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

Re: [Xen-users] Intermittent network in HVM domU



You checked the MAC and IP addresses, right?

--
Michael South

On May 6, 2011, at 12:38, Roberto Bifulco <roberto.bifulco2@xxxxxxxx> wrote:

Some more infos!
I run a simple ping test from my laptop to both domUs runinng on my server. The ping test shows that both domU suffers intermittent connectivity. More interestingly, when a domU network works, the other domU's network doesn't and viceversa.
So it seems a "network scheduler" problem. 

I'm starting searching about how can I customize the behavior of the IO scheduler for VNICs.
Roberto

2011/5/6 Roberto Bifulco <roberto.bifulco2@xxxxxxxx>
Todd,
thanks for your fast answer.
I'm going to install PVonHVM drivers for the linux domU to give them a try.
Today I realized that the intermittent connectivity is an issue only when a second domU is started, i.e, running just the linux domU works well.

Anyway, I should solve the issue even in the case of HVM domUs (without PV drivers), so still looking for any hint.
Thanks,
Roberto

2011/5/5 Todd Deshane <todd.deshane@xxxxxxx>
On Thu, May 5, 2011 at 8:17 AM, Roberto Bifulco
<roberto.bifulco2@xxxxxxxx> wrote:
> Dear Xen Community,
> I'm using Xen to virtualize some Windows and Linux servers.
> Currently I'm using two machines, one for production, which runs Xen 3.2.1
> with Debian5 Dom0 and the second one (Xen 4.0.1, openSuse 11.4 dom0) for
> development/testing.
> I'm experiencing some problems with an HVM linux domU over the development
> machine.
> In particular, the linux domU runs network services (web server, ssh), and,
> periodically, in an intermittent manner, the domU's network goes off.
> I realized that the vnc console attached to the domU still works, i.e., the
> domU is responsive, while the network doesn't work.
> The host machine is underloaded, i.e., it runs just two hvm domUs (the linux
> domU and an idle windows domU) and is configured in this way:
> one Intel processor with 4 cores,
> dom0 with 1 vcpu pinned to core 0
> other domains with 1 vcpu pinned to cores 1-3
> changing the dom0 credit scheduler configurations doesn't change this
> behaviour (e.g., assigning to dom0 a weight of 2048 while other domains have
> weight set to 256).
> Any suggestion on how I can solve this problem, or how can I figure out
> WHERE is the problem?

Have you tried installing the PVonHVM drivers in the Linux domU?

Have you tried disabling transmit checksum offloading?

Does a linux PV guest have the problem or just HVM?


Thanks,
Todd

> Thanks,
> Roberto
> --
> Roberto Bifulco, Ph.D. Student
> robertobifulco.it
> COMICS Lab - www.comics.unina.it
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users
>



--
Todd Deshane
http://www.linkedin.com/in/deshantm
http://www.xen.org/products/cloudxen.html
http://runningxen.com/




--
Roberto Bifulco, Ph.D. Student
robertobifulco.it
COMICS Lab - www.comics.unina.it



--
Roberto Bifulco, Ph.D. Student
robertobifulco.it
COMICS Lab - www.comics.unina.it
_______________________________________________
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®.