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

Re: [Xen-users] domU loses network after a while

  • To: Pasi Kärkkäinen <pasik@xxxxxx>
  • From: Olivier Hanesse <olivier.hanesse@xxxxxxxxx>
  • Date: Fri, 19 Nov 2010 10:27:05 +0100
  • Cc: xen-users@xxxxxxxxxxxxxxxxxxx, Erwan RENIER <erwan.renier@xxxxxxxxxxx>
  • Delivery-date: Fri, 19 Nov 2010 01:28:19 -0800
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; b=TDMmTP/RFNa0gfQWjdIN+BczL7PrhUw7H6LlDi/Yq7lg2b7N0QQX1Sf1ezOQxO2EPr XpaiuktYU22SKQcXC9B7R9Hua8/mpGuZ+paibHUf87crRSG9SLadMvqRO06N+3dVtfO/ x2BFdCmJ2j8xmRoMPMrdab6yc/TnfZZfhH5nw=
  • List-id: Xen user discussion <xen-users.lists.xensource.com>

Hello Pasi,

Are you thinking about the smartpoll patch ?

* [x86/xen] Disable netfront's smartpoll mode by default.

If so, yes the debian kernel we are using, is fine about this issue.



2010/11/19 Pasi Kärkkäinen <pasik@xxxxxx>
On Thu, Nov 18, 2010 at 10:35:46PM +0100, Erwan RENIER wrote:
>    Le 18/11/2010 19:33, Stein van Oevelen a écrit :
>        Nathan Friess wrote:
>          Hi,
>          I have a host running Debian Squeeze with Xen 4.0.1 installed from
>          the Debian repository.  My host is working fine running 10 domUs,
>          except that after some number of days one or more domUs lose their
>          network interface.  In dom0 and domU I'm running kernel
>          2.6.32-5-xen-amd64 (PV kernel in Debian repo), and interfaces are in
>          a bridged config.
>        In case anyone runs into something similar, I did figure out how to
>        resolve my issue without having to reboot the domU... which is a
>        slight improvement I suppose.
>        Running
>           xm network-detach domUName 0
>        Causes eth0 to disappear in the domain, and the following to show up
>        in dmesg:
>          net eth0: xennet_release_rx_bufs: fix me for copying receiver.
>        Then running an xm network-attach will bring eth0 back and everything
>        will be okay.
>        I also tried rmmoding xen_netfront in domU before I figured out the
>        network-detach, but this doesn't help.  When modprobing again, the
>        domU says "eth0: link is not ready".  This also causes any subsequent
>        attempt to network-detach in dom0 to get stuck and eventually timeout
>        with "Device 0 (vif) could not be disconnected."
>        Also, when the vif is stuck, the TX dropped packets shown in dom0 is
>        steadily increasing.  It seems that dom0 is having trouble sending
>        packets to domU because one of the two gets stuck in an unexpected
>        state.  At least doing a network-detach will reset the state and allow
>        things to run properly again.
>        Nathan
>        _______________________________________________
>        Xen-users mailing list
>        [1]Xen-users@xxxxxxxxxxxxxxxxxxx
>        [2]http://lists.xensource.com/xen-users
>      I have a similar issue, however my DomU's seem to get their connection
>      back after a while automatically, most of the time it happens when I
>      physically login to the Dom0 :/

Are you guys running the latest kernel?

This sounds like a bug that was fixed two months ago (or so)
in the upstream "xen/stable-2.6.32.x" git branch..

-- Pasi


Xen-users mailing list



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