Fedora core 7/8 filesystem corruption

Hello all,

I've come out of lurking to file a bug report. I've got 2 VMs with the same issues: their filesystems seem to mysteriously corrupt themselves after about a week of use.

I'm running Parallels 5584 on a C2D Macbook Pro (4Gb 256mb Vram model). The affected VMs are a Fedora 7 VM and a Fedora 8 VM. Both VMs have all development packages installed in addition to the base install. Both VMs are using Ext3 as their filesystem.

VMs are set as follows:
20 GB expanding HDD
128mb Partition sda1: /boot
2gb partition sda2: swap
8 gb partition sda3: /
rest of drive: /home
500mb RAM (not 512 to avoid old kernel bug that is probably fixed now anyway)
32mb VRAM
USB and sound are enabled
shared networking
Parallels tools are installed (after updates have been run, and I dont install new updates after installing the tools)

My use case includes using svn/gcc/emacs under KDE. I suspend the VM and close Parallels before suspending the Mac, and I have not seen any crashes of Parallels or the Mac itself. I use this VM for work only, so basically nothing else is run.

I leave the VM running (i.e. only suspend not reboot) and after about a week I start to see major corruption (I/O error reading file, empty file after write, disappearing file) after about a week. When I run fsck on the VM it fixes the issue for about another week.

I've reinstalled both VMs (delete VM, .hdd files everything -> create new vm with newly downloaded install ISOs) several times and this is still happening.

So: Is this a known bug that I didnt hear about or something new?

I can not send these VMs to the Parallels team (unfortunately) due to it being full of company sourcecode.

EDIT: I forgot to mention that this corruption is not happening to the Windows VM that I also use for work. One Windows VM and one of the two Linux VM's are nearly always running during my work day.

Also, I did not notice this corruption happening until after upgrading to OSX 10.5.2. I tried running the Combo updater manually in case it was a bugged OS update but that did not help.