[CentOS] Re: Bonding and Xen

Axel Thimm Axel.Thimm at ATrpms.net
Fri Jul 18 08:25:37 UTC 2008


On Tue, Jul 15, 2008 at 03:34:56PM -0500, Victor Padro wrote:
> Does anyone has implemented this sucessfully?

Yes and no. :/

> I am asking this because we are implementing Xen on our test lab machines,
> which they hold up to three 3com and intel Nics 10/100mbps based.
> 
> These servers are meant to replace MS messaging and intranet webservers
> which holds up to 5000 hits per day and thousands of mails, and probably the
> Dom0 could not handle this kind of setup with only one 100mbps link, and
> could not afford changing all the networking hardware to gigabit, at least
> not yet.
> 
> Any pointers perhaps?

This is not CentOS specific, nor RHEL & clones, but a general Linux
issue: bonding and bridging is broken leading to loops on the switch
unless the switch is intelligent enough to do trunking on its side.

The problem is that the outgoing packages from the virtual xen bridge
are seen by the other bonding memebers and the learned mac addresses
on the xen bridge toggle from the VM to the outside interface.

I had posted this issue on the respective lists, but nothing happened
- ideally the bridge code would allow for static macs.

This then indirectly affects anything that uses a Linux bridge,
including xen and most other virtual solutions. If you google for
bonding on each of them you will find trouble reports all over.

So you options are:

a) use only active/backup type solutions to avoid loops.
b) use an inteligent swicth that is able to trunk ports and therefore
   does not generate the loops. But then these servers cannot be
   PXE/DHCP booted anymore (for reinstalling them).

I had these issues with the 2x1GB setup on the ATrpms servers and lost
a lot of hair over it. The increased throughput was there at the end,
but maybe I'd preferred to keep my hair ...
-- 
Axel.Thimm at ATrpms.net



More information about the CentOS mailing list