Hi, 贺鹏:
It seems you are right. but ovs has correct routes. here is the infomation of my environment. there is only one difference, 192.168.6.133 goes over gateway 192.168.0.11 which is on vlan1904 rather than non-vlan netdev-bridge. could you have a look at the details , web link http://www.pasteme.cn/47815 thanks bridge("dpdktun") ----------------- 0. priority 32768 NORMAL -> learned that 52:54:00:d7:11:82 is on port patch-tun-br0 in VLAN 0 -> no learned MAC for destination, flooding -> output to native tunnel >> native tunnel routing failed Port "vxlan0" Interface "vxlan0" type: vxlan options: {key="20", local_ip="192.168.5.133", remote_ip="192.168.6.131"} [root@0133 ~]# ovs-appctl ovs/route/lookup 192.168.6.133 src 192.168.0.10 gateway 192.168.0.11 dev vlan1904 At 2020-08-08 22:10:40, "贺鹏" <[email protected]> wrote: >Have you used ofproto/trace to see the results? > >according to my experience, you probably forgot to set up the default >route for br-tun. >you can check by ovs-appctl ovs/route/show. > >yuelongguang <[email protected]> 于2020年8月7日周五 下午11:35写道： >> >> Hi, all >> two nodes: 3.10.0-1062.4.3.el7.x86_64, openvswitch-2.12.0 , each of which >> has one running vm. >> >> >> the flows of br-int and br-tun are NORMAL, local_ip and remote_ip are >> reachable. >> vxlan tunnal vxlan0 , options: {key="20", local_ip="192.168.5.133", >> remote_ip="192.168.6.131"} >> >> >> vhostclient--->br-int------->br-tun --->dpdk0(192.168.5.133), all bridge's >> openflows are normal. >> >> >> ovs-tcpdump can capture ping packets , but which are not encapsulated on >> vxlan0 , and not forwarded to dpdk0(dpdk binded physical nic) >> >> >> my question is does vxlan needs special openflow rules ? >> >> >> thanks >> >> >> >> >> >> _______________________________________________ >> dev mailing list >> [email protected] >> https://mail.openvswitch.org/mailman/listinfo/ovs-dev > > > >-- >hepeng _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
