for me. The disk corruption was only hindering boot, there was no data loss in my home folder.


Launch the old virtual machine after inserting an Ubuntu ISO. Ubuntu will analyze the machine. Choose "Try ubuntu" (obviously don't install anything and don't erase any data). Once it's booted up (and it will because it's using the ubuntu iso not your old broken install), in the browser (Nautilus) choose "Other locations" to navigate to your old data and move them to somewhere safe.

I did not try to fix the machine, but I was only interested in recovering data and move them to the new virtual machine. It's probably possible to do what the top answer is suggesting directly from the live-booted old machine that way.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .