Hello!
SHORT VERSION
Cannot defrag my RAID 5 rig - OS freezes when it hits VMDK file. Read long version for details.
DON'T TELL ME TO
Before you say "turn off delayed writing" via device properties - ain't gonna happen, that section is greyed out, and YES I get plenty of "delayed write failure" messages if I try and do much with the VM.
QUESTIONS (Pick 1 or more...just want to get this diagnosed and fixed!)
* Will Diskeeper Server or Enterprise deal with this better (I own 1 license of server 2008 - not installed on this system)? (Defrag is the "cheap version", per Microsoft)
* I have seen some posts for VMWare to use CLI to mount and defrag their VMDK... would that help? If I do that will the RAID not lock up?
* nVidias tools are pathetic - I have tried to install/run 6.7/9.18 tools and I cannot get a GUI to run on the server so I cannot manage the RAID array except during a boot operation. Is there a trick to get these to run?
* Should I just get a 3Ware SataII PCI-E or PCI-X card and use their stuff instead? Maybe the controller on the MB is bad?
* Are there tests I should be running on the media? Or diagnostics on the RAID to determine if it is OK?
SYSTEM
Tyan GT24, Dual Opteron 2212, 8GB RAM (only 4 addressed, thanks nVidia for your crummy RAID support), 3x 500GB Seagate Barracuda 7200.10 (SATA), nVidia NFP3600 Chipset.
THE LONG VERSION
In August 2007, I built a host and a guest using a server I had built. At the time, I knew nothing about RAID, and this server sported the nVidia NFP3600 chipset which I found had no decent Linux support, no decent Windows support, and the best I could do was use infrarecorder to slipstream the 26MB worth of drivers into a Windows Server 2003 install CD.
The server has worked OK with VMWare (albeit Disk I/O has been definitively sluggish)...and over time the performance started to degrade. VMWare logs started noting read(10) and write(10) taking 1, 5, 10, 20 seconds to complete. It culminated in the BSOD on the VM, but the host never blue-screened, so I suspected something was corrupt on the VM.
Further research lead me to check into defragmentation (bing! the brain-lightbulb hit 5 Watts!) - and I found that the VM was 85% fragmented, but would BSOD before I could complete a defrag. The host OS also reported 85% file fragmentation, but the stats were skewed because it was mad the VMDK had a whole 3 fragments?! That should not be a big deal.
But if I fire up defrag on the host, it reports 3%, hits the VMDK file and the OS says things are running, but TaskManager stops responding, Defrag cannot be stopped, no other apps can receive focus, and the lights on the HDDs are out (they have power, but zero activity). So that tells me (another 2 Watts for my light bulb, please) uh, ain't much defragging going on!
This happens even if I stop all the VMWare services.
Start Free Trial
