Periodic catastrophic failures of VB under macOS Big Sur

Discussions related to using VirtualBox on Mac OS X hosts.
Post Reply
fabriziogiudici
Posts: 4
Joined: 15. Oct 2021, 17:01
Primary OS: Mac OS X other
VBox Version: OSE other
Guest OSses: Linux, Windows, macOS
Location: Genoa, Milan (Italy)
Contact:

Periodic catastrophic failures of VB under macOS Big Sur

Post by fabriziogiudici »

Hello.

The problem is so strange I feel even unable to describe it. Anyway I'm trying...

I've been using VirtualBox on macOS for I think a decade without ever seeing something like that. My latest laptop is a MacBook Pro 16", bought in February 2021. No problems for months, even though I didn't use VB extensively (but I did use it). Everything started three weeks ago, when all VMs (all Ubuntu 20) started to have troubles: they booted, but soon showed failures such as crashes etc. Reinstalling VB from scratch didn't help. Having also VMWare Fusion and a tight deadline I used Fusion to deliver my duties in time, postponing the diagnosis of the problem.

Then I started working with a Vagrant project that requires VB. I expected to deal with the problem again, but strangely enough everything was fine and also the VMs that previously used to fail were ok.

I've been working extensively (several hours per day) on the new project for ten days with no problems at all. Then, all of a sudden, the bug showed up again. This time VB crashes (with different kinds of core dumps, from the classic segmentation fault to failing code signatures) immediately, even before the guest boots (with a few exceptions, that anyway don't complete the boot).

That's why I've written "periodic" and not "intermittent", since it seems it has good and bad periods of days.

Honestly the first thing I'd think about such a problem would be hardware failure (especially the failing code signature smells about it), possibly the disk. But:

+ no other application is failing (in general: LibreOffice, photo editing tools, Java stuff, IDEs, Firefox... Even VMWare Fusion can boot without problems a number of Linux guests);
+ I have about 1TB of RAW photos on the same SSD - they are subjected to non-destructive processing, so never change, and I have a tool that periodically checks the MD5 of the files - so they are deeply tested for integrity. No problems detected. An SSD problem should likely appear in other parts of the file system. Consider that with Vagrant I'm constantly destroying and recreating VMs, so they shouldn't affect the same SSD area.
+ A tool that checks S.M.A.R.T. status show no faults.

macOS Big Sur 11.5.1 + VirtualBox 6.1.26.
scottgus1
Site Moderator
Posts: 20965
Joined: 30. Dec 2009, 20:14
Primary OS: MS Windows 10
VBox Version: PUEL
Guest OSses: Windows, Linux

Re: Periodic catastrophic failures of VB under macOS Big Sur

Post by scottgus1 »

We'd need to know the exact error message the next time Virtualbox crashes on you.

Note that Vagrant uses Virtualbox in ways we don't grok, so it would have to be a Virtualbox-only session, not a Vagrant-controlled session.
fabriziogiudici
Posts: 4
Joined: 15. Oct 2021, 17:01
Primary OS: Mac OS X other
VBox Version: OSE other
Guest OSses: Linux, Windows, macOS
Location: Genoa, Milan (Italy)
Contact:

Re: Periodic catastrophic failures of VB under macOS Big Sur

Post by fabriziogiudici »

Here it is: a bunch of crash logs created yesterday.
In the meantime I tried for the n-th time (without any specific rationale) to reinstall VirtualBox. Now VirtualBox doesn't crash, but the guest Linux VMs do a little bit before or after completing the boot.
Attachments
Archive.zip
(155.35 KiB) Downloaded 6 times
fabriziogiudici
Posts: 4
Joined: 15. Oct 2021, 17:01
Primary OS: Mac OS X other
VBox Version: OSE other
Guest OSses: Linux, Windows, macOS
Location: Genoa, Milan (Italy)
Contact:

Re: Periodic catastrophic failures of VB under macOS Big Sur

Post by fabriziogiudici »

Follow up. In the past days I've seen no more crashes on VirtualBox, but guest machines still keep crashing. A simpler one (Ubuntu Server, no desktop) seemed to survive enough to the point of being able to log in, but it showed other malfunctions, such as wrong checksums during an apt update. This seems to me memory corruption or such.
In the meantime I was able to configure Vagrant to use VMWare Fusion and I'm going that way to respect my deadline. I'm still absolutely interested in fixing the VirtualBox issue - just to say that with Fusion there are no issues (as there are no issues with other applications), so I get this as a confirmation that there are no hardware failures.

PS I even tried to run VirtualBox from a guest inside Fusion by enabling VT-x/EPT. VirtualBox still has problem inside Fusion.
paulmc
Posts: 72
Joined: 28. Aug 2019, 18:43
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Mac OS X, Linux, Windows
Location: Earth (Guyana / USA / South Africa)
Contact:

Re: Periodic catastrophic failures of VB under macOS Big Sur

Post by paulmc »

@fabriziogiudici, just a quick shot in the dark here, but is it at all possible that third-party enhancement software somehow got accidentally installed, that doesn't play nicely with VirtualBox? E.g., several threads in this forum advise against the Intel Power Gadget {a kernel extension with bundle-id 'com.intel.driver.EnergyDriver'}; anything perhaps along those lines?
fabriziogiudici
Posts: 4
Joined: 15. Oct 2021, 17:01
Primary OS: Mac OS X other
VBox Version: OSE other
Guest OSses: Linux, Windows, macOS
Location: Genoa, Milan (Italy)
Contact:

Re: Periodic catastrophic failures of VB under macOS Big Sur

Post by fabriziogiudici »

It's not really a shot in the dark: I have Intel Power Gadget installed. The strange thing is that it has been installed since the beginning (when I got my latest laptop), that is February, and VirtualBox has worked since then, until the incident I described. In any case it could be non deterministic, so it's definitely something to test. I even started a comparative test this night, but - LOL - at the moment VirtualBox worked perfectly again. I'll wait until the next time it crashes, or crashes the guest, then I'll remove Intel Power Gadget and see whether it change things.

Thanks.
Post Reply