20% clock drift in the guest

[meixner]

The clock of my guest runs at 80% of the speed of the host clock: Invoking date and 60s later invoking it again, only 48s have passed in the guest.

Host: Windows 7 Professional 64bit Guest: Linux Mint 17.1 64bit

[meixner]

A CentOS 6.x (32 bit) guest on the same host does not show this clock drift.

[meixner]

Log of the CentOS 6.x guest

[Frank Mehnert]

Sorry, 4 phyiscal host cores but 8 VCPUs configured, that will not reliably work.

[meixner]

I switched back to 4 VCPUs, without any difference. Still losing 12s per minute.

[meixner]

After switching to 4 VCPUs

[meixner]

So what next? Even with the updated configuration the bug is still there, but the ticket has been closed and the resolution set to invalid. Do I have to open a new ticket?

[Frank Mehnert]

Re-opened the ticket for you :-) The CPU measurement on the host is somehow incorrect. It's incorrect for the CentOS 6 guest as well but has apparently no effect there.

[meixner]

BTW. CPU is Core i7-6700 with 3.40GHz

[Frank Mehnert]

I know, Skylake, that's visible from the log as well.

[Frank Mehnert]

You might want to try the following workaround:

VBoxManage setextradata VM_NAME VBoxInternal/
TM/TSCTicksPerSecond 3400000000

where VM_NAME is the VM you attempt to start.

[meixner]

Replying to frank:

You might want to try the following workaround:

VBoxManage setextradata VM_NAME VBoxInternal/
TM/TSCTicksPerSecond 3400000000

where VM_NAME is the VM you attempt to start.

Yes, that seems to fix the issue.