VirtualBox v3.0.X system lock up issue

[randshuntzinger]

So, ummmm, is it safe to use VirtualBox v3.0.8 now?

I hope some of you managed to get a week of uptime under heavy file/network I/O...it would be good news!

I managed to get a week of uptime under the 3.0.7 test release with 4 guests running on a host. I didn't have high I/O though. On the positive side, with 4 guests running I didn't manage a day uptime with 3.0.0 to 3.0.6 so things are definitely better. Are the completely fixed? I don't know - time will tell.

[ljw1]

Are your guests set to single or multiprocessor? In all the versions from 3.0.6 back as soon as multiprocessor for the guest is enabled they lock up very quickly and take down the host. I don't really want to test this version and hard lock the host again.

[sportq]

So, ummmm, is it safe to use VirtualBox v3.0.8 now?

I hope some of you managed to get a week of uptime under heavy file/network I/O...it would be good news!

Well, prior to 3.0.8 I would get about an hour before a hard lock up (see my post above). I installed 3.0.8 yesterday, rebooted to make sure the kernel modules were the new ones and hit it hard with several VM's doing some (much needed) OS updates simultaneously and touch wood, it's been up for 26 hours now. Even the guaranteed-to-crash-it Windows Home Server instance has been stable.

I'll post if anything bad happens.

-Pete

[rpro]

Thanks Pete for testing it out!

[Soupy]

I'm surprised that there isn't more activity from people trying out 3.0.8, given how much frenzy there was over the issues with the previous versions. Anyone else have status to report?

[ljw1]

After all the problems I am not keen to keep testing versions and crashing my server. Probably others are having the same thoughts. Although in single cpu 3.0.8 seems to be working ok.

[arantala]

A lot of good work has been done. VirtualBox seems a lot more stable now on 3.0.8.
My 8 gig copy test passed many times properly without crashes. Unfortunately, just as I was ready to uncork the champagne, it froze again. I had a lot of traffic and disk activity going back and forth. I'm going to enable core dumps and see if I can reproduce.

[randshuntzinger]

Are your guests set to single or multiprocessor? In all the versions from 3.0.6 back as soon as multiprocessor for the guest is enabled they lock up very quickly and take down the host. I don't really want to test this version and hard lock the host again.

Sorry I didn't respond sooner. I'm using single-CPU mode. I've not tested multiple CPU because VirtualBox doen't want to configure it in my environment. It is still good news because with 3.0.6 and earlier it crashed frequently in single-cpu mode.

-- Rand

[half12]

Hi,

Having tested v3.0.7 + patch for a while, I switched to v3.0.8 a week ago and not had any system lock ups, even while running three VBox and downloading a Solaris 10u8 DVD iso file, something that caused problems previously. The earlier versions of v3.0.X certainly made me cautious about upgrading from v2.2.4, it would not be reasonable to say they should have been released only as beta versions.

[steinar]

Have now been running 3.0.8 with debian, windows home server and freenas guests for about one week without issues. All guests were single cpu, VTx enabled, host internal network disabled. Reconfigured debian guest to dual cpu for about one day ago. Stability remins to see.

Host: Opensolaris 2009.06 on Intel Core 2 Quad Q9400, Gigabyte EG45M-DS2H, 4 GB RAM.

This uptime was science fiction on 3.0.6, which would hang the host completely or boot with kernel panic after between 24 to 48 hours.

[zebra]

Hi all.

Well, bad news ahoy. After thinking we were home free, with 3.0.8, I'm still observing the same sad lock-up behaviour with my OpenSolaris b125 host + Win2k8 SP2 guest, with both single and multiple CPU's allocated.

Granted, it took a week before the whole system locked up this time, rather than a day...but it still locks up .

Getting there. Getting there...

Keep up the good work. I'm sure it'll be worked out, in the long run!

Cheers.

z

[tschoenemann]

For us all is good now.

10 days without any problem on sol10u7 , sun fire x4140.
20 VMs: some linux, some os2 some win (xp, 2000, 2003, nt4). Two running SMP.

thomas

[steinar]

Have now been running 3.0.8 with debian, windows home server and freenas guests for about one week without issues. All guests were single cpu, VTx enabled, host internal network disabled. Reconfigured debian guest to dual cpu for about one day ago. Stability remins to see.

Host: Opensolaris 2009.06 on Intel Core 2 Quad Q9400, Gigabyte EG45M-DS2H, 4 GB RAM.

This uptime was science fiction on 3.0.6, which would hang the host completely or boot with kernel panic after between 24 to 48 hours.

Still running after 'round two weeks:) Had one incident where planned reboot of my debian guest would drop all network interfaces, both host and guests. Took about 2 minutes and network was up again.

[dri]

Is this issue considered closed? I don't want to upgrade my 2.2.4 just to revert back tomorrow. Running OpenSolaris 200906 on AMD-V.

[ljw1]

No this issue is still not resolved two hard locks since updating to 3.0.10. It is pretty frustrating as there appears to be no progress in solving this problem for opensolaris.