On Wed, Jan 23, 2013 at 12:56 PM, Ben Pfaff <b...@nicira.com> wrote: > On Wed, Jan 23, 2013 at 06:18:52PM +0100, Markus Schuster wrote: >> On Wednesday 23 January 2013 17:49:09 Ben Pfaff wrote: >> > It looks like you have each interface on an SLB bond connected to two >> > different switches. >> >> Correct. >> >> >> > That's not a supported configuration and won't work reliably. >> >> OK, any explanation why? Or to say it the other way: I can't see any >> technical reason why it should work on the same switch but not on two >> different? The MAC will jump between the ports the very same way and the VM >> will suffer from the same connectivity issues, no matter what. > > Jesse, can you remind me why? I can never remember. This time, I'll > add it to the FAQ.
It does seem risky given the other problems inherent in SLB bonding but I can't actually think of a problem off the top of my head. It's been at least 3 years since I worked on that code though, so I don't really remember if I said something before. The behavior described seems odd though and not really related to the upstream switch configuration. Were there any modifications made to the flow table? _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
