Linux Mint Debian and Win7 guests lose and cannot reacquire mouse and keyboard capture

[Karmic]

Here's the relevant platform information:

VirtualBox version:
Version 5.1.22 r115126 (Qt5.3.2)
VirtualBox 5.1.22 Oracle VM VirtualBox Extension Pack

Host platform:
Linux Mint Debian

Guest platforms:
Windows 7
Linux Mint Debian (same as host)
(Guest extensions have been updated on both)

Output of "uname -a" on host:
Linux 3.16.0-4-amd64 #1 SMP Debian 3.16.43-2 (2017-04-30) x86_64 GNU/Linux

Backround information:
I'm running two different guest virtual machines simultaneously on my Linux host as noted above. I have configured both guest virtual machines to automatically capture and release the mouse and keyboard when the guest window comes into and goes out of focus.

After upgrading to 5.1.22, I have observed several instances where one or the other guest virtual machines permanently lose the mouse and keyboard capture. When this happens the VirtualBox application menu and 'x' in the upper right hand corner of the window become unresponsive. Judging by the type and color of the mouse and keyboard guest status icons in the lower right hand corner of the window, mouse integration (MI) is enabled and the keyboard is not captured. Hovering with the mouse over these icons does not produce the typical informational popup window.

In addition, the guest window does not respond to host key commands such as Host+h to go full screen. I am able to minimize and change the size of the application window. The guest continues to run (i.e. cursors are still blinking, etc.) but there is no response from the mouse or keyboard. The only recourse is to kill the guest virtual machine from a host terminal.

I upgraded VirtualBox from version 5.1.18 to the current version noted above. I have not observed this behavior in versions equal and prior to version 5.1.18.

I know that the VirtualBox logs could potentially provide additional clues as to what is happening but I do not know how to capture them when this occurs or even where to find them on the host.

Any ideas as to what might be happening?
Thanks in advance.

[halfervirt]

This sounds much like a problem I am having.

Host: Linux Mint 17 Cinnamon
Guest: Linux Mint 18 Mate with latest Guest Additions
Also same behaviour with Ubuntu 14.04 LTS guest
VirtualBox: 5.1.22

This behaviour has been exhibited for me in the above guest systems, in various versions of VirtualBox 5.1.x. I have several times uninstalled and gone back to 5.0.x, but I am conscious that eventually 5.0.x support will cease. I am therefore biting the bullet and hoping I can help developers discover the bug so it can be included in a future 5.1.x release. I am aware that it is going to be hard to pin down, since it is very unpredictable as to when it occurs.

For me, the mouse capture from the guest stops working, and I cannot click on anything in the guest. The keyboard still works, and the guest is still running. I can minimise the full-screened guest using the Ctrl-Right-M, and I can toggle full-screen using Ctrl-Right-F. Unfortunately VB is still hanging onto the capture, so if I do minimise the window, I cannot click on anything in the host either. I have to use Alt-Tab, Alt-Esc and keyboard shortcuts to use either the host or the guest. If am lucky enough to be able to get the right keyboard shortcuts to get back into the guest, restarting the machine using sudo shutdown -r now does not alleviate the situation, but sudo shutdown now helps, and everything is back to normal, and I can then restart the guest.

This phenomena probably happens once or twice a day. Can someone advise what logs would be useful in discovering why this happens? I will look for a related ticket now, but if anyone knows of related tickets, I would be grateful for links.

[socratis]

I haven't seen any related tickets, but don't take my word for it, I may have missed it, I usually glance over them to see the more active ones.

You need to find a pattern for when that happens. Full screen? Weird/unusual screen sizes, not from within the guest but from dragging/resizing the window? Window manager? Kernel? X11? Widgets? Anything else?

I'm not on Linux myself, so I can't actively contribute to this, I'm simply throwing ideas what to look for, trying to find a common denominator. And most certainly, the VBox.log for a session when that happens. ZIPed.

[halfervirt]

And most certainly, the VBox.log for a session when that happens. ZIPed.

OK, thanks for your advice. I've also searched the tickets and can't find anything that feels related. If When it happens again I will raise a new ticket and will supply the log file at that point.

I can't see from the Downloads page when 5.0.x support will cease. You don't happen to know, do you? I am considering going back again, but not if it will be dropped in the next month or so - I have to bite the bullet at some point!

[socratis]

I can't see from the Downloads page when 5.0.x support will cease. You don't happen to know, do you?

Not the slightest idea. The developers do not mention "details" like that, which IMHO, they should. The product lifetime is something that companies usually make public...

BTW, I'm going back and forth between 5.1.x and 5.0.x (even 4.3.x) all the time, it takes two minutes to switch...

[halfervirt]

BTW, I'm going back and forth between 5.1.x and 5.0.x (even 4.3.x) all the time, it takes two minutes to switch...

Interesting. Out of curiosity, what makes you switch so frequently? Are you getting bugs in the later versions as well?

Since I've just had another crash of the type I referred to, I am going to file a report now. If I get another one by the end of the day, I'm going back to 5.0 until support finishes.

[socratis]

Out of curiosity, what makes you switch so frequently? Are you getting bugs in the later versions as well?

Well, bugs exist in every version. And if it is a bug, when did it start happening? What was the behavior before that? I try to avoid saying "well, this started happening a while back, I don't remember when...". If another user says that, I'm working with them to see when it actually started happening. Trying to see whether it's a regression or not helps a lot, mainly the developers. And a helped developer is a happier developer. They can target the bug easier, faster.

I usually run the latest test builds. I had a bug once in between 5.1.10 and 5.1.12 (#16260). I pinpointed the bug between specific revisions of the build. It was fixed in between the official releases and never "released" wider to the public. That made me feel that I was really helping

[halfervirt]

I hear your plea. I am a software engineer too, and yes, helpful bug reports are great. I really do try to be as detailed as possible. However, I am pressed for time, and struggle with productivity at the best of times, without my core operating systems also causing me hours worth of extra work.

I will try to work with a dev to get to the bottom of this one. Frank on the bug tracker has been really good in the past, even with the worst of intermittent behaviours.