P2V trouble WIN2K Server

slinuxuzer

Ok, I am lost here. This is what I did

P2v win2k machine into my Vmware Vsphere 4 environment, running two esx servers and one V-center server. running on Dl380 G6 quad cores with hyper threading. 48 GB of mem.

NFS Storage to Netapp appliance, Netapp currently on different network (waiting on new one) and my storage connctions are not in a private Vlan currently

Guest runs SQL, Stopped SQL Services and ran P2V.

Problem - Guest machine processor is running at 100% and I can't get Vmware tools to install, Guest also try's to pickup the host hardware I.E. Install drivers for the actual NIC card instead of Vmware card and some hardware like the video driver wasn't auto-detected

Guest has 1GB of mem and one virtual processor

Any ideas?

astorrs

Did you modify the number of CPUs between the physical/virtual copies of the server? (sounds like it could be a uniprocessor/multiprocessor thing on Win2k).

I'm guessing it's detecting the NICs as Intel Pro/1000 cards - that's not the host hardware but the default NIC VMware presents to guests in vSphere. Without the VMware Tools installed you're missing the VGA drivers, etc on such an old O/S.

slinuxuzer

No, its a single processor on both. I thought that something flakey might have happened so I tried a cold migration and the cold migration failed.

Not sure why, this hardware is very old, is there anything extra I need to do to the cold migartion disk to make sure it works?

astorrs

Install VMware Tools on it before cloning it.

slinuxuzer

astorrs wrote: »

Install VMware Tools on it before cloning it.

I was thinking that. I will try it.

I spoke with Vmware support today and they seem to think it's an issue because of my SNMP monitoring client. So I decided to try a cold migration

Problem with this is, I thought I didn't have the correct drivers. turns out the network .sys file and my CPQarray.sys file are already on the disk.

My cold migration is stopping at 2%

Question 1. would it get this far if I didn't have the correct drivers?

Question 2. Every time I edit the .ISO with Petools I get a txtsetup.oem wasn't found from my server and any other machine I try to boot with this disk. Is there a better way to edit this?

Sunshine_54

Hard to say why it failed at 2 percent without the logs. I doubt its a driver issue. If it was a driver issue it would not detect your Windows installation and/or your disk/network adapter. Usually failures at 2 percent are caused by networking. Check speed/duplex settings.

slinuxuzer

Prolly should have mentioned, the machine I am P2V'ING is a really old server 7+ years, an internal resource here tells me what there are some settings he thinks need to be changed around in order to get this VM to work with the new Gen processors, but he can't remember where the documentation is, we are both looking for it.

Does this ring a bell to anyone?

slinuxuzer

[2010-06-04 18:00:08.828 'App' 168 info] [stub,42] error [Windisk_CheckNonFixedMediaDevice] Failed to retrieve media type for \\?\Volume{55fbdb3f-7002-11df-b03f-806e6f6e6963}\: Incorrect function (1)
[#1] [2010-06-04 18:00:08.828 'App' 564 verbose] [attachedComputerWin32,678] [GetVolumesForComputer] Checking if volume \\?\Volume{55fbdb3f-7002-11df-b03f-

[#1] [2010-06-04 18:00:20.843 'App' 564 error] [NetworkAdaptersInfoEnum,945] Failed to intialize locator to WMI. Err code = -2147221164
[#1] [2010-06-04 18:00:20.843 'App' 564 error] [NetworkAdaptersInfoEnum,196] Failed to initialize WMI
[#1] [2010-06-04 18:00:20.843 'App' 564 verbose] [NetworkAdaptersInfoEnum,155] Failed. Use MIB Interface way to get the list of Network Adapters
[#1] [2010-06-04 18:00:20.843 'App' 564 warning] [NetworkAdaptersInfoEnum,156] Virtual NICs will not be filtered out
[#1] [2010-06-04 18:00:22.031 'App' 564 verbose] [NetworkAdaptersInfoEnum,361] Detected 2 NICs
[#1] [2010-06-04 18:00:22.031 'App' 564 verbose] [NetworkAdaptersInfoEnum,374] NIC1 has description "Compaq NC3163 Fast Ethernet NIC"
[#1] [2010-06-04 18:00:22.031 'App' 564 verbose] [NetworkAdaptersInfoEnum,396]

astorrs

Is there a hidden utilities volume on the physical server (was common with older Compaq's especially)? skip it if you are p2v'ing.

slinuxuzer

Well I made some progress on this, I now have the cold migration moving past 2% and at 20% as of 30 minutes ago.

To get this moving forward I did three things, not sure which helped but all probably needed to be done anyway.

My Vmware resource here told me I should go ahead and tell it to move the pagefile and hibernation file (no hibernation file as this is Win2k), also moved the network card on this server to a different port putting it in the same VLan as the V-center server, also unchecked the small .4 GB utility partition that was present on this older compaq server.

So I will see how the newly cold migrated VM guest behaves and post and update soon.

slinuxuzer

Still the same issue, the processor stays pegged out at 100%, it has way more resources as a VM than it ever did as a physical box. The physical box was quite a bit older than the new ESX host, prolly 7+ years old. I was told there might be some processor settings that I need to look at, can anyone give me a clue?

astorrs

Assuming you've already got VMTools installed, take a snapshot of the running VM, then start by cleaning up the old hardware profiles/drivers.

From a command prompt (not the Run box) type the following:

set devmgr_show_nonpresent_devices=1

then

devmgmt.msc

This will launch Device Manager under the View menu Show Hidden Devices and you should now see all the non-present hardware (it appears grayed out). Right-click on each one an Uninstall it. Also look for drivers such as "Compaq iLO Remote Management Card", "HP Physical Card That No Longer Exists", etc. Once you've removed everything reboot again and watch the system event log for errors (you can always GoTo the previous snapshot if you bugger things up and start again - quick warning, in the vSphere Client GUI when dealing with snapshots, Delete != Revert, it actually equals Commit).

Once things are cleaned up and looking healthy let us know what the CPU performance profile looks like.

MentholMoose

What is using the CPU? Check with Process Explorer, procexp.exe here:
live.sysinternals.com

kriscamaro68

slinuxuzer wrote: »

Still the same issue, the processor stays pegged out at 100%, it has way more resources as a VM than it ever did as a physical box. The physical box was quite a bit older than the new ESX host, prolly 7+ years old. I was told there might be some processor settings that I need to look at, can anyone give me a clue?

Have you tried disabling hyper-threading? If not I would try that. I know there use to be issues with ht and older server os's.

slinuxuzer

kriscamaro68 wrote: »

Have you tried disabling hyper-threading? If not I would try that. I know there use to be issues with ht and older server os's.

At what level are you referring to turning hyper threading off at? I know It can be turned off in the BIOS, but if at all possible I would like to disable it for this Guest only and not for the entire ESX host.

kriscamaro68

slinuxuzer wrote: »

At what level are you referring to turning hyper threading off at? I know It can be turned off in the BIOS, but if at all possible I would like to disable it for this Guest only and not for the entire ESX host.

I am referring to the bios. If it can be disabled at the guest host then try that if thats a no go then try the bios. Either way it should be a quick way if you disable it in the bios to atleast test the theory. I saw this happen with various older os's on hyper-threading with p4's it would use 100% cpu all the time just sitting there. I guess will see if thats it. Hope it works out for ya.

astorrs

kriscamaro68 wrote: »

I am referring to the bios. If it can be disabled at the guest host then try that if thats a no go then try the bios. Either way it should be a quick way if you disable it in the bios to atleast test the theory. I saw this happen with various older os's on hyper-threading with p4's it would use 100% cpu all the time just sitting there. I guess will see if thats it. Hope it works out for ya.

There's no point in this as hyperthreading has worked great since ESX 3.5 and the guest can't tell the difference between a physical and/or logical processor. Like I said before, cleanup the garbage you brought over first (including Add/Remove Programs for any RAID software, HBA software, etc)

Sunshine_54

astorrs is correct. Remove all of the non present hidden devices and anything dealing with the old hardware from add/remove programs. It's normal for the CPU to be pegged after a conversion. If you check process explorer it's most likely hardware interrupts to the missing hardware. Make the changes above and this should be resolved.

slinuxuzer

This turned out to be an issue with a piece of compaq hardware monitoring software, can't remember the name, but the process was sysdown.exe once I stopped this service it took a minute for the server to calm down, (this was likely other software I have loaded) and then it went down to about 8% processor

I also had trouble getting booted into safe mode, wich would have allowed me to tinker with the VM without the processor being pegged, the boot up was just too fast for me, there is a delay option under edit settings for the VM that I set to 5000 milisecond or 5 second that allowed me enough time to hit F8 and go into safe mode.

Also some notes on the Cold Migration CD,

1. The source and destination systems need to be in the same Vlan, or atleast they did on my network, this seems right since Ghost and all other imaging software I have used works best if both systems are on the same Vlan.

2. Haven't gotten the PETOOL to work properly yet, but one caviat is when you add drivers to it the last directory needs to be CAPITALIZED why I don't know, but it will error out.

Thanks to everyone for all the help.