Hello, The data center didn't maintain its networking. The reason of disconnection is that my advisor commanded his staff to power off those servers forcibly. I think that he wouldn't like me to continue my research.
Anyway, Thank you all. 陳奕任 <[email protected]> 於 2020年5月22日 週五 下午4:33寫道： > Hello, > > Sorry about that I didn't notice the difference between ovs-dpctl and > ovs-ofctl. > I redo the four scenarios, and get the output of "ovs-dpctl dump-flows" > and "dmesg | tail". > > I look up the source code of Ryu on GitHub, and observe that Ryu didn't > implement selection_method yet. > I know the issue should be feedback to Ryu team. It is not your > responsibility. > > I want to try if I can use ovs-ofctl to change the selection_method on the > group, > but unfortunately, I cannot connect to my servers because the data center > may start to maintain its networking. Maybe until next Monday, I cannot > continue my work. > > > * ====== scenario 0: packet-out to "select" group and set > in_port=CONTROLLER ====== * > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:148, used:0.472s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:3, bytes:222, used:3.228s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:3, bytes:222, used:4.852s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:3, bytes:222, used:6.404s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:296, used:1.788s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:296, used:4.660s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=55,rule_cookie=0,controller_id=0,max_len=65509)) > > root@host1# dmesg | tail > [31460.287478] device eth2 entered promiscuous mode > [31620.582914] device eth2 left promiscuous mode > [31620.590767] device adapter left promiscuous mode > [31620.599356] device ovs-system left promiscuous mode > [31625.852877] device ovs-system entered promiscuous mode > [31625.853525] No such timeout policy "ovs_test_tp" > [31625.853525] Failed to associated timeout policy `ovs_test_tp' > [31625.856165] device adapter entered promiscuous mode > [31626.283349] device eth2 entered promiscuous mode > [31626.288887] IPv6: ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready > > > > * ====== scenario 1: packet-out to "fast-failover" group and set > in_port=CONTROLLER ====== * > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:0.888s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:5.152s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:5.956s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:6.760s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:140, used:2.196s, flags:P., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:140, used:2.892s, flags:P., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:140, used:3.564s, flags:P., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:140, used:4.173s, flags:P., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:272, used:1.492s, flags:FP., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:272, used:2.084s, flags:FP., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:272, used:2.688s, flags:FP., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:272, used:3.284s, flags:FP., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=71,rule_cookie=0,controller_id=0,max_len=65509)) > > root@host1# dmesg | tail > [31460.287478] device eth2 entered promiscuous mode > [31620.582914] device eth2 left promiscuous mode > [31620.590767] device adapter left promiscuous mode > [31620.599356] device ovs-system left promiscuous mode > [31625.852877] device ovs-system entered promiscuous mode > [31625.853525] No such timeout policy "ovs_test_tp" > [31625.853525] Failed to associated timeout policy `ovs_test_tp' > [31625.856165] device adapter entered promiscuous mode > [31626.283349] device eth2 entered promiscuous mode > [31626.288887] IPv6: ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready > > > > *====== scenario 2: packet-out to "select" group and set in_port=LOCAL > ====== * > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:0.928s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=65,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:6.624s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=65,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:0, bytes:0, used:never, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=67,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:0, bytes:0, used:never, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=67,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:1.212s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=69,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:66, used:2.120s, flags:., > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=69,rule_cookie=0,controller_id=0,max_len=65509)) > > root@host1# dmesg | tail > [31460.287478] device eth2 entered promiscuous mode > [31620.582914] device eth2 left promiscuous mode > [31620.590767] device adapter left promiscuous mode > [31620.599356] device ovs-system left promiscuous mode > [31625.852877] device ovs-system entered promiscuous mode > [31625.853525] No such timeout policy "ovs_test_tp" > [31625.853525] Failed to associated timeout policy `ovs_test_tp' > [31625.856165] device adapter entered promiscuous mode > [31626.283349] device eth2 entered promiscuous mode > [31626.288887] IPv6: ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready > > > > * ====== scenario 3: packet-out to "TABLE" port and set > in_port=CONTROLLER ====== * > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:1, bytes:74, used:0.708s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=45,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:2, bytes:148, used:3.573s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=45,rule_cookie=0,controller_id=0,max_len=65509)) > root@host1# ovs-dpctl dump-flows system@ovs-system > > recirc_id(0),in_port(2),eth(src=52:54:00:6d:8e:2f,dst=56:b6:d7:40:c3:42),eth_type(0x0800),ipv4(src= > 128.0.0.0/128.0.0.0,dst=128.0.0.0/128.0.0.0,proto=6,frag=no),tcp(dst=8080), > packets:4, bytes:296, used:9.004s, flags:S, > actions:userspace(pid=2708385104,controller(reason=1,dont_send=0,continuation=0,recirc_id=45,rule_cookie=0,controller_id=0,max_len=65509)) > > root@host1# dmesg | tail > [31460.287478] device eth2 entered promiscuous mode > [31620.582914] device eth2 left promiscuous mode > [31620.590767] device adapter left promiscuous mode > [31620.599356] device ovs-system left promiscuous mode > [31625.852877] device ovs-system entered promiscuous mode > [31625.853525] No such timeout policy "ovs_test_tp" > [31625.853525] Failed to associated timeout policy `ovs_test_tp' > [31625.856165] device adapter entered promiscuous mode > [31626.283349] device eth2 entered promiscuous mode > [31626.288887] IPv6: ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready > > > > > > > Han Zhou <[email protected]> 於 2020年5月22日 週五 下午2:20寫道： > >> Thanks for the test scenarios. It seems to be a problem related to >> dp_hash when in_port=CONTROLLER. I guess it would work if you specify >> selection_method=hash in the group. Could you try that? By default it is >> using dp_hash, which requires recirculation as the trace output shows. It >> is possible that the dp_hash didn't get any value (0) from kernel datapath >> so that it would keep recirculating, something similar as [0]. However, it >> needs to be verified by ovs-dpctl dump-flows (instead of ovs-ofctl >> dump-flows) and dmesg | tail. I didn't find it included in your logs. This >> command needs to be executed immediately after the packet-out before the >> cache is cleared. >> >> Although I suspect it could be the recirculation problem, I don't think >> it is the exact same root cause of [0] because in your case the packet >> doesn't need to be slowpathed, so I guess the patch would not help for your >> case, and I can't explain yet why in_port=LOCAL works. I will stop >> speculating before getting more information. >> >> [0] - >> https://patchwork.ozlabs.org/project/openvswitch/patch/[email protected]/ >> >> On Thu, May 21, 2020 at 9:58 PM 陳奕任 <[email protected]> wrote: >> >>> Hello, >>> >>> Thanks for your quick response. The following contains 4 scenarios (0 ~ >>> 3): >>> >>> scenario 0: packet-out to "select" group and set in_port=CONTROLLER >>> scenario 1: packet-out to "fast-failover" group and set >>> in_port=CONTROLLER >>> scenario 2: packet-out to "SELECT" group and set in_port=LOCAL >>> scenario 3: packet-out to "TABLE" port and set in_port=CONTROLLER >>> >>> In scenario 1 and 2, my composed packets are sent, but they cannot be >>> sent in scenario 0 and 3. >>> >>> ==== >>> root@host1# ovs-ofctl -O OpenFlow15 show adapter >>> OFPT_FEATURES_REPLY (OF1.5) (xid=0x2): dpid:0000bad0babe0051 >>> n_tables:254, n_buffers:0 >>> capabilities: FLOW_STATS TABLE_STATS PORT_STATS GROUP_STATS QUEUE_STATS >>> BUNDLES >>> OFPST_PORT_DESC reply (OF1.5) (xid=0x3): >>> 1(eth2): addr:06:ef:e9:1a:0c:96 >>> config: 0 >>> state: LIVE >>> current: 10GB-FD COPPER >>> speed: 10000 Mbps now, 0 Mbps max >>> LOCAL(adapter): addr:a2:2b:ed:d2:1b:4b >>> config: 0 >>> state: LIVE >>> speed: 0 Mbps now, 0 Mbps max >>> OFPT_GET_CONFIG_REPLY (OF1.5) (xid=0x9): frags=normal miss_send_len=0 >>> >>> >>> *====== scenario 0: packet-out to "select" group and set >>> in_port=CONTROLLER ====== * >>> ======== part of the code (OpenFlow 1.5)======== >>> buffer_id=ofproto.OFP_NO_BUFFER >>> in_port = ofproto.OFPP_CONTROLLER >>> actions = [ parser.OFPActionGroup(1) ] >>> msg = parser.OFPPacketOut(datapath=datapath, buffer_id=buffer_id, >>> match=parser.OFPMatch(in_port=in_port), actions=actions, data=packet.data) >>> datapath.send_msg(msg) >>> >>> ======== after RyuApp runs, before packet-out ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=11.384s, table=0, n_packets=0, n_bytes=0, >>> idle_age=11, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 >>> actions=group:1 >>> cookie=0x0, duration=11.384s, table=0, n_packets=0, n_bytes=0, >>> idle_age=11, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-groups adapter >>> OFPST_GROUP_DESC reply (OF1.5) (xid=0x2): >>> group_id=1,type=select,bucket=bucket_id:0,actions=output:eth2 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=94.104s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x1) >>> >>> ======== after packet-out (telnet to port 8080 from other host) ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=148.805s, table=0, n_packets=0, n_bytes=0, >>> idle_age=148, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 >>> actions=group:1 >>> cookie=0x0, duration=148.806s, table=0, n_packets=4, n_bytes=296, >>> idle_age=6, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=208.300s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x18) >>> >>> >>> >>> *====== scenario 1: packet-out to "fast-failover" group and set >>> in_port=CONTROLLER ====== * >>> ======== part of the code (OpenFlow 1.5)======== >>> buffer_id=ofproto.OFP_NO_BUFFER >>> in_port = ofproto.OFPP_CONTROLLER >>> actions = [ parser.OFPActionGroup(1) ] >>> msg = parser.OFPPacketOut(datapath=datapath, buffer_id=buffer_id, >>> match=parser.OFPMatch(in_port=in_port), actions=actions, data=packet.data) >>> datapath.send_msg(msg) >>> >>> ======== after RyuApp runs, before packet-out ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=20.749s, table=0, n_packets=0, n_bytes=0, >>> idle_age=20, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 >>> actions=group:1 >>> cookie=0x0, duration=20.748s, table=0, n_packets=0, n_bytes=0, >>> idle_age=20, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-groups adapter >>> OFPST_GROUP_DESC reply (OF1.5) (xid=0x2): >>> >>> group_id=1,type=ff,bucket=bucket_id:0,watch_port:eth2,actions=output:eth2 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=173.081s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> using bucket 0 >>> bucket 0 >>> output:1 >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: 2 >>> >>> ======== after packet-out (telnet to port 8080 from other host) ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=333.717s, table=0, n_packets=0, n_bytes=0, >>> idle_age=333, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 >>> actions=group:1 >>> cookie=0x0, duration=333.716s, table=0, n_packets=5, n_bytes=346, >>> idle_age=32, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=365.835s,ref_count=1,packet_count=3,byte_count=194,bucket0:packet_count=3,byte_count=194 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> using bucket 0 >>> bucket 0 >>> output:1 >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: 2 >>> >>> >>> >>> *====== scenario 2: packet-out to "SELECT" group and set in_port=LOCAL >>> ====== * >>> ======== part of the code (OpenFlow 1.5)======== >>> buffer_id=ofproto.OFP_NO_BUFFER >>> in_port = ofproto.OFPP_LOCAL >>> actions = [ parser.OFPActionGroup(1) ] >>> msg = parser.OFPPacketOut(datapath=datapath, buffer_id=buffer_id, >>> match=parser.OFPMatch(in_port=in_port), actions=actions, data=packet.data) >>> datapath.send_msg(msg) >>> >>> ======== after RyuApp runs, before packet-out ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=65.134s, table=0, n_packets=0, n_bytes=0, >>> idle_age=65, priority=8192,tcp,in_port=LOCAL,tp_src=8080 actions=group:1 >>> cookie=0x0, duration=65.134s, table=0, n_packets=0, n_bytes=0, >>> idle_age=65, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-groups adapter >>> OFPST_GROUP_DESC reply (OF1.5) (xid=0x2): >>> group_id=1,type=select,bucket=bucket_id:0,actions=output:eth2 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=172.172s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=LOCAL,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=LOCAL,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=LOCAL,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x1a) >>> >>> ======== after packet-out (telnet to port 8080 from other host) ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=296.942s, table=0, n_packets=0, n_bytes=0, >>> idle_age=296, priority=8192,tcp,in_port=LOCAL,tp_src=8080 actions=group:1 >>> cookie=0x0, duration=296.942s, table=0, n_packets=5, n_bytes=346, >>> idle_age=17, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=334.464s,ref_count=1,packet_count=3,byte_count=194,bucket0:packet_count=3,byte_count=194 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=334.464s,ref_count=1,packet_count=3,byte_count=194,bucket0:packet_count=3,byte_count=194 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=LOCAL,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=LOCAL,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=LOCAL,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x1f) >>> >>> >>> >>> *====== scenario 3: packet-out to "TABLE" port and set >>> in_port=CONTROLLER ====== * >>> ======== part of the code (OpenFlow 1.5)======== >>> buffer_id=ofproto.OFP_NO_BUFFER >>> in_port = ofproto.OFPP_CONTROLLER >>> actions = [ parser.OFPActionOutput(ofproto.OFPP_TABLE) ] >>> msg = parser.OFPPacketOut(datapath=datapath, buffer_id=buffer_id, >>> match=parser.OFPMatch(in_port=in_port), actions=actions, data=packet.data) >>> datapath.send_msg(msg) >>> >>> ======== after RyuApp runs, before packet-out ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=21.992s, table=0, n_packets=0, n_bytes=0, >>> idle_age=21, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 >>> actions=group:1 >>> cookie=0x0, duration=21.992s, table=0, n_packets=0, n_bytes=0, >>> idle_age=21, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-groups adapter >>> OFPST_GROUP_DESC reply (OF1.5) (xid=0x2): >>> group_id=1,type=select,bucket=bucket_id:0,actions=output:eth2 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=93.852s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x21) >>> >>> ======== after packet-out (telnet to port 8080 from other host) ======== >>> root@host1# ovs-ofctl -O OpenFlow15 dump-flows adapter >>> cookie=0x0, duration=242.223s, table=0, n_packets=5, n_bytes=370, >>> idle_age=2, priority=8192,tcp,in_port=CONTROLLER,tp_src=8080 actions=group:1 >>> cookie=0x0, duration=242.223s, table=0, n_packets=5, n_bytes=370, >>> idle_age=2, priority=4096,tcp,in_port=eth2,tp_dst=8080 >>> actions=CONTROLLER:65509 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-groups adapter >>> OFPST_GROUP_DESC reply (OF1.5) (xid=0x2): >>> group_id=1,type=select,bucket=bucket_id:0,actions=output:eth2 >>> >>> root@host1# ovs-ofctl -O OpenFlow15 dump-group-stats adapter >>> OFPST_GROUP reply (OF1.5) (xid=0x6): >>> >>> >>> group_id=1,duration=279.572s,ref_count=1,packet_count=0,byte_count=0,bucket0:packet_count=0,byte_count=0 >>> >>> root@host1# ovs-appctl ofproto/trace-packet-out adapter >>> 'tcp,in_port=CONTROLLER,tp_src=8080' 'group:1' >>> Flow: >>> tcp,in_port=CONTROLLER,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,nw_src=0.0.0.0,nw_dst=0.0.0.0,nw_tos=0,nw_ecn=0,nw_ttl=0,tp_src=8080,tp_dst=0,tcp_flags=0 >>> >>> bridge("adapter") >>> ----------------- >>> group:1 >>> -> no live bucket >>> >>> Final flow: unchanged >>> Megaflow: recirc_id=0,eth,ip,in_port=CONTROLLER,nw_frag=no >>> Datapath actions: hash(l4(0)),recirc(0x2c) >>> >>> >>> >>> >>> >>> Han Zhou <[email protected]> 於 2020年5月22日 週五 上午8:56寫道： >>> >>>> >>>> >>>> >>>> On Thu, May 21, 2020 at 12:20 PM Ben Pfaff <[email protected]> wrote: >>>> > >>>> > On Thu, May 21, 2020 at 10:50:21PM +0800, 陳奕任 wrote: >>>> > > Hello, >>>> > > >>>> > > >>>> > > ==== What you did that make the problem appear. ==== >>>> > > I use Ryu as the OpenFlow controller. >>>> > > I try to use Packet-Out with Actions=ActionGroup(1) and >>>> in_port=Controller, >>>> > > where the type of group 1 is "select". >>>> > > There is only one bucket and the bucket has only one ActionOutput >>>> to a >>>> > > physical port, because I am doing a small test for a project. >>>> > > >>>> > > I use OpenFlow13 firstly, and then I try to use OpenFlow15. both of >>>> them >>>> > > have the same problem. >>>> > > >>>> > > ==== What you expected to happen. ==== >>>> > > My composed packets should be sent >>>> > > >>>> > > ==== What actually happened. ==== >>>> > > no packet is sent. >>>> > > >>>> > > ==== other trials (other information that I think might be >>>> relevant.) ==== >>>> > > (1) If I change the group type to "FastFailover", it works. >>>> > > (2) if I change the in_port to "Local", it works. >>>> > > (3) if I change the Action=ActionOutput(Table), and add a flow >>>> entry >>>> > > which match is "in_port=controller" and "actions=group(1)", it >>>> doesn't >>>> > > work. I use " ovs-ofctl dump-group-stats" to check packet_count and >>>> > > byte_count, but they are not increased. >>>> > > However, the packet_count and byte_count are increased on the flow >>>> entry >>>> > > when I use ovs-ofctl dump-flows to check them. >>>> > >>>> > That's quite odd. Have you tried ofproto/trace-packet-out to see >>>> what's >>>> > going on? See ovs-vswitchd(8) for documentation. >>>> >>>> In addition, could you attach the output of "ovs-dpctl dump-flows" and >>>> "ovs-ofctl dump-groups" when it happens? >>>> >>>
_______________________________________________ discuss mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
