VB freezes with linux guest

[seattle vic]

Version 5.1.26 r117224 (Qt5.7.1), Ubuntu Mate 17.04 host and guest. After an hour or two it freezes and I can no longer ssh into the host (I'm running headless). The host syslog log files indicate that the VM was powered down.
The hardware is a new Dell I7 desktop. Next time this happens I'll hook up a monitor to see if the host is still responsive, but for now any clues would be appreciated. I'm going to try and attach the guest log file.

[seattle vic]

This time it crashed within an hour. Before I restarted it I disabled audio as I saw that suggestion in another crash thread. I hooked up a monitor and everything was frozen so I had to reboot the host. I'm also continuously running top and see no excessive CPU hogging from virtualbox. I've also attached the syslog from the host. Ignore all the UFW stuff as it's a windows 10 machine doing constant broadcasts. In the logs, things appear to start happening around 11:54. Messages about specific CPU's and things like: "NMI watchdog: BUG: soft lockup - CPU#1 stuck for 22s! [kworker/1:0:5421]" etc..

[Perryg]

I don't see any outstanding reason for this from your log or host syslog. I would like to see the syslog and the dmesg log from the guest the next time this happens though.

[seattle vic]

I froze about an hour later. One thing that I did see in the host syslog repeatedly that I had not seen in a couple days is:

The canary thread is apparently starving. Taking action.
Aug 12 14:09:46 backup3 rtkit-daemon[3644]: Demoting known real-time threads.
Aug 12 14:09:46 backup3 rtkit-daemon[3644]: Successfully demoted thread 3643 of process 3643 (n/a).

Here is the guest syslog. There is no dmesg log file created in /var/log:

[seattle vic]

In looking at the logs, it appears that the processor and software are still running even after the guest dies, I can't ssh to the host and the X screen is frozen.

Top does now show a debilitating CPU load at the end. I always thought that linux was pretty much immune to an app crashing the system but that's what's happening here to the host. Just a thought.

[Perryg]

I think that what you are seeing is a red haring but not sure yet. What again is causing all the flooding of the host and guest syslog? I also wanted to see the guests dmesg log.

[seattle vic]

The host just froze again, though I'm still connected and looking at Top and syslog via ssh (though I cannot ssh in). I see that there is a kworker process at 100%, which I assume to be kernel work. There are also a ton of kernel messages I see on syslog (using tail -f). I'm going to reboot and see if there is a dmesg log file in the guest /var/log. Didn't see one before.

[seattle vic]

I think that what you are seeing is a red haring but not sure yet. What again is causing all the flooding of the host and guest syslog? I also wanted to see the guests dmesg log.

It's a windows 10 machine on the network that does constant multicast/broadcasts, I think for discovery purposes. I thought it was odd as well and at first figured it was a virus but it appears to be normal after some googling. I have the exact same setup on another machine (different hardware, an ASUS i5 desktop, about 2 years old), but with the same ubuntu mate host and ubuntu mate guest and it's been running fine. The only other difference I can think of is maybe the graphics card. The Asus has a standard on board graphics where the newer Dell that I'm having the problem with is a separate higher powered card.

[seattle vic]

Solved. In looking at the syslog, things started unraveling after the message: "NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [ATA-1:15492]"

Some googling revealed a thread that it was happening to a Dell computer with an Nvidia graphics card and that they needed to install the latest Nvidia driver.
Likely a coincidence that my unit is a Dell but I checked my driver and it was the standard xorg nouveau driver. I installed the latest 375.66 driver and everything seems to be fine after 3 hours.
QED - Not

[seattle vic]

PS - How do you mark a thread as solved?