compacting virtual disk corrupts disk => fixed in SVN

[Quinn Stevenson]

HOST: Windows 7 64-bit GUEST: CentOS 5.5 64-bit

Steps I used to reproduce:

• Install CentOS in VM
• Allow CentOS to download updates
• Install zerofree in VM
• reboot to runlevel 1
• ran the following commands mount -n -o remount,ro -t ext3 /dev/mapper/VolGroup00-LogVol00 / zerofree /dev/mapper/VolGroup00-LogVol00 mount -n -o remount -t ext3 /dev/mapper/VolGroup00-LogVol00 / reboot
• After successful reboot, showdown the VM and run VBoxManage modifyhd --compact <virtualdisk.vdi>
• Now when I try and boot the VM the first time, the screen resizes for the X login, but it stays black and nothing is ever displayed. After resetting the virtual machine, I get a "Kernel panic - not syncing: Attempted to kill init!"

One other note - this process works just fine with the IDE virtual controller/disks. I've tested with IDE, SATA and SAS virtual disks, and the error occurs with both the SATA and SAS disk types.

I'll attach all the vbox.log files for the test run with a SAS virtual disk.

[Quinn Stevenson]

Most current log

[Quinn Stevenson]

Older Log

[Quinn Stevenson]

Even Older Log

[Quinn Stevenson]

Even Older Log

[Quinn Stevenson]

Oldest Log

[Quinn Stevenson]

VBox.log.2 and VBox.log.2.2 are dups - I upload the file twice by mistake, and I couldn't figure out how to delete it.

[Quinn Stevenson]

Also - I first encountered this problem in 3.2.6, and it is still there in 3.2.8

[aeichner]

We probably found the issue. Will be fixed in the next maintenance release. Thanks for the report!

[aeichner]

As a workaround try to enable the host I/O cache for the controller or attach the image to the IDE controller.

[Klaus Espenlaub]

BTW, if you try the workaround you'll see that the disk actually isn't corrupted at all.