[fedora-virt] Virtio-win Windows NIC goes offline during bulk copies

[Greg Scott]

Help!

 

Long story short - I need to do a bulk copy of 200+ GB from a physical
Windows server to a virtual Windows server hosted on a RHEL 6.1 host
with libvirt.   The guest Windows VM is running Windows Server 2008R2
with two virtio NICs.  One virtio NIC is connected to RHEL bridge br0,
the other to RHEL bridge br1.  Two virtual NICs because they will
eventually connect to two different networks.  

 

So I am trying to do a bulk copy of 200+ GB from a physical Windows
server to this virtual machine.  At random times, generally after a few
dozen or so GB - but sometimes after more than 100 GB - the virtual NIC
goes offline.  From the Windows VM, I am unable to ping anything,
including the RHEL host it's running under.  

 

No, it's not a firewall rule.  The only workaround I can find is, inside
the Windows VM, disable the virtio NIC and then re-enable it.
Essentially, reload the driver from Windows' point of view.  Then it's
good for a while and everyone can ping everyone for a while again.  

 

I am running this bulk copy through a gb switch.  I've tried a couple of
different switches, but no change in symptoms.  Plus, even if the
switches are flakey, I should not have to keep reloading that virtual
NIC driver.

 

Any advice appreciated.

 

Thanks

 

-          Greg Scott

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/virt/attachments/20110902/1a3afb5c/attachment.html