Discuss the 3.2.0 release

[mykes]

I have windows 7 x64 host, Ubuntu Lucid (Kubuntu) guest.

I upgraded from 3.1.8 to 3.2

I figured I'd try out the dual monitor support, so changed monitors to 2.

Booted my VM (after backing up the .vdi file!)

The guest doesn't see the dual monitors. D'oh, it says it's only supported for windows guests in the manual. Please fix this for upcoming releases!

I notice, though, in seamless mode, that my desktop seems to be centered and about 1/2 the width of my primary monitor. If I drag linux windows too far right, they are clipped at about 2/3 the screen width.

So I go to shutdown and I notice a 2nd VM window is open, probably for the 2nd monitor. It's black though. Kubuntu says I only have one monitor to configure...

I shutdown and change the monitor count back to 1.

Boot kubuntu in the VM, seamless mode works again. But when I drag a linux window, the linux rendered things on my whole screen flash. The inactive windows, the toolbar, etc.

First thing I did when I booted the new VM was install guest additions.

Also, the bottom panel in lucid / kubuntu isn't rendered right. See attachment. The bar is supposed to be the silver color on the right, but renders black on the left for some reason.

[sej7278]

sej: he did not say "slow down" - he said his VMs refuse to work.

oh ok, i've seen that happen on my core2quad which doesn't have npt, when i try to run a vdi from my core-i5 which does, but that was 3.1 something, i've not seen it on the 3.2 betas on linux.

are you sure the core i5 661 supports nested paging - its one of those mobile processors with the gpu onboard isn't it? it supports vt-x for sure. if it doesn't support npt it ain't gonna run vm's configured for it.

bit worrying as part of the recent changes was to enable the nested paging option by default.

[Technologov]

All Core iX (Nehalem class) processors support EPT (Nested pages). Even i3.

[petrossa]

I'm pretty impressed actually.
Win7 x64 host, windows quests multiple monitors works like a charm.
A few minor drawing issues i mentioned before (the vbox sessions aren't skinnable).
And for the other os's, from ubuntu to osx guests work quite well. Weird thing, i can't get the os version of osx. If i click on about this mac i get a blue screen ( really) in the guest for about 20 seconds and then it comes back.

Forgot:
Processor load is way down. Perfect. Were it used to be around 25% on my phenom II even when the quest was idle, now it idles down to practically zero.

[ilborg]

On arch linux ose version of the release doesn't seem to work. Only " Guru Meditation -2403 (VERR_TRPM_DONT_PANIC)" seems to happen on every guest(centos, debian, vista). I did try to change various switches without success. Bad thing that Arch linux doesn't have rollback-option to get guests working, and you can't compile easily VB in 64-bit environment because arch is missing multilib(and virtualbox is using some 32-bit stuff).

I'll include log from guru meditation. It always seems to occur in linux when it is starting kernel. Last line is the "Booting kernel." -line, and then the guru meditation happens.

[cyruspy]

Anybody tested import/export feature?. I tried to export from a Linux host (v3.2.0) and import into a Windows host "v3.2.0" and the UUID defined in the OVF is different from the real UUID of the VMDK file (queried with "VBoxManage showhdinfo").

Modified the OVF by hand to replace the UUID with the correct one and on the first retry VBox crashed on the Windows Host. On the third try it imported the machine.

Edit: It happened with a second machine export, so I think it's a bug....

[Sasquatch]

Anybody tested import/export feature?. I tried to export from a Linux host (v3.2.0) and import into a Windows host "v3.2.0" and the UUID defined in the OVF is different from the real UUID of the VMDK file (queried with "VBoxManage showhdinfo").

Modified the OVF by hand to replace the UUID with the correct one and on the first retry VBox crashed on the Windows Host. On the third try it imported the machine.

So the normal import worked like a charm, but after you modified the UUID in the OVF image you got a crash? That's a user error, not a VB problem. You change the file VB creates, it's asking for it to barf at you.

[cyruspy]

Hmm, nope, please read again. The just-exported-original-OVF got a wrong UUID for the related VMDK disk. I had to modify the file and set the correct UUID to make it work. On the first retry (second attempt) it crashed, i'm not sure why. Retried the import (a third time) without touching the OVF and it worked. A second exported machine also got the wrong UUID for its disk.

[Perryg]

Seems to work for me.
From the ovf file

Code: Select all

 <Disk ovf:capacity="8589934592" ovf:diskId="vmdisk1" ovf:fileRef="file1" ovf:format="http://www.vmware.com/interfaces/specifications/vmdk.html#streamOptimized" vbox:uuid="65984d8a-85cd-4f12-85e6-4de340ea3edb"/>

From the showinfo

Code: Select all

Storage Controller (0, 0): /home/perry/.VirtualBox/HardDisks/Debian-Squeeze.vdi (UUID: 65984d8a-85cd-4f12-85e6-4de340ea3edb)

[cyruspy]

Seems to work for me.
From the ovf file

Code: Select all

 <Disk ovf:capacity="8589934592" ovf:diskId="vmdisk1" ovf:fileRef="file1" ovf:format="http://www.vmware.com/interfaces/specifications/vmdk.html#streamOptimized" vbox:uuid="65984d8a-85cd-4f12-85e6-4de340ea3edb"/>

From the showinfo

Code: Select all

Storage Controller (0, 0): /home/perry/.VirtualBox/HardDisks/Debian-Squeeze.vdi (UUID: 65984d8a-85cd-4f12-85e6-4de340ea3edb)

Did you try to import it?. That's your original VDI disk, check the generated VMDK they're not the same and VBox is writing the original UUID on the OVF, not the new one from the generated disk.

[Perryg]

Well I did try to import it and it puked just after I posted this. See Screen shot below.
I am not so concerned about the UUID at the moment. AFAIK that should not make the import fail, but I am looking into it now.

Here is the HDinfo from the vmdk file

Code: Select all

perry@Ubuntu-10:~$ VBoxManage showhdinfo /home/perry/Documents/Debian-Squeeze.vmdk
Oracle VM VirtualBox Command Line Management Interface Version 3.2.0
(C) 2005-2010 Oracle Corporation
All rights reserved.

UUID:                 58469214-51d4-4f36-9231-bc567b731979
Accessible:           yes
Description:          
Logical size:         8192 MBytes
Current size on disk: 2967 MBytes
Type:                 normal (base)
Storage format:       VMDK
Location:             /home/perry/Documents/Debian-Squeeze.vmdk

DS-Import.jpg (47.95 KiB) Viewed 7646 times

[Perryg]

OK second attempt worked. I deleted the SATA controller that I was not using and it appears it went as planned.

Also the UUID is regenerated on import and the fact that they are not the same does not matter (AFAIK). The DEVs would need to answer that one.
So the question I have is did you receive the same error I did and what are the actual settings of the VM? Did you try to select the legacy .9 or just the default?
I will of course continue to investigate this but your input would help.

As you can see there are a total of 3 different UUIDs on this working import.

DS2-Import.jpg (60.34 KiB) Viewed 7644 times

[cyruspy]

Well I did try to import it and it puked just after I posted this. See Screen shot below.
I am not so concerned about the UUID at the moment. AFAIK that should not make the import fail, but I am looking into it now.

Here is the HDinfo from the vmdk file

Code: Select all

perry@Ubuntu-10:~$ VBoxManage showhdinfo /home/perry/Documents/Debian-Squeeze.vmdk
Oracle VM VirtualBox Command Line Management Interface Version 3.2.0
(C) 2005-2010 Oracle Corporation
All rights reserved.

UUID:                 58469214-51d4-4f36-9231-bc567b731979
Accessible:           yes
Description:          
Logical size:         8192 MBytes
Current size on disk: 2967 MBytes
Type:                 normal (base)
Storage format:       VMDK
Location:             /home/perry/Documents/Debian-Squeeze.vmdk

DS-Import.jpg

From the error, apparently the disk checksum don't match. Check with sha1sum on the vmdk.

[cyruspy]

OK second attempt worked. I deleted the SATA controller that I was not using and it appears it went as planned.

Also the UUID is regenerated on import and the fact that they are not the same does not matter (AFAIK). The DEVs would need to answer that one.
So the question I have is did you receive the same error I did and what are the actual settings of the VM? Did you try to select the legacy .9 or just the default?
I will of course continue to investigate this but your input would help.

As you can see there are a total of 3 different UUIDs on this working import.

DS2-Import.jpg

In my case, the UUID on the disk definition and the storage attachment match. It's the one from the original VDI. It didn't complain about the VMDK checksum in my tests.

Changing the UUID on the OVF to match the VMDK UUID makes the import successful.

I tested default OVF format only (not 0.9)

Bug report with error screenshot and OVF files: http://www.virtualbox.org/ticket/6802

[humdinger70]

3.1.8 works. 3.2.0 has a mess of problems. OK, what did you break??