Re: [Xen-users] Multiple Network Cards + Multiple Bridges on debian lenny

[Alexander Pirsig <alex@xxxxxxxxxx>]

Hi Fajar,

I think it could be the conntrack tables but at the moment its hard tell 
because the system load is back to normal.
At the moment my conntrack table and settings look like this:

cat /proc/sys/net/ipv4/netfilter/ip_conntrack_max
4097152

cat /proc/net/ip_conntrack | wc -l
17186

> I wonder about your vif-script though. Why do you have it commeted
> out? Which vif-script did you use?
> # (vif-script vif-bridge)
The answer to this question is:

cat /etc/xen/scripts/vif-bridge | head
#!/bin/bash
#============================================================================
# /etc/xen/vif-bridge
#
# Script for configuring a vif in bridged mode.
# The hotplugging system will call this script if it is specified either in
# the device configuration given to Xend, or the default Xend configuration
# in /etc/xen/xend-config.sxp.  If the script is specified in neither of those
# places, then this script is the default.

So that means if you comment out vif-bridge vif-bridge is used via 
default in Debian.
I just wonder because in our application we're getting a large amount of 
different Ip's. A normal client connect takes 0.3ms and then they are 
gone :) probably I'll should raise conntrack table settings.
Regards,

Alex

Fajar A. Nugraha wrote:
> On Mon, Aug 31, 2009 at 1:23 PM, Alexander Pirsig<alex@xxxxxxxxxx> wrote:
>
>   
> > Bridge config on Dom0 looks like this:
> >
> > xenbr1          8000.001b213b8a69       no              eth1
> >                                                       vif6.0
> >     
>   
> > Yesterday I discovered that sometimes during Highload on DomU's part of the
> > network is hard rechable with ssh or connections stalls even if I try to
> > connect to dom0 with has seperate uplink.
> >     
> There was a thread on this list sometime ago, regarding iptables. Do
> you have /proc/net/ip_conntrack on dom0? Does the number of lines
> close to /proc/sys/net/ipv4/netfilter/ip_conntrack_max?
>
>   
> > checking arp sometimes shows strange information (during highload) like:
> >
> > arp -na
> > ? (192.168.0.158) at 00:1b:21:3b:8a:68 [ether] on eth0
> > ? (192.168.0.158) at 00:1b:21:3b:8a:68[ether] on xenbr1
> >
> > So my question are:
> >
> > 1. Is it normal that bridge xenbr1 claims ip mac address of eth1?
> >     
> Yes.
>
>   
> > 2. Do I have to reset mac address on the bridge (xenbr1) or network
> > interface (eth1) to FF:FF:FF:FF bevor starting it.
> >     
> Not really. Your standard bridge setup using OS init scripts should
> work. That's what I use, with RHEL.
>
>   
> > 3. What could cause the problem with the arp table?
> > 4. Do I have to turn of  arp/multicast on eth1 or xenbr1 like this ->
> >   ip link set xenbr1 arp off
> >   ip link set xenbr1 multicast off
> >     
> I'm not sure about these two. I don't arp and multicast use their
> default settings on RHEL, YMMV.
> Since you mention "high load" I suspect it has something to do with
> iptables connection tracking.
>
> I wonder about your vif-script though. Why do you have it commeted
> out? Which vif-script did you use?
> # (vif-script vif-bridge)
>
>   
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users