Migrate XP or other Windows OS’s to VirtualBox

 




I needed to migrate a Windows XP box because it has a special piece of software…what special piece of software you ask?  A real museum piece!  Lotus 1-2-3 with a special quote form I didn’t have time to migrate into something more contemporary.

I ran into lots of issues, here’s how I worked past those issues and made it happen:
* Fair warning here, you MAY have to re-register Windows when it boots up (I had to for XP but it worked to do so)

First:  Read the how to from VirtualBox:  https://www.virtualbox.org/wiki/Migrate_Windows

* After renaming the agp440.sys and the intelppm.sys then running the MergeIDE bat file image the machine per their instructions

1 – Make an image of the current physical box with clonezilla. It will bust it up into different size files, the default is 2000 megabytes (2 gig chunks).  I’d make it much larger, if you have an 80 gig drive make it 85000 (85 gig just to be safe).  I didn’t happen to do that but I think it’s an arbitrary thing ultimately since when you read the image back into the disk clonezilla automatically does it’s thing regardless of chunk size.

2 – Create a new XP virtual machine in VirtualBox (can’t hold your hand on this one…it would take too long to tell you how to do that so I assume you’re not a total derp and you can manage this step).

3 – Select the new XP image > settings > storage > select the CD drive (Empty) > Click the drop down to the far right of “Optical Drive” and select your clonezilla ISO (I assume you have this in a folder on the new machine running VirtualBox).

4 – Boot the machine and image your saved image back to the virtual drive.  Once done shut the machine down.

5 – Here’s where I ran into a lot of problems on boot up:

Virtual Box boots up and it’s just a black screen.  For this you need to be quick on the draw and hit F8 repeatedly to bring up the safe mode prompt.  Then boot into safe mode, you should see either more black screen OR files flying accross the screen.
If you also see the lines at the bottom just haning there and doing nothing this is also the fix for this problem:

I found that in c:\windows\inf\ <– that folder on my imaged machine was EMPTY!  Copy the data from your old physical box into your new one.  I had to rename the INF folder to INF.kill > make a new INF folder then dump my files in there.  You might also want to delete any “oem” inf files, in Linux you can do a:  “ls | grep oem” to see the files with oem in their names then “rm -f oem*.inf” to delete all the INF files, note there are “PNF” files still left.  I left those…not sure what they’re for!

Once you do that…more problems but we can fix them.

I got a blue screen that said:  {Registry File Failure}:
Good one!  It says:  \SystemRoot\System32\Config\DEFAULT or its log or alternate.  It is corrupt, absent, or not writable.

All you need to do here on XP is restore the registry backup!  In c:\windows\system32\config there’s a file named “default” and “default.save”.  Rename the “default” to “default.kill” so the system won’t look at it then rename “default.save” to “default”.  The .save file is your registry’s “last known good” ;)
You can also go to your old physical box > boot up a Linux Live disk > migrate to that folder and copy your current “default” file onto removable media (USB disk) > then sneaker net it over to your other box > overwrite the default in the c:\windows\….config folder.

Reboot!  Try safe mode just to be safe.
Success (at least for me).  You then login and you get errors about:  “The file or directory c:\windows\system is corrupt and unreadable. Please run the Chkdsk utility.”.

** Fair warning again – this totally HOSED me when it booted up because when it booted up it didn’t want to run the registration wizard so I re-did the migration knowing what to check (see above notes so far) and it went perfectly the second time…I think the original image back was borked since the INF folder was empty and the registry file was bad.

Computer > right click c: > properties > tools > error checking > check both boxes > start > reboot
Windows will go through a litany of file systems checks and repair the error.  Windows should now come up normally.
After that’s all done > install your guest additions > make sure your drivers area good and things work.  If it mostly works I’d call it a success!

One last thing – you need to delete any OEM inf files in c:\windows\inf folder > you can ignore the *.pnf files.  Delete all the OEM one’s.  When I booted up my Live Linux distro I went to the terminal > cd’d to the directory (wherever yours mounts it) then did a:

ls | grep oem

or you might need to do:  ls | grep OEM

Just to make sure you have all the oem INF files in the list and your syntax is correctly typed THEN you can do:  rm -f oem*.inf <– enter

Done!

Now reboot and let it boot up normally.

This is basically what I did – just do it!  You have the saved image and as long as it imaged correctly you can do this all day ;)

I had an interesting thing I hadn’t seen before on a VMware guest that was a Windows Server 2012 machine.

Symptoms:
System process (PID 4) 50% cpu or higher:  vmware 5.x + server 2012.

I looked at another 2012 server and noticed it had the e1000 NIC in it and I remember when building this current server I had changed it to the vmxnet3 thinking, oh…vmnet3 NIC must be better than the e1000.  I was wrong.  I don’t think Server 2012 likes the vmxnet3 for whatever reason (driver related) regardless of whether you have VMware Tools installed or not because I did; and it wasn’t having problems.  So I Changed the NIC from vmxnet3 to e1000.

Again; it’s related to a networking issue (References):

http://blogs.technet.com/b/yongrhee/archive/2009/08/07/how-to-troubleshoot-high-cpu-in-the-system-process.aspx
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1077
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2231

I don’t think any of those happen to actually say, switch from vmxnet3 to e1000 BUT that’s what I did and the problem went away!

I experienced something interesting the past week for a client.  Slow performance on iSCSI, SAN and VMWare when large throughput or heavy network i/o was occurring.  They had a setup like this:

VMware > All VM’s on a big fat volume iSCSI to the SAN > SAN < 1 TB volume

Inside the guest (Server 2012 for example), you then make an iSCSI connection to the SAN to run their db’s, file shares, etc.  They were using Sage 2014 and it was VERY slow especially when printing or running reports.   It took me what seemed like FOREVER to figure it out BUT in the end it was all about the MTU.  If you don’t know what MTU is, go look it up :P

In the end make sure your MTU matches this with a Dell setup:

SAN interfaces = MTU 9000 (they had a Dell Equallogix PS4100 – that might be auto configured, I think it is)
Switch where SAN & VMWare connect = MTU 9252 (they had a Dell Force10 S60)
VMware = MTU 9000 (that means ALL vswitches)

Note that you ALSO want MTU on the interface where your VM’s talk to the rest of your network.  For me that was the vmkernel port and the physical switch ports on the core switch that connected the physical to the virtual – MTU 9252.

There’s a decent kb from VMWare that pointed me in the right direction.  Oh, I also believe I disabled the ACK on the VMware iSCSI connector inside the VMWare server based on another KB but this didn’t really help me as far as I know, I have not turned it back on.  Again, check that out online if the MTU confirmation doesn’t work for you it might be the ACK on the iSCSI connector.

Working on a project for stuff I personally own and manage (VMWare Servers and various guest OS’s) I had to convert machines using VMWare Stand Alone Converter.

A quick note about that – if you’re converting Linux Machines that have the “root” account disabled by default like Ubuntu has then you need to log into your Ubuntu target and make the root account able to login via the console and ssh.

Some by default deny root logon to ssh (a good idea!) but you’d have to set that – it’s not set by default in the ssh server config.  If you haven’t done that then all you need to do is this:

1 – Login to your Ubuntu workstation as your regular non privileged user

2 – sudo -i

3 – Type in the password for the non privileged user – this will then elevate you to “root”

4 – Type:  passwd

5 – Type in your new password two times

6 – Logout

Now you can use those credentials in VMWare converter to migrate your machine from one server to another!  Also note, when doing this you’ll need to adjust your NIC since it won’t get an IP address (see the post before this one).

Also note – once your machine is converted set your root account back to default so it can’t login.  To “disable” the account type in:

passwd -dl root

* Note – you should be logged in as root to run this command!  Then “exit” and log out.  Try logging in as root now – you shouldn’t be able to but you can sudo to still elevate your non privileged user to root privileges.

Reference: https://help.ubuntu.com/community/RootSudo

Just a quick note to the masses, I used VMWare Converter 5.x Stand Alone (It’s not version specific) to migrate an Ubuntu 8.04 Server I have doing SPAM filtering and when I type in “ifconfig” to determine if the IP address came up OK all I saw was localhost (127.0.0.1).  I made sure to adjust DHCP to see the new MAC address and I KNEW this was going to be an issue – not with DHCP but with converting Ubunutu like this – it looses eth0’s IP address.  Why?  Because the MAC address changes from the old VMWare Servers MAC address allocation to the new VMWare Servers MAC address allocation – essentially, the new server issues the new / imported VM a new MAC.  This makes the OS install the new MAC as an additional NIC…because it is!

I found the answer to my problems here, I used this post before to fix this same issue in the past but now I’m documenting it on my blog so I can remember :P

Post:  Eth0 disapears in VMware and Ubuntu Server

SuperkiKim is the bomb!

Essentially you shut down two services and edit a file, then bring those services up.

Here are the commands thanks for the VMWare Community Forum and SuperkiKim:

Stop services:
/etc/init.d/udev stop
/etc/init.d/networking stop
Edit the file:
nano etc/udev/rules.d/70-persistent-net.rules

Comment out the first PCI devices in the list, at the end of the line it'll say "eth0" with a "#" symbol
# SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="00:0c:29:36:xx:xx", ATTR{type}=="1", NAME="eth0"

Then - on the next PCI device (SUBSYSTEM) where it probably says "eth1" change that to "eth0"
Save it by exiting NANO, ctrl + x sometimes it's ctrl + alt + x
Hit "Y" to save

restart your services by typing:
/etc/init.d/udev start
/etc/init.d/networking start

ifconfig command should show you your IP that you expected!