VB always aborting.

Discussions related to using VirtualBox on Linux hosts.
vista killer
Posts: 9
Joined: 11. Sep 2007, 14:00

Post by vista killer »

I have upgrade my system from feisty to edgy and it seems tha there is no more abort problem with virtual box.I have use it for hours and is very stable.
freimann
Posts: 14
Joined: 8. Sep 2007, 09:44

latest kernels as crashes culprit ??

Post by freimann »

Hi !
I also had many crashes after upgrading my Slackware box to the latest version - 12.0 - with kernel 2.6.21.5 ... I have been wrestling with these crashes for some two weeks but finally gave up and recalled someone mentioning that kernel upgrade from 2.6.18 to 2.6.19 started the crashes. I have already played with 2.6.18 a little because of XEN so now I have just compiled a pristine 2.6.18 - and !! - no more problems !! even my logitech webcam works fine inside virtual win-XP using isochronous USB xfers ... great for skype video ...
sitor
Posts: 20
Joined: 28. Sep 2007, 13:20
Contact:

Post by sitor »

Could this not all be the tickless timer problem as mentioned on the FAQ (http://virtualbox.org/wiki/User_FAQ:

Code: Select all

Tickless Linux timer: VirtualBox currently does not work well on Linux kernels with tickless timer support (CONFIG_NO_HZ enabled). We are investigating the problem but in the meantime you should disable the tickless timer by disabling that config option or by adding nohz=off to the Linux kernel command line.
In that case the reverting to an older kernel that has not implemented the tickless timer would indeed solve the problem, but using the current kernel with tickless timer disabled would as well.

I'm having the same problem on a Mandriva 2008 RC2, but I don't know how to disable the tickless timer option in the kernel.

Ciao,

Sitor
freimann
Posts: 14
Joined: 8. Sep 2007, 09:44

tickless Linux kernel ...

Post by freimann »

well - i will test if disabling tickless timer helps ... you can disable it either by setting a config option and recompiling your kernel - OR - by appending this string "nohz=off" at the end of kernel command line (in you lilo.conf or menu.lst for grub)
freimann
Posts: 14
Joined: 8. Sep 2007, 09:44

disabling tickless kernel does not prevent crashes ..

Post by freimann »

ok - so going back to 2.6.21-smp on Slack 12.0 with "nohz=off" command line did not help - it took about 20 minutes for the virtual win-xp machine to crash :-((
sitor
Posts: 20
Joined: 28. Sep 2007, 13:20
Contact:

Post by sitor »

:cry:

I don't think that Mandriva has Virtualbox kernel drivers for older kernels in 2008, and I need 2008 for Belgian eID support.

So my only hope left now is on the 1.5.2 version of Virtualbox.

Otherwise I will either have to stay on Windows as main OS, or move to VMWare server.

Would be a pity.

Ciao,

Sitor
sitor
Posts: 20
Joined: 28. Sep 2007, 13:20
Contact:

Post by sitor »

And it just crashed again. Oh, well. Good that I didn't get rid of the physical multi-boot Windows. So I guess that I'll still be stuck in Windows for some time :evil:
Let's hope that 1.5.2 does bring a fix to the problem and that it doesn't take long anymore to become available.

Ciao,

Sitor
holvenstot
Posts: 12
Joined: 15. Sep 2007, 04:58

I Believe

Post by holvenstot »

[quote="sitor"]:

I don't think that Mandriva has Virtualbox kernel drivers for older kernels in 2008, and I need 2008 for Belgian eID support.


Have you tried to build the driver - I do a lot of kernel test work and with each new kernel guess what - its time to rebuild the driver: As root:

/etc/init.d/vboxdrv setup

Of course, on your mandriva system you may have to look around a little for the vboxdrv script.

Falling back to the 2.6.18 series of kernels is not really an option - at least not for me. None of my SATA hard drives work at the 2.6.18 level - my motherboard uses the NVIDIA chipset - condolences not required - results may vary with other chipsets.

I have been hearing about the 1.5.2 release and how it is going to fix this for a while now - does anyone have a target date and is the beta source available for download so I can at least try it out?

Have a great day
holvenstot
Posts: 12
Joined: 15. Sep 2007, 04:58

Another thing ...

Post by holvenstot »

Just in case anyone is playing around looking at options here let me tell you another one which was NOT a solution for me.

I used to run VirtualBox 1.3.8 on my Ubuntu 7.04 system - I think that it had a 2.6.16 series kernel - but I am getting old and can't remember back that far.

But 1.3.8 was a stable as a rock in that environment - so I said to myself lets drop back to 1.3.8 - the clipboard is nice but you can't use it to copy data onto an aborted system.

I found that 1.3.8 running on a 2.6.23 or 2.6.22 series kernel aborts just as much as the 1.5 version of VirtualBox.

So at the least, I think that it would be safe to say that this is not a problem introduced into 1.5.0 but rather the failure to adapt to some change made in the kernel.

Just a thought
holvenstot
Posts: 12
Joined: 15. Sep 2007, 04:58

Not a solution

Post by holvenstot »

Based on the thought expressed in a previous post that since 1.3.8 - which used to be rock solid, failed as frequently as 1.5.0 when run on a "current" kernel I sat down last night and built a fresh kernel (2.6.23-rc9) with CONFIG_NO_HZ turned off

At this point I removed the instance of VirtualBox 1.5.0 which was installed on my system - leaving behind only the virtual disks containing my XP image.

I then grabbed the source tarball for VirtualBox-OSE 1.5.0 and after putting together the required toolchain I did a build and when complete I defined a "machine" using my previously installed instance of Windows XP and brought it up without any problems.

I then setup a long running task (re-encoding some large mpeg files with Pinnacle's Studio 10) and went to bed.

When I woke up 5 hours later my virtual XP system was still up and running - since I had previously considered myself to be lucky to get even an hour of uptime before having the system abort I thought I might be on to something.

Alas, after running a few more tasks it aborted an hour later. So this was NOT the solution, or at least the complete solution - I say complete solution because I have been unable to locate enough information to determine if all these aborts have the same root cause.

Could it be that we are dealing with multiple issues here causing the aborts, and that rebuilding VirtualBox from its source solved some of those issues? I don't know.

Since it has been a LONG time since I got anywhere near six hours of uptime before getting an abort just to see if it continues to run longer between aborts - i.e.: have I made any progress or am I blowing smoke?
sitor
Posts: 20
Joined: 28. Sep 2007, 13:20
Contact:

Post by sitor »

For the moment I give up => back in Windows as main OS (at work at least).

First of all I'll wait for the final release of Mandriva 2008. And when that is live I'll see what to do. If Virtualbox 1.5.2 is out by then, I'll give it a try and see if it does any better then the current version. If not, then Virtualbox is useless to me and I'll use VMWare server (which is proven and should work without problems).

Ciao,

Sitor
holvenstot
Posts: 12
Joined: 15. Sep 2007, 04:58

I feel you ...

Post by holvenstot »

Sitor -

I know how you are feeling - while I have not given up on VirtualBox I did drop an older Sony laptop back to Windows so that I could perform some Windows specific tasks I need for work.

I guess what bothers me most here is not the problem, all systems have problems from time to time, but the seeming lack of communication from the developers on this issue - where are they, what can we do minimize the issue, what fixes we can try.

Or even when 1.5.2 is expected out.

Virtual box is one heck of a product, and before all this happened, it had a good track record (at least with me) but if something is not done soon, including improving communications, there will be a mass migration to another solution.
jbeiter
Posts: 12
Joined: 14. Sep 2007, 17:09

Re: I feel you ...

Post by jbeiter »

holvenstot wrote: I guess what bothers me most here is not the problem, all systems have problems from time to time, but the seeming lack of communication from the developers on this issue - where are they, what can we do minimize the issue, what fixes we can try.
This is bothering me too. Even some acknowledgment that the developers know there is a problem would at least give the hope it might be fixed in the next release.
I don't know if virtualbox developers comment on any other threads but I was expecting at least a bone or two of things to check in a thread like this.
freimann
Posts: 14
Joined: 8. Sep 2007, 09:44

disabled hyperthreading helps with crashes ...

Post by freimann »

after 24 hours on my Slack 12.0 box with Pentium-4 and HT disabled I am still running without any crash - having two VM - Win-XP and Win-2000 running all the time - doing heavy graphics editing in Adobe Photoshop & website editing in Dreamweaver (512 Mb VM) and with W2K running MS Exchange 2003 ...
sitor
Posts: 20
Joined: 28. Sep 2007, 13:20
Contact:

Post by sitor »

Freimann,

What do you mean with HT disabled? What is HT?

Ciao,

Sitor
Post Reply