[vbox-dev] disk I/O crashes of FreeBSD 9.1-P7 guest on Mac OS X 10.8.5 host
Klaus Espenlaub
klaus.espenlaub at oracle.com
Mon Sep 23 10:08:27 GMT 2013
Hi Axel,
On 22.09.2013 13:38, Axel Rau wrote:
> Hi,
>
> This VM exists 6 month now and worked perfect.
> Recently I upgraded the host from 10.8.4 to 10.8.5.
> Today, I started the VM after a break of 2 month after upgrading VBox to 4.2.18r88780.
> After that, I updated freebsd 9.1 to 9.1-P7 using freebsd-update.
> Everything worked perfect so far.
> Then I did a portsnap fetch and a portsnap update, which froze the VM immediately (lots of directories/file had to be removed).
> After a few seconds the VM started rebooting but was unable to find the root disk (a SATA drive), until I powered down and restarted.
> This was reproducible.
> After that I switched the disks to a IDE controller: Same crash on portsnap update. Only difference: It completed its reboot.
Sounds like the host is not able to handle the I/O within the timeout
imposed by the guest OS. OSX is known for bad I/O performance, it is
easily overloaded, resulting in unbearable response time for I/O
requests (we've seen a 64K write request taking 3 minutes, and there are
probably worse situations). There is very little VirtualBox can do about
this.
> Nothing interesting in the logs.
Which logs? Host (normal OSX logging and VBox.log) and guest logs?
> This is a 6-core Xeon, 2 cores dedicated to the VM.
Sounds unlikely that you have dedicated cores for the VM... guess you
simply set the VCPU count to 2.
> Questions: Has this combination of host/guest OS versions been tested?
Probably not - there is a near infinite number of combinations, and you
use a quite unusual one.
> What can I do to further help finding the cause?
Clarifying which log looks normal would provide a starting point.
Klaus
>
> Thanks, Axel
> ---
> PGP-Key:29E99DD6 ☀ +49 151 2300 9283 ☀ computing @ chaos claudius
More information about the vbox-dev
mailing list