snat in PREROUTING chain?
"Stephen M. Rumble" <stephen.rumble-H217xnMUJC0sA/[email protected]> Tue, 24 Apr 2007 21:24:32 -0400
| Newsgroups | gmane.linux.network.bridge.ebtables.user |
|---|---|
| Message-ID | <[email protected]> |
Hi all, Apologies in advance if this is covered elsewhere, but I haven't yet found anything that seems to match what I'm trying to accomplish. I'm working on a project that involves large numbers of Xen virtual machines, which need to be sandboxed network-wise into multiple private subnets. For efficient and minimally restrictive containment it seemed that using ebtables would be a good way to go. The idea is simply to map each VMs internal MAC address to one we allocate externally. We use some of the upper bits to represent a 'virtual network' identifier. Then we employ ebtables to ensure that each VM may only receive packets with source MAC addresses matching its virtual network. However, we've run into a problem. Since ebtables only snats after the bridging code has seen the internal VM MAC address, the bridge is confused when two virtual machines with the same internal MAC address are enslaved on it. What we'd like is to change the MAC address before it comes out of the Xen vif and into the bridge, keeping whatever original internal address exists completely isolated. (This also permits us to have VMs with the same MAC address internally - a nicety since we're actually forking them!) Is this possible already? Am I missing something? If not, should I investigate patching ebtables to allow snat in PREROUTING, or is there a good reason why this is not already done? Thanks in advance, Steve P.S. Please CC me as I'm not subscribed to the list. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/