Virtualization inside a virtualizer, discussions
-
- Posts: 2
- Joined: 23. Jun 2010, 15:58
- Primary OS: Ubuntu other
- VBox Version: PUEL
- Guest OSses: Ubuntu, Windows, Fedora
Re: Virtualization inside a virtualizer, discussions
Afaik, KVM can do nested virtualization on AMD CPU's. Not sure how exactly does it work tho.
-
- Posts: 4
- Joined: 11. Aug 2010, 23:56
- Primary OS: Ubuntu other
- VBox Version: OSE Debian
- Guest OSses: Ubuntu
Re: Virtualization inside a virtualizer, discussions
Well, in my case I want to have a testrig guest that has the same configuration including IP addressing as a live server (i.e. so I don't have to change config files that explicitly mention IP addresses). And I don't want to change my desktop routing to talk to this guest, because I need to talk to the real live server too.DaveHCYJ wrote:Just curious, why would you run virtualization inside of virtualization? What are the benefits or what would be scenario where you would need this? Couldn't all the inner VM's just run in the outer instance?
So I have an "access" guest, where I can play with the routing table. This access machine can't talk to the real server addresses, only to the test version ones. Now, I *could* run the testrig as a VM in the outer instance, and communicate with the access machine over intnet; but to do this I'd have to remember to startup two separate machines (yeah, I know, I could script this trivially).
But, if I have the testrig running as a guest inside the access machine, I can just boot it up once and ignore it; just suspend/start the access machine whenever I want them both. This counts as a benefit, for me
Now, I had hoped that I could run VBox inside a VBox, but that doesn't seem to be working even when I switch loads of options off. But the default qemu springs into life just fine, giving me VNC access to the console of the testrig, so I can connect to it from the desktop if I need to (i.e. during installation). So even though I'm told that qemu "is not a virtualizer" it does achieve the use-case I wanted.
+-- desktop ------------------------+ | +-- (VBox OSE) access ---------+ | | | +-- (qemu) testrig -------+ | | | | | | | | | | +-------------------------+ | | | +------------------------------+ | +-----------------------------------+
-
- Volunteer
- Posts: 3342
- Joined: 10. May 2007, 16:59
- Location: Israel
Re: Virtualization inside a virtualizer, discussions
JimCheetham: For your testcase, much more reasonable is to use "internal" network and 2 VBox VMs.
VM1 will have 2 NICS - 1 bridged (or NAT), and 1 internal (connected to VM2).
VM2 will have 1 NIC - internal (connected to VM1)
like you said "Now, I *could* run the testrig as a VM in the outer instance"
All the complex network stuff can be done with 1 layer of virtualization. Performance will be better.
about the : qemu "is not a virtualizer" -- read my article here:
http://chorgan.provo.novell.com/lfl/.ht ... ation.html
-Technologov
VM1 will have 2 NICS - 1 bridged (or NAT), and 1 internal (connected to VM2).
VM2 will have 1 NIC - internal (connected to VM1)
like you said "Now, I *could* run the testrig as a VM in the outer instance"
All the complex network stuff can be done with 1 layer of virtualization. Performance will be better.
about the : qemu "is not a virtualizer" -- read my article here:
http://chorgan.provo.novell.com/lfl/.ht ... ation.html
-Technologov
-
- Posts: 9
- Joined: 6. Oct 2010, 19:24
- Primary OS: MS Windows XP
- VBox Version: OSE other
- Guest OSses: xp
Re: Virtualization inside a virtualizer, discussions
I've had perfect results with this configuration: VmWare ESX hosting Windows XP SP3, which runs VirtualBox hosting Windows XP SP3.
My problem is that it is slow. Not just slow, but OMG slow. VmWare ESX hosting Windows XP SP3 is snappy and ideal as my development environment. It makes my life easier if I can run VirtualBox to host Windows XP SP3 for parts of my development. This works fine on my desktop Windows XP SP3 development machine. It also works fine in the VmWare ESX hosting Windows XP SP3. But, as I said, its so slow. 30x slower, I'm guessing.
So, any hints on how to improve performance?
Any suggestions are welcome.
Email is good, because I probably will forget to check this forum.
Regards.
Mark K Vallevand
mark.vallevand@unisys.com
My problem is that it is slow. Not just slow, but OMG slow. VmWare ESX hosting Windows XP SP3 is snappy and ideal as my development environment. It makes my life easier if I can run VirtualBox to host Windows XP SP3 for parts of my development. This works fine on my desktop Windows XP SP3 development machine. It also works fine in the VmWare ESX hosting Windows XP SP3. But, as I said, its so slow. 30x slower, I'm guessing.
So, any hints on how to improve performance?
Any suggestions are welcome.
Email is good, because I probably will forget to check this forum.
Regards.
Mark K Vallevand
mark.vallevand@unisys.com
-
- Posts: 2
- Joined: 7. Oct 2010, 12:20
- Primary OS: MS Windows 7
- VBox Version: OSE Debian
- Guest OSses: Windows7
Re: Virtualization inside a virtualizer, discussions
I also agree with you that very low amount of users has a nested paging capable CPU and some users don't even have hardware-v available.
-
- Posts: 12
- Joined: 21. Dec 2009, 10:41
- Primary OS: MS Windows 7
- VBox Version: OSE other
- Guest OSses: Windows Server 2008 R2, Windows 7 x64, Windows XP x86
Re: Virtualization inside a virtualizer, discussions
In VMware, the code looks like this to run a Virtual Machine:
Put that in a VMX file.
I can run VirtualBox's virtual machine inside my VMware machine but no VT.
But in virtualbox, hmm hard to say I don't really know how, and VirtualBox must have an option to limit the CPU like limit the CPU to 50%
There are still a way to Virtualize inside a virtualizer (see my code).
Code: Select all
isolation.tools.getPtrLocation.disable = "TRUE"
isolation.tools.setPtrLocation.disable = "TRUE"
isolation.tools.setVersion.disable = "TRUE"
isolation.tools.getVersion.disable = "TRUE"
monitor_control.disable_directexec = "TRUE"
monitor_control.disable_chksimd = "TRUE"
monitor_control.disable_ntreloc = "TRUE"
monitor_control.disable_selfmod = "TRUE"
monitor_control.disable_reloc = "TRUE"
monitor_control.disable_btinout = "TRUE"
monitor_control.disable_btmemspace = "TRUE"
monitor_control.disable_btpriv = "TRUE"
monitor_control.disable_btseg = "TRUE"
I can run VirtualBox's virtual machine inside my VMware machine but no VT.
But in virtualbox, hmm hard to say I don't really know how, and VirtualBox must have an option to limit the CPU like limit the CPU to 50%
There are still a way to Virtualize inside a virtualizer (see my code).
-
- Posts: 17
- Joined: 3. Jan 2011, 12:08
- Primary OS: MS Windows 7
- VBox Version: OSE other
- Guest OSses: Debian 5
Re: Virtualization inside a virtualizer, discussions
Hello,
Like bornmw, I developed an application that I distribute as a vdi file. I am wondering if it would work in a Citrix environment : would the clients be able to use my application if it ran under XenServer ?
Some posts below made me think running VB under Xen is not possible, some others made me think the contrary. Who's right or wrong ?
If somebody managed to run VB under XenServer (or knows that it is impossible), I'd be glad to know it
Like bornmw, I developed an application that I distribute as a vdi file. I am wondering if it would work in a Citrix environment : would the clients be able to use my application if it ran under XenServer ?
Some posts below made me think running VB under Xen is not possible, some others made me think the contrary. Who's right or wrong ?
If somebody managed to run VB under XenServer (or knows that it is impossible), I'd be glad to know it
-
- Posts: 2
- Joined: 20. Jan 2011, 11:45
- Primary OS: FreeBSD
- VBox Version: OSE self-compiled
- Guest OSses: too many
Re: Virtualization inside a virtualizer, discussions
I successfully ran Windows 2000 in VirtualBox-OSE 3.2.12 on FreeBSD 8.1 in VMware Fusion 3.1.2. Host CPU is a Core i3 with VT. No editing of the outer guest's VMX file was required, it fired right up. I haven't done much with it yet, but it seems stable and surprisingly fast.
-
- Posts: 2
- Joined: 23. Jan 2011, 03:06
- Primary OS: MS Windows 7
- VBox Version: OSE other
- Guest OSses: win xp/jolicloud
Re: Virtualization inside a virtualizer, discussions
Ive just finished installing win xp inside win xp inside win 7.
latest version of oracle virtual box. standard install of everything.
http://img38.imageshack.us/i/3layersdown.jpg/
looks nice.
response time is excellent, except the mouse jumps quite a bit.
latest version of oracle virtual box. standard install of everything.
http://img38.imageshack.us/i/3layersdown.jpg/
looks nice.
response time is excellent, except the mouse jumps quite a bit.
-
- Posts: 2
- Joined: 23. Jan 2011, 03:06
- Primary OS: MS Windows 7
- VBox Version: OSE other
- Guest OSses: win xp/jolicloud
Re: Virtualization inside a virtualizer, discussions
scrub that. fixed the mouse issue. its running perfectly fine now.
doing some stress tests on everything now.
Im trying to install a light weight OS jolicloud as a fourth layer now, but its not having it
doing some stress tests on everything now.
Im trying to install a light weight OS jolicloud as a fourth layer now, but its not having it
-
- Posts: 9
- Joined: 26. Jan 2011, 11:23
- Primary OS: MS Windows 7
- VBox Version: PUEL
- Guest OSses: windows xp
Re: Virtualization inside a virtualizer, discussions
Hello,
I am trying to run a VirtualBox VM on top of installing VirtualBox within another VM (created by other means, tools)...
I get the following error, Any Ideas?
Thanks,
Niranjan
Failed to open a session for the virtual machine winXPNoVT
cpum#1 "CPU model" mismatch: host=0x2c saved=0x17 [ver=12 pass=final]
(VERR_SSM_LOAD_CPUID_MISMATCH)
Result Code: E_FAIL (0x80004005)
Component: Console
Interface: IConsole {<Some Random Number Id>}
I am trying to run a VirtualBox VM on top of installing VirtualBox within another VM (created by other means, tools)...
I get the following error, Any Ideas?
Thanks,
Niranjan
Failed to open a session for the virtual machine winXPNoVT
cpum#1 "CPU model" mismatch: host=0x2c saved=0x17 [ver=12 pass=final]
(VERR_SSM_LOAD_CPUID_MISMATCH)
Result Code: E_FAIL (0x80004005)
Component: Console
Interface: IConsole {<Some Random Number Id>}
-
- Posts: 1
- Joined: 16. Feb 2011, 19:33
- Primary OS: Fedora other
- VBox Version: OSE Fedora
- Guest OSses: centos
Re: Virtualization inside a virtualizer, discussions
Just got it to work:
Fedora 14 host, CentOS5.4 guest (aka Level 1 guest) successfully built a CentOS5.4 VM (aka Level 2 guest). Key for me was to make sure the guest additions and shared folder kernel modules were not loaded in the Level 1 CenOS5.4 guest (the one used to build the second VM). Once I renamed those kernel modules and rebooted the Level 1 guest, i was able to create a new VM from within the Level 1 guest.
Mouse is kind of wonky and you lose the ability (at least I did) to access the Level 1 guest while the Level 2 guest is up (including during the build).
I was quite surprised, actually, to get it to work for it seems like there are many reasons why it would not. Goes to show that the VirtualBox developers got things right.
Fedora 14 host, CentOS5.4 guest (aka Level 1 guest) successfully built a CentOS5.4 VM (aka Level 2 guest). Key for me was to make sure the guest additions and shared folder kernel modules were not loaded in the Level 1 CenOS5.4 guest (the one used to build the second VM). Once I renamed those kernel modules and rebooted the Level 1 guest, i was able to create a new VM from within the Level 1 guest.
Mouse is kind of wonky and you lose the ability (at least I did) to access the Level 1 guest while the Level 2 guest is up (including during the build).
I was quite surprised, actually, to get it to work for it seems like there are many reasons why it would not. Goes to show that the VirtualBox developers got things right.
-
- Posts: 4
- Joined: 24. Sep 2010, 00:01
- Primary OS: Ubuntu other
- VBox Version: OSE Debian
- Guest OSses: Multiple
Re: Virtualization inside a virtualizer, discussions
Successfully started & installed VMware vSphere Hypervisor (ESXi - build 348481) inside Virtualbox 4.0.4 on Ubuntu 10.04.2 LTS
* What was key for me were errors inside VMware's original failed boot in regards to measured cpu and bus speeds in conflict (press ESC from purple screen, press l at prompt, and scroll away) ; so focused on anything that would involve measurement code (re, CPUs frequencies, system timers, etc...)
- Note-worth-ies on my hardware
Pavilion dv6 (dv6-3006TX)
WW961PA#ABG
* Notebook PC
* Intel (R) Core (TM) i5-430M processor 2.26 GHz with Turbo Boost up to 2.53 GHz
* 15.6" diagnoal High Definition HP BrightView LED Display (1366x768)
* 500GB (7200RPM) Hard Drive
* 4096MB DDR3 SDRAM (2 Dimm) ==> upgraded to 8GB
* 802.11b/g/n WLAN
* LightScribe Super Multi 8X DVD+/-RW with Double Layer Support
* 6-Cell 55WHr Lithium-Ion Battery
* HP QuickWeb
* ATI Mobility Radeon (TM) HD 5650 Switchable Graphics with 1024MB DDR3 dedicated graphics memory
* 5-in-1 Digital Media Reader
* HP TrueVision webcam
* HP MediaSmart
Host build order
- Fresh UBUNTU install from Ubuntu Desktop 10.04.1 LTS (AMD64) CD
- Fully patched, incl kernel - here's `uname -a` ...
Linux csmith-laptop 2.6.32-24-generic #43-Ubuntu SMP Thu Sep 16 14:58:24 UTC 2010 x86_64 GNU/Linux
- Packages installed at once (most probably not related to VMware or Virtualbox, but have not formally tested out components) ...
openjdk-6-jre
rcconf
build-essential
ncurses-base
libncurses5-dev
flashplugin-installer
- (Old school M!cro$oft habits) rebooted before installing virtualbox via instructions here = "http://www.virtualbox.org/wiki/Linux_Downloads"
- The following config / changes have come from lots of Googling - mostly specific to my hardware / config
!!!!! IMPORTANT !!!! ===>>>
After each of the following changes, would advise combination of re-running /etc/init.d/vboxdrv setup (to update against DKMS) and normal rebooting (to catch boot time parameter changes) (yeah, more old school M!cro$oft habits, plus I haven't formally tested what's critically dependent on what)
!!!!! IMPORTANT !!!! <<<===
* Updated /etc/default/grub for this line ...
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash clocksource=hpet"
* Updated HPET user freqency at command line ...
sysctl -w dev.hpet.max-user-freq=1024
* Updated vboxdrv TSC to ASYNC; creation of /etc/modprobe.d/vbox.conf with following ...
options vboxdrv force_async_tsc=1
* Finally, installed package 'cpufrequtils' ; noticed it bound kernel parameters in relation to my number of cores
* Re, this last package and another /etc/init.d/vboxdrv setup did something good - for me, it was the clincher to get things going
- I found i7z which was v. helpful in seeing live numbers (summary) on how my CPUs were behaving; needs ncurses
https://code.google.com/p/i7z/
Guest parameters
- Again, after lots of Googling, key changes were (if you want the XML of the Guest, let me know) ...
* Kept Operating System to "Other"; Kept Version to "Other/Unknown"
* Base Memory: 4GB
* 4 Processors, with PAE/NX
* Enable VT-x / AMD-V, with Nested Paging
* Chipset: ICH9
* Enable IO APIC
* Hardware clock in UTC time
* 16GB Fixed Disk on SATA controller
* Intel PRO/1000MT Desktop NIC
- Successfully booted from VMware-VMvisor-Installer-4.1.0.update1-348481.x86_64.iso and installed VMware
I am not a guru or anything, but more than happy to chat about any of this -
Regards,
Christopher
* What was key for me were errors inside VMware's original failed boot in regards to measured cpu and bus speeds in conflict (press ESC from purple screen, press l at prompt, and scroll away) ; so focused on anything that would involve measurement code (re, CPUs frequencies, system timers, etc...)
- Note-worth-ies on my hardware
Pavilion dv6 (dv6-3006TX)
WW961PA#ABG
* Notebook PC
* Intel (R) Core (TM) i5-430M processor 2.26 GHz with Turbo Boost up to 2.53 GHz
* 15.6" diagnoal High Definition HP BrightView LED Display (1366x768)
* 500GB (7200RPM) Hard Drive
* 4096MB DDR3 SDRAM (2 Dimm) ==> upgraded to 8GB
* 802.11b/g/n WLAN
* LightScribe Super Multi 8X DVD+/-RW with Double Layer Support
* 6-Cell 55WHr Lithium-Ion Battery
* HP QuickWeb
* ATI Mobility Radeon (TM) HD 5650 Switchable Graphics with 1024MB DDR3 dedicated graphics memory
* 5-in-1 Digital Media Reader
* HP TrueVision webcam
* HP MediaSmart
Host build order
- Fresh UBUNTU install from Ubuntu Desktop 10.04.1 LTS (AMD64) CD
- Fully patched, incl kernel - here's `uname -a` ...
Linux csmith-laptop 2.6.32-24-generic #43-Ubuntu SMP Thu Sep 16 14:58:24 UTC 2010 x86_64 GNU/Linux
- Packages installed at once (most probably not related to VMware or Virtualbox, but have not formally tested out components) ...
openjdk-6-jre
rcconf
build-essential
ncurses-base
libncurses5-dev
flashplugin-installer
- (Old school M!cro$oft habits) rebooted before installing virtualbox via instructions here = "http://www.virtualbox.org/wiki/Linux_Downloads"
- The following config / changes have come from lots of Googling - mostly specific to my hardware / config
!!!!! IMPORTANT !!!! ===>>>
After each of the following changes, would advise combination of re-running /etc/init.d/vboxdrv setup (to update against DKMS) and normal rebooting (to catch boot time parameter changes) (yeah, more old school M!cro$oft habits, plus I haven't formally tested what's critically dependent on what)
!!!!! IMPORTANT !!!! <<<===
* Updated /etc/default/grub for this line ...
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash clocksource=hpet"
* Updated HPET user freqency at command line ...
sysctl -w dev.hpet.max-user-freq=1024
* Updated vboxdrv TSC to ASYNC; creation of /etc/modprobe.d/vbox.conf with following ...
options vboxdrv force_async_tsc=1
* Finally, installed package 'cpufrequtils' ; noticed it bound kernel parameters in relation to my number of cores
* Re, this last package and another /etc/init.d/vboxdrv setup did something good - for me, it was the clincher to get things going
- I found i7z which was v. helpful in seeing live numbers (summary) on how my CPUs were behaving; needs ncurses
https://code.google.com/p/i7z/
Guest parameters
- Again, after lots of Googling, key changes were (if you want the XML of the Guest, let me know) ...
* Kept Operating System to "Other"; Kept Version to "Other/Unknown"
* Base Memory: 4GB
* 4 Processors, with PAE/NX
* Enable VT-x / AMD-V, with Nested Paging
* Chipset: ICH9
* Enable IO APIC
* Hardware clock in UTC time
* 16GB Fixed Disk on SATA controller
* Intel PRO/1000MT Desktop NIC
- Successfully booted from VMware-VMvisor-Installer-4.1.0.update1-348481.x86_64.iso and installed VMware
I am not a guru or anything, but more than happy to chat about any of this -
Regards,
Christopher
-
- Posts: 12
- Joined: 21. Dec 2009, 10:41
- Primary OS: MS Windows 7
- VBox Version: OSE other
- Guest OSses: Windows Server 2008 R2, Windows 7 x64, Windows XP x86
Re: Virtualization inside a virtualizer, discussions
You ran an hypervisor in VirtualBox ? I thought it was impossible because it uses VT-x and this last inside a virtual machine is impossible...
How do I run Hyper-V (Windows Server 2008 R2) without having VT-x support?
How do I run Hyper-V (Windows Server 2008 R2) without having VT-x support?
Re: Virtualization inside a virtualizer, discussions
Not all hypervisors need vt-x for them to work, ea. 64bit guests will be a problem.
[This space is intentionally left blank]
If you can read this, you can read the VirtualBox Manual, the Forum FAQ, and the QuickClick FAQ
-=[ Search this forum with Keywords, VirtualBox solutions at you're fingertips]=-
If you can read this, you can read the VirtualBox Manual, the Forum FAQ, and the QuickClick FAQ
-=[ Search this forum with Keywords, VirtualBox solutions at you're fingertips]=-