On Fri, Feb 12, 2010 at 2:51 PM, Todd Deshane <[email protected]> wrote: > So I think the next problem that I would like to solve is getting the > host network to work correctly. When I do the obvious of simply > copying the flows for the VM and changing the mac address to that of > eth0, the host is not able to work as I would have expected. It can > dhcp, but that seems like all it can do. It can't lookup any over dns, > or anything else. The strange thing is that its flows don't seem to be > showing up anywhere (I check the switch logs), I only seem ARP traffic > for the eth0 mac and nothing else.
I think I've found a specific case of normal traffic from the host that is getting deleted. First, a dump of the default flow table for dp0/externalbr: ovs-ofctl dump-flows dp0 Feb 13 22:31:51|00001|ofctl|INFO|datapath dp0 is named externalbr Feb 13 22:31:51|00002|ofctl|INFO|connecting to unix:/usr/local/var/run/externalbr.mgmt stats_reply (xid=0xf7b36636): flags=none type=1(flow) duration=95286s, table_id=1, priority=0, n_packets=281959, n_bytes=28181037, actions=NORMAL and to refresh you, for sake of convenience, the switch config is: bridge.internalbr.port=internalbr bridge.externalbr.port=externalbr bridge.internalbr.mac=00:11:22:aa:cc:ee bridge.externalbr.mac=00:11:22:bb:aa:dd bridge.externalbr.port=eth0 So, with this in place, is there any reason why this flow is being deleted: Feb 13 22:24:23|746001|dpif|DBG|dp0: delete flow in_port0000:vlan65535 mac00:11:22:bb:aa:dd->00:1c:0f:95:1d:42 type0800 proto17 ip128.153.144.73->128.153.N.N port51679->53, packets:0, bytes:0, used:never It seems to be DNS traffic from externalbr to a local DNS server (IP masked with N.N.) My suspicion is that the problem has to do with mac address and/or some mis-coordination between eth0 and externalbr eth0 has a mac of 00:23:54:87:21:3c, and similarly has a useful flow being deleted. This looks like a DHCP response: Feb 13 22:22:43|744867|dpif|DBG|dp0: delete flow in_port0001:vlan65535 mac00:1c:0f:95:1d:42->00:23:54:87:21:3c type0800 proto17 ip128.153.144.1->128.153.144.198 port67->68, packets:1, bytes:350, used:5.864s I also found other similar examples of flows being deleted. This testing is all after a fresh reboot, and only loading the switch with the config given above, then just simply trying to get the host's networking to work. Note that for this particular test, I did not delete or add any flows form the default setup of the switch. The only way that I have found that I can get networking to work is to dpctl del-if externalbr eth0, then restart networking on eth0. Any ideas or things for me to try? Thanks, Todd -- Todd Deshane http://todddeshane.net http://runningxen.com _______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss_openvswitch.org
