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

RE: [Xen-users] Xen 3.1 + CentOS 5 + HP ProLiant G5 = DomU packetdrops

  • To: <xen-users@xxxxxxxxxxxxxxxxxxx>
  • From: "Boudreau Luc" <luc.boudreau@xxxxxxxxxxxx>
  • Date: Tue, 13 Nov 2007 09:14:54 -0500
  • Delivery-date: Tue, 13 Nov 2007 06:15:44 -0800
  • List-id: Xen user discussion <xen-users.lists.xensource.com>
  • Thread-index: Acgl6CCvrgD1e9OBR3KMDW5WTxnadgAFkFRQ
  • Thread-topic: [Xen-users] Xen 3.1 + CentOS 5 + HP ProLiant G5 = DomU packetdrops

I've had (and still have...) this problem with one of my servers. The setup is 
special though. It has 2 NICs, one Broadcom and one Intel. The onboard Broadcom 
was used as a bridge between my domUs and the Intel was dedicated to a MySQL 
domU running CentOS 5. Dom0 is also CentOS 5.

This issue appeared when I started the transfer of large files between a Win 
2003 domU to any external device (by which I mean not a domain of the server). 
Any transfer of a file over approx 2gigs made the connexion fail on all network 
cards and packets were being dropped.

After I updated both drivers, the Intel card started to behave correctly. Now, 
I'm looking for replacing the Broadcom for an Intel one. This is a HARD issue 
to troubleshoot, and so far, having been a subscriber of this list for a while, 
no one was able to resolve it. My guess is that the Broadcom driver has some 
trouble with Xen... this is just a guess that needs to be tested.

Please keep this list informed of your progress, I'm sure other people are 
reading it and looking for the answer to this same issue.


Luc Boudreau
Registrariat, Université de Montréal

-----Message d'origine-----
De : xen-users-bounces@xxxxxxxxxxxxxxxxxxx 
[mailto:xen-users-bounces@xxxxxxxxxxxxxxxxxxx] De la part de Ivan Porro
Envoyé : 13 novembre 2007 06:27
À : xen-users@xxxxxxxxxxxxxxxxxxx
Cc : Pezza
Objet : Re: [Xen-users] Xen 3.1 + CentOS 5 + HP ProLiant G5 = DomU packetdrops

Dear all,

I've the same problem on xen 3.1, PE 1950 server (on dom0 I've broadcom 
nextreme driver updated to 1.5) I'm experiencing a strange behaviour.

On windows HVM (2003 and XP) I don' have problem via the built-in VNC 
server but, if I connect via RDP (terminal server), after few minutes of 
connection, the terminal server client freeze (on windows and on linux) 
and the HVM does not respond. After about half minute of panic, it go 
back working just to freeze again after few minutes.
It freeze even if from a VNC session on Win2k3 server machine I open a 
terminal server to the very same server... the RDP session crashes, 
while the VNC is perfect.
I need to virtualize a server for terminal server access so I cannot 
rely on VNC only...

During the freeze, if I click a button (i.e. sw installation on windows) 
it "remember" the event and as soon as the system unlock it execute the 
requested commands.

during the worst freeze I've experienced my laptop (Ubuntu 7.10) became 
totally locked too... to complete this e-mail I had to CTR+ALT+F1 switch 
to text console and kill tsclient.

The only useful info I found is:

Net0 RX:   0bytes        0pkts        0err    23759drop
TX:        0bytes        0pkts        0err        0drop

Net0 RX:   0bytes        0pkts        0err    23757drop
TX:        0bytes        0pkts        0err        0drop

but I don't know how to further investigate in and relate it to other 
system hw/sw components. Network inside the VM seems working good.

I'll google around until new advice :)


Pezza wrote:
> Hi all,
> I've a Xen server (Xen 3.1 compiled from sources [stable downloaded last
> wed...], CentOS 5 as the host OS, HP Proliant G5 6Gb ram with Broadcom
> NetXtreme 2) with 4 guests, each running Windows 2003 just installed, with
> nothing else.
> Network is configured with the default bridging script, no changes. I'm
> using 1 network card only for the moment.
> All guests run apparently fine, but if I try to download from a server on
> the same subnet and/or from a share on the same subnet, I get strange
> corruption errors.
> Poking around, I found out that all the guests are experiencing packet drops
> on their network cards: even just after booting the guest, with the guest os
> completely idle, I can see from xm top that the guest network card is
> dropping packets.
> Googling around, I've found some similar situations and thus I tried:
>  * updating the network card drivers: I'm now using version 1.5.xxx
> downloaded from the broadcom website and installed following instructions
> found on this mailing list
>  * enabling IP forwarding in the host kernel (instead of using the
> automatically generated rules in the forward chain in iptables)
>  * disabling the management mode on the card firmware, following
> instructions found on this mailing list
>  * using the ne2k_pci driver in the guest instead of the default
>  * trying different vm schemes (1, 2 or 3 VM at a time instead of 4)
>  * pinning Dom0 to one CPU and the guest on another (as suggested somewhere
> on this mailing list)
> None of the above fixed the problem.
> I don't know what else could I do to fix this.
> Is this situation normal? Is using a PV enabled host/guest system the only
> solution?
> Thanks in advance,
>                                                               M.

voice: +39 010 353 2789
fax:   +39 010 353 2948

Xen-users mailing list

Xen-users mailing list



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