[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

AW: [Xen-users] Netback grant fails


  • To: <xen-users@xxxxxxxxxxxxxxxxxxx>
  • From: Leo Krüger <leo.krueger@xxxxxxxxxxxxx>
  • Date: Thu, 23 Jul 2009 20:50:00 +0200
  • Delivery-date: Thu, 23 Jul 2009 11:51:01 -0700
  • List-id: Xen user discussion <xen-users.lists.xensource.com>
  • Thread-index: AcoLqm+1kTTp/MNcR9W49IooGmGDeQAGbMxw

Hello,

this might be a problem with interrupt sharing between virtual machines
(including dom0).
If an interrupt is assigned to more than one machine it can happen from time
to time that
one of the machines disables the interrupt. This causes other resources on
this interrupt (and in other machines)
to be disabled too.

Have a look at /proc/interrupts in all of your machines to get to know if
there are interrupts assigned to more than one.
It's nothing special if there are several resources sharing one interrupt,
it's just not good if this happens across machines.

I had have had this problem with an usb controller sharing it's interrupt
with an onboard network controller
and solved it by disabling the usb controller in the bios.
So if you have an interrupt shared between machines, try to disabled some of
the resources using it.
I really don't know if there are ways to reassign interrupts,
but if they are "assigned" in hardware there is probably no way to do that.

Excuse my (sometimes) poor English and potential wrong statements (if there
are any).

Greetings,
leo


> -----Ursprüngliche Nachricht-----
> Von: xen-users-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-users-
> bounces@xxxxxxxxxxxxxxxxxxx] Im Auftrag von Pepe Barbe
> Gesendet: Donnerstag, 23. Juli 2009 17:28
> An: xen-users@xxxxxxxxxxxxxxxxxxx
> Betreff: [Xen-users] Netback grant fails
> 
> Hello,
> 
> For the past year I have been trying to debug a problem with my Xen
> setup that involves a passthrough PCI card. If there was too much
> network traffic (where too much, could be around 5 Mbps) the network
> would freeze and I would loose all my connectivity.
> 
> Initially I had the dom0 on Ubuntu 8.04 and things were terrible then.
> If I reached this state, the OS would have something in Ininturrptible
> I/O and I couldn't gracefully shut down, creating the possibility of
> breaking havoc on my software RAID. Since I moved the dom0 to Centos
> 5.3 using the latest stable Xen, the problem has abated but still
> happens on rare occasions; at least the system is responsive and I
> restart all the domUs and have the network come back.
> 
> It happened recently again I think I have found something that could
> be interesting in the Kernel Log (Posted at the end of the email).
> 
> The card in question uses the tg3 driver and requires it to be setup
> in PCI permissive mode. As you will see, the first line of my paste
> below says 'ACPI: PCI interrupt for device 0000:04:00.0 disabled' and
> then the netback fails.
> 
> Any ideas of what is going on or what could I do so I can reduce the
> likelihood of this happening again even more?
> 
> Thanks,
> Pepe
> 
> ACPI: PCI interrupt for device 0000:04:00.0 disabled
> eth0: port 2(vif2.0) entering disabled state
> device vif2.0 left promiscuous mode
> eth0: port 2(vif2.0) entering disabled state
> #### netback grant fails
> 
> < snip >
> 
> #### netback grant fails
> eth0: port 4(vif4.0) entering disabled state
> device vif4.0 left promiscuous mode
> eth0: port 4(vif4.0) entering disabled state
> eth0: port 3(vif3.0) entering disabled state
> device vif3.0 left promiscuous mode
> eth0: port 3(vif3.0) entering disabled state
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.