Xen users - Apr 2007 - network error, even when reboot into non-Xen smp kernel

After installing xen-3.0.4_1-install-x86_32p.tgz on a CentOS 4.4, I was
getting a network error:
Device eth0 has different MAC address than expected

I have searched for hours here and other places, tried all the recommended
solutions and no luck. However, my real question is this... Even after I
reboot into the previous CentOS kernel ( CentOS (2.6.9-42.0.10.ELsmp)), I
get the same network error. I have tried this on 2 different boxes and it is
completely reporducible. (Different network cards, too.)

Strangely, I can boot into the non-smp kernel: CentOS (2.6.9-42.0.10.EL) and
get networking.

What is Xen doing that is persistent and causing network failure even after
fallback to the original smp kernel?? (Maybe it is worth saying that the
network was 100% fine before Xen install, in fact, sadly it is a completely
remote server, so that is the only way to get into the box unless I want to
pay for an expensive remote hands service!!)

How can I undo these changes and restore networking to my smp kernel? Help
please!!


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

Robert Ficcaglia wrote:
> After installing xen-3.0.4_1-install-x86_32p.tgz on a CentOS 4.4, I 
> was getting a network error:
> Device eth0 has different MAC address than expected
First, why aren''t you using the RHEL 4 RPM? It''s true they
accidentally
left out the documentaiton in that, but it works fine, and has RPM 
kernels for both Dom0 and DomU.

Second, unless you hard-code your MAC address in your Xen configuration 
file, it''s going to get changed randomly in your DomU''s every
freaking
time you reboot them. And RedHat''s network setup tools look in 
/etc/sysconfig/network-scripts/ifcfg-eth* to associate the HWADDR 
setting with specific network devices. This keeps eth0 and eth1 from 
getting swapped around for different kernel configurations, and helps 
keep the network device names consistent when adding wi-fi or USB 
network devices.

You can also simply comment out the HWADDR lines in 
/etc/sysconfig/networ-kscripts/ifcfg-eth* if you have only one network 
device and thus aren''t worried about swapping them.

Is the problem in your Dom0? Do you have multiple network devices?
> I have searched for hours here and other places, tried all the 
> recommended solutions and no luck. However, my real question is 
> this... Even after I reboot into the previous CentOS kernel ( CentOS 
> (2.6.9-42.0.10.ELsmp )), I get the same network error. I have tried 
> this on 2 different boxes and it is completely reporducible. 
> (Different network cards, too.)
>  
> Strangely, I can boot into the non-smp kernel: CentOS 
> (2.6.9-42.0.10.EL) and get networking.
See above.
>  
> What is Xen doing that is persistent and causing network failure even 
> after fallback to the original smp kernel?? (Maybe it is worth saying 
> that the network was 100% fine before Xen install, in fact, sadly it 
> is a completely remote server, so that is the only way to get into the 
> box unless I want to pay for an expensive remote hands service!!)

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

No reason I used the tarball other than it was first on the list :) I would
be glad to remove what''s on there now and reinstall with RPMs, part of
the
problem is that I don''t know the steps to completely remove Xen and
start
clean; this is the best list I found:
http://lists.xensource.com/archives/html/xen-users/2006-04/msg00617.html
or download the src tarball and make uninstall (probably will be my first
try)

To answer your other questions, yes - the trouble is with Dom0, and yes I
have multiple (2) NICs (same hw).  So with 2 NICs, I presume I cannot simply
comment out the HWARDDR (cause I tried that and it didn''t help); so
what is
the procedure for 2 NICs?


On 4/9/07, Nico Kadel-Garcia <nkadel@gmail.com>
wrote:
>
> Robert Ficcaglia wrote:
> > After installing xen-3.0.4_1-install-x86_32p.tgz on a CentOS 4.4, I
> > was getting a network error:
> > Device eth0 has different MAC address than expected
> First, why aren''t you using the RHEL 4 RPM? It''s true
they accidentally
> left out the documentaiton in that, but it works fine, and has RPM
> kernels for both Dom0 and DomU.
>
> Second, unless you hard-code your MAC address in your Xen configuration
> file, it''s going to get changed randomly in your DomU''s
every freaking
> time you reboot them. And RedHat''s network setup tools look in
> /etc/sysconfig/network-scripts/ifcfg-eth* to associate the HWADDR
> setting with specific network devices. This keeps eth0 and eth1 from
> getting swapped around for different kernel configurations, and helps
> keep the network device names consistent when adding wi-fi or USB
> network devices.
>
> You can also simply comment out the HWADDR lines in
> /etc/sysconfig/networ-kscripts/ifcfg-eth* if you have only one network
> device and thus aren''t worried about swapping them.
>
> Is the problem in your Dom0? Do you have multiple network devices?
>
> > I have searched for hours here and other places, tried all the
> > recommended solutions and no luck. However, my real question is
> > this... Even after I reboot into the previous CentOS kernel ( CentOS
> > (2.6.9-42.0.10.ELsmp )), I get the same network error. I have tried
> > this on 2 different boxes and it is completely reporducible.
> > (Different network cards, too.)
> >
> > Strangely, I can boot into the non-smp kernel: CentOS
> > (2.6.9-42.0.10.EL) and get networking.
> See above.
>
> >
> > What is Xen doing that is persistent and causing network failure even
> > after fallback to the original smp kernel?? (Maybe it is worth saying
> > that the network was 100% fine before Xen install, in fact, sadly it
> > is a completely remote server, so that is the only way to get into the
> > box unless I want to pay for an expensive remote hands service!!)
>
>

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

Robert Ficcaglia wrote:
> No reason I used the tarball other than it was first on the list :) I 
> would be glad to remove what''s on there now and reinstall with
RPMs,
> part of the problem is that I don''t know the steps to completely 
> remove Xen and start clean; this is the best list I found:
> http://lists.xensource.com/archives/html/xen-users/2006-04/msg00617.html
> or download the src tarball and make uninstall (probably will be my 
> first try)
>  
> To answer your other questions, yes - the trouble is with Dom0, and 
> yes I have multiple (2) NICs (same hw).  So with 2 NICs, I presume I 
> cannot simply comment out the HWARDDR (cause I tried that and it 
> didn''t help); so what is the procedure for 2 NICs?
Well, a "tar tzf" of the tarball would be a good start to know what to
remove.

Are the NIC''s identical models to each other? Do you have available the
right modules for it, as listed in /etc/modprobe.conf and listed by lsmod?

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

On Mon, Apr 09, 2007 at 08:26:56AM -0700, Robert Ficcaglia
wrote:
> To answer your other questions, yes - the trouble is with Dom0, and yes I
> have multiple (2) NICs (same hw).  So with 2 NICs, I presume I cannot
simply
> comment out the HWARDDR (cause I tried that and it didn''t help);
so what is
> the procedure for 2 NICs?
Could you please run the following command in both cases? (It works,
non-smp centos kernel; it doesn''t, dom0.)

/sbin/ifconfig -a

And send us the output? 

-- 
lfr
0/0


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