virtualbox.org

I'm struggling to get a CentOS minimal install (64-bit) to boot reliably. Periodically it successfully boots, but most of the time (99 out of 100 times or so), it does not boot. It fails with the message "Switching to clocksource tsc".

I managed to create a VM, install CentOS version 7, along with a GNOME GUI. The install completed successfully and I could install additional software. I could also restart the guest OS and it would reboot successfully every time. The install was problematic in the same way - booting the VM with the Centos iso image often hung at "Switching to clocksource tsc".

I installed the Guest additions (v4.3.14, not 4.3.12) after several attempts as the kernel build was failing. That I fixed and the install completed, although the OpenGL support module failed. I did try installing the guest additions for 4.3.12, but seeing a post about how 4.3.14 might solve my kernel issues, I tried that instead and it did indeed appear to install better. At the completion of the Guest Additions install, I rebooted the VM and it successfully booted. A few minutes later, I try rebooting the VM and it fails almost always.

I say "almost always" as I had left it powered down for several hours this morning after not being able to get it started on multiple attempts and it started perfectly. But as soon as I try to repeat the power down and start of the VM, it fails to start. (On one occassion, I changed the memory from 1024 MB to 2048 MB and it booted. But after that, the size of memory made no difference. I suspect it was a coincidence that it happened to boot that time when I changed the memory size for the VM).

I am running VirtualBox 4.3.12_OSEr93733 on a FreeNAS host (version 9.3).

I attach my VBox.log from a recent boot attempt.

Any ideas how I could get my Guest OS to boot every time? I feel I'm so close, but something is not quite right. Any guidance would be much appreciated.

Thanks!

00:00:00.011632 Host RAM: 16040MB total, 2595MB available
00:00:00.084685   RamSize         <integer> = 0x0000000080000000 (2 147 483 648, 2 048 MB)
00:00:00.084956   VRamSize         <integer> = 0x0000000000c00000 (12 582 912, 12 MB)

You're allocating memory that's awfully close to the available memory on your host. Your host is not going to breathe easily. Either shut down unwanted applications on your host, or reduce the memory requirements of your guest.

I changed it down to 1024 MB, well above the requirements for CentOS, but that makes no difference. A new iso install and the installed operating system both still hang on boot.

In any case, my FreeNAS server is running fine. It's the guest VMs that are showing no signs of life...

Since you are running *BSD you would get better help from them since this is their build and not ours.

In case anyone is looking for how this story continues, just posted a topic in the FreeNAS forums:
https://forums.freenas.org/index.php?th ... tsc.40239/

Strange part is that in about 1% of cases, a VM does boot, letting me install CentOS, GNOME GUI and Crashplan. If I restart the VM, it won't boot, hanging at the same message.

One thing that I didn't notice at the beginning but I noticed from your other post, is that you have set the guest as 'Other_64:

00:00:00.068030 Guest OS type: 'Linux_64'

In reality, CentOS should have the 'RedHat' template applied to it. Please change it, but before you do, zip and post the '/home/vbox/VirtualBox VMs/CentOS-3/CentOS-3.vbox' file (the "recipe" for the VM). There we can see the changes (if any) between your recipe and the template.

Oh, and I do not think that the problem is due to the "Switching to clocksource tsc" error message...

First off, I attach the vbox for the CentOS VM that has installed CentOS, Gnome and Crashplan, but that fails to start.

Now, testing the use of the RedHat-64 bit template, I created 3 VMs, all with RH-64bit, 1024 MB RAM, each with it's own 256GB VDI and bridged networking. The first booted and started to install CentOS. That looked promising. So I powered off the VM and started it again to check that it restarts. No joy. It hangs on the clocksource message.

Created a second VM. That too booted straight into CentOS install. This is ongoing right now. Once it is installed, I'll power it down and see if it restarts.

I then created a third VM. That fails to install with the boot hanging at the clocksource message.

Two VMs booting within the space of 5 mins is a record so that looks promising, but something is still amiss.

Do you see anything perculiar in the .vbox file?

Btw, from what I have read so far, I agree that the clocksource may well not be the issue. It's just that that is the visible symptom.

I have also tried to change the template for my successfully installed jail to RedHat 64 bit. It still hangs at the clocksource message.

One thing I noticed when my RedHat templates lead to successful boots: I see green OK messages at the start of most rows during the boot in the time before the clocksource message. On the occasions when the VM doesn't boot, there is not a single green OK message on any lines as it quickly reaches the clocksource message. Might be irrelevant, but may not...

CentOS 7 install completed. I then ACPI shutdown the VM and then pressed start. The boot hangs on the clocksource message.

It would appear that the RedHat 64-bit template alone is not a solution, but perhaps it helped.

Are you still using the BSD fork? If yes then your updates don't belong here.

I am still using the VirtualBox on FreeNAS so I assume that is "the BSD fork".

I have the topic open with the FreeNAS community. I only posted here in response to others.