Adding OpenFlow plugin. On Thu, Apr 28, 2016 at 8:25 PM, Luis Gomez <ece...@gmail.com> wrote:
> After further manual testing: > > 1) Flow stats: There is consistent failure of some flows (mostly LLDP) not > being reported in stats the first time you start mininet after controller > starts. After that flow report fails sporadically. > 2) Address Tracker: I cannot reproduce the issue of host IP addresses are > seeing in all nodes outside CI. > 3) Host Tracker: I cannot reproduce the issue of no host is learned > anymore after remove/add port. > 4) Loop Remover: It is a bit slower but it works after adding 1 sec sleep. > > So I think we only have issue 1) for now. > > BR/Luis > > > On Apr 28, 2016, at 5:05 PM, Sai MarapaReddy <sai.marapare...@gmail.com> > wrote: > > Hi Luis, > > I tried with following patches in my local system and see LLDP flows on > all switches , as well i see no duplicate hosts. > > In local system i tried building projects in this order OFP -> l2switch -> > distribution. > > I request to re-run the jenkins with following patches. > > *Distribution repo* with https://git.opendaylight.org/gerrit/#/c/38047/ > > *L2Switch repo* with https://git.opendaylight.org/gerrit/#/c/33303/ > > *Openflowplugin repo* with https://git.opendaylight.org/gerrit/#/c/35892 > > > > > LLDP Flows:- > > mininet> sh ovs-ofctl dump-flows s1 -O OpenFlow13 > OFPST_FLOW reply (OF1.3) (xid=0x2): > cookie=0x2b00000000000002, duration=52.993s, table=0, n_packets=24, > n_bytes=2040, priority=100,dl_type=0x88cc actions=CONTROLLER:65535 > cookie=0x2b00000000000006, duration=49.679s, table=0, n_packets=82, > n_bytes=11519, priority=2,in_port=1 actions=output:2 > cookie=0x2b00000000000007, duration=49.674s, table=0, n_packets=76, > n_bytes=11079, priority=2,in_port=2 actions=output:1 > mininet> sh ovs-ofctl dump-flows s2 -O OpenFlow13 > OFPST_FLOW reply (OF1.3) (xid=0x2): > cookie=0x2a00000000000004, duration=21.73s, table=0, n_packets=10, > n_bytes=812, idle_timeout=1800, hard_timeout=3600, > priority=10,dl_src=7e:41:98:63:6b:c3,dl_dst=62:25:db:d3:c1:0d > actions=output:1 > cookie=0x2a00000000000005, duration=21.729s, table=0, n_packets=7, > n_bytes=686, idle_timeout=1800, hard_timeout=3600, > priority=10,dl_src=62:25:db:d3:c1:0d,dl_dst=7e:41:98:63:6b:c3 > actions=output:2 > cookie=0x2b00000000000001, duration=56.953s, table=0, n_packets=13, > n_bytes=1105, priority=100,dl_type=0x88cc actions=CONTROLLER:65535 > cookie=0x2b00000000000004, duration=53.616s, table=0, n_packets=117, > n_bytes=18360, priority=2,in_port=3 actions=output:2,output:1 > cookie=0x2b00000000000005, duration=53.616s, table=0, n_packets=21, > n_bytes=1470, priority=2,in_port=1 > actions=output:2,output:3,CONTROLLER:65535 > cookie=0x2b00000000000003, duration=53.634s, table=0, n_packets=19, > n_bytes=1386, priority=2,in_port=2 > actions=output:3,output:1,CONTROLLER:65535 > mininet> sh ovs-ofctl dump-flows s3 -O OpenFlow13 > OFPST_FLOW reply (OF1.3) (xid=0x2): > cookie=0x2a00000000000000, duration=30.894s, table=0, n_packets=4, > n_bytes=336, idle_timeout=1800, hard_timeout=3600, > priority=10,dl_src=72:60:32:2e:10:6f,dl_dst=72:21:37:35:aa:b3 > actions=output:2 > cookie=0x2a00000000000001, duration=30.892s, table=0, n_packets=4, > n_bytes=336, idle_timeout=1800, hard_timeout=3600, > priority=10,dl_src=72:21:37:35:aa:b3,dl_dst=72:60:32:2e:10:6f > actions=output:1 > cookie=0x2b00000000000000, duration=61.23s, table=0, n_packets=14, > n_bytes=1190, priority=100,dl_type=0x88cc actions=CONTROLLER:65535 > cookie=0x2b00000000000002, duration=57.969s, table=0, n_packets=117, > n_bytes=17796, priority=2,in_port=3 actions=output:1,output:2 > cookie=0x2b00000000000000, duration=57.983s, table=0, n_packets=22, > n_bytes=1680, priority=2,in_port=1 > actions=output:2,output:3,CONTROLLER:65535 > cookie=0x2b00000000000001, duration=57.978s, table=0, n_packets=22, > n_bytes=1680, priority=2,in_port=2 > actions=output:1,output:3,CONTROLLER:65535 > > > Regards, > Sai MarapaReddy > Brocade Communications > > On Mon, Apr 25, 2016 at 11:06 AM, Luis Gomez <ece...@gmail.com> wrote: > >> OK, here is the resulting distribution: >> >> >> https://jenkins.opendaylight.org/releng/view/integration/job/integration-multipatch-test-boron/4/org.opendaylight.integration$distribution-karaf/artifact/org.opendaylight.integration/distribution-karaf/0.5.0-SNAPSHOT/distribution-karaf-0.5.0-SNAPSHOT.zip >> >> And here are the test results: >> >> >> https://jenkins.opendaylight.org/releng/view/l2switch/job/l2switch-csit-1node-switch-only-boron/332/robot/report/log.html >> >> Failure analysis is similar as before: >> >> - Flow push: There is no LLDP flows being pushed. Also flood flows seem >> to be in config space because operational shows flow ID. >> - Address Tracker: All host IP addresses are seeing in all nodes, that is >> not right, only node connected to host should show the host. >> - Host Tracker: After remove/add OpenFlow port (ovs-vsctl add-port), no >> host is learned anymore through that port. >> - Loop Remover: Not working, using a loop topology, even when we observe >> some ports are in DISCARD mode, mininet sees DUP packets. >> >> BR/Luis >> >> >> On Apr 22, 2016, at 7:01 PM, Sai MarapaReddy <sai.marapare...@gmail.com> >> wrote: >> >> Hi Luis, >> >> Patch (https://git.opendaylight.org/gerrit/#/c/33303/) is good now. >> >> Could you please run check once. >> >> Regards, >> Sai MarapaReddy >> >> On Fri, Apr 22, 2016 at 4:22 PM, Amit Mandke (ammandke) < >> amman...@cisco.com> wrote: >> >>> Oh ok. >>> >>> -Amit >>> >>> >>> >>> >>> >>> >>> >>> >>> On 4/22/16, 11:31 AM, "Luis Gomez" <ece...@gmail.com> wrote: >>> >>> >Hi Amit, the idea we discussed in the l2switch call is to combine both: >>> ofplugin patch to change features [1] + l2switch patch do deal with new >>> plugin [2]. For that we had to remove any feature change in [2]. >>> > >>> >[1] https://git.opendaylight.org/gerrit/#/c/35892/ >>> >[2] https://git.opendaylight.org/gerrit/#/c/33303/ >>> > >>> >BR/Luis >>> > >>> >> On Apr 22, 2016, at 11:05 AM, Amit Mandke (ammandke) < >>> amman...@cisco.com> wrote: >>> >> >>> >> @Sai - what happened to the pom and feature changes in this patch. I >>> don’t see those anymore in the commit? How would this change in to Li >>> feature? Am I missing something? >>> >> >>> >> -Amit >>> >> >>> >> >>> >> From: Sai MarapaReddy <sai.marapare...@gmail.com> >>> >> Date: Thursday, April 21, 2016 at 10:17 AM >>> >> To: Amit Mandke <amman...@cisco.com> >>> >> Cc: Luis Gomez <ece...@gmail.com>, Abhijit Kumbhare < >>> abhijitk...@gmail.com>, OpenDayLight-L2switch-Dev < >>> l2switch-...@lists.opendaylight.org> >>> >> Subject: Re: [L2switch-dev] L2switch Li issues >>> >> >>> >> @Amit/Abhijit - Agree with you, we will likely abandon patch. >>> >> @Luis - I will rebase and push a patch so that it is ready for merge. >>> >> >>> >> Regards, >>> >> Sai MarapaReddy >>> >> >>> >> On Thu, Apr 21, 2016 at 10:12 AM, Amit Mandke (ammandke) < >>> amman...@cisco.com> wrote: >>> >> Please abandon the patch on beryllium if decision is made to not >>> change it. >>> >> >>> >> -Amit >>> >> >>> >> >>> >> From: <l2switch-dev-boun...@lists.opendaylight.org> on behalf of >>> Luis Gomez <ece...@gmail.com> >>> >> Date: Wednesday, April 20, 2016 at 7:31 PM >>> >> To: Abhijit Kumbhare <abhijitk...@gmail.com>, Sai MarapaReddy < >>> sai.marapare...@gmail.com> >>> >> Cc: OpenDayLight-L2switch-Dev <l2switch-...@lists.opendaylight.org> >>> >> Subject: Re: [L2switch-dev] L2switch Li issues >>> >> >>> >> Correct, we will only change plugin in master, so we will only >>> consider: https://git.opendaylight.org/gerrit/#/c/33303/ >>> >> >>> >> Also Sai, do you know why this patch says it cannot be merged? >>> >> >>> >> BR/Luis >>> >> >>> >> >>> >>> On Apr 20, 2016, at 4:59 PM, Abhijit Kumbhare <abhijitk...@gmail.com> >>> wrote: >>> >>> >>> >>> Hi Sai, >>> >>> >>> >>> Please don't change the default on the stable/beryllium to the Li >>> version. We are only changing it to default in Boron. >>> >>> >>> >>> Thanks, >>> >>> Abhijit >>> >>> >>> >>> On Wednesday, April 20, 2016, Sai MarapaReddy < >>> sai.marapare...@gmail.com> wrote: >>> >>> Hi Luis, >>> >>> >>> >>> Following are the patches for Stable/beryllium & master for >>> L2switch. I have change feature.xml & pom.xml to correctly point the li >>> version. >>> >>> >>> >>> Stable/beryllium - https://git.opendaylight.org/gerrit/#/c/33304/ >>> >>> Master - https://git.opendaylight.org/gerrit/#/c/33303/ >>> >>> >>> >>> Could you please give it one more try with these patches ? >>> >>> >>> >>> Regards >>> >>> Sai MarapaReddy >>> >>> Brocade Communications >>> >>> >>> >>> On Fri, Apr 15, 2016 at 3:57 PM, Luis Gomez <ece...@gmail.com> >>> wrote: >>> >>> I think the table is more like this: >>> >>> >>> >>> Case 1: Current L2 switch (should point to OFP-He) & Current OFP >>> (default plugin is OFP-He) --> Should pass, current situation >>> >>> Case 2: L2 switch with patch (should point to OFP-Li) & Current OFP >>> (default plugin is OFP-He) --> Should pass if current L2switch patch >>> contained all fixes for Li plugin. >>> >>> Case 3: Current L2 switch (should point to OFP-He) & OFP with >>> patch (default plugin is OFP-Li, no -li features) --> Should pass if >>> L2switch did not care about OF plugin version >>> >>> Case 4: L2 switch with patch (should point to OFP-Li) & OFP with >>> patch (default plugin is OFP-Li, no -li features) --> This will never >>> work, you are pulling -li features which do not exist anymore in OFP. >>> >>> >>> >>> I tested 2 & 3. >>> >>> >>> >>>> On Apr 15, 2016, at 2:42 PM, Sai MarapaReddy < >>> sai.marapare...@gmail.com> wrote: >>> >>>> >>> >>>> Hi Luis, >>> >>>> >>> >>>> Please correct me if i am wrong here. >>> >>>> >>> >>>> You have checked below Case 3 & Case 4 and found errors. >>> >>>> >>> >>>> Case 3 errors :- >>> https://jenkins.opendaylight.org/releng/view/l2switch/job/l2switch-csit-1node-switch-only-boron/291/robot/report/log.html >>> >>>> Case 4 erros :- >>> https://jenkins.opendaylight.org/releng/view/l2switch/job/l2switch-csit-1node-switch-only-boron/296/robot/report/log.html >>> >>>> >>> >>>> L2Switch Patch - https://git.opendaylight.org/gerrit/#/c/33304/ >>> & https://git.opendaylight.org/gerrit/#/c/33303/ >>> >>>> >>> >>>> OFP Patch - https://git.opendaylight.org/gerrit/#/c/35892/ >>> >>>> >>> >>>> Current L2 switch code (with out patch ) points to >>> features-openflowplugin which is equivalent to OFP-he >>> >>>> >>> >>>> Case 1: Current L2 switch (should point to OFP-He) & Current OFP >>> (should point to OFP-He) --> Should pass >>> >>>> Case 2: L2 switch with patch (should point to OFP-Li) & Current >>> OFP (should point to OFP-He) --> Should Fail >>> >>>> Case 3: Current L2 switch (should point to OFP-He) & OFP with >>> patch (should point to OFP-Li) --> Should Fail >>> >>>> Case 4: L2 switch with patch (should point to OFP-Li) & OFP with >>> patch (should point to OFP-Li) --> Should Pass >>> >>>> >>> >>>> >>> >>>> >>> >>>> Regards, >>> >>>> Sai Reddy >>> >>>> Brocade Communications >>> >>>> >>> >>>> On Thu, Apr 14, 2016 at 4:03 PM, Luis Gomez <ece...@gmail.com> >>> wrote: >>> >>>> And this second test report is when I just apply ofplugin Li patch >>> [2]: >>> >>>> >>> >>>> >>> https://jenkins.opendaylight.org/releng/view/l2switch/job/l2switch-csit-1node-switch-only-boron/296/robot/report/log.html >>> >>>> >>> >>>> I see SAME issues as with l2switch Li plugin patch + some weird >>> topology display issue (internal topology works as otherwise ping would not >>> work but switch links are not shown) the second time we run Host Tracker >>> test. >>> >>>> >>> >>>> BR/Luis >>> >>>> >>> >>>> [2] https://git.opendaylight.org/gerrit/#/c/35892/ >>> >>>> >>> >>>> >>> >>>>> On Apr 13, 2016, at 7:32 PM, Luis Gomez <ece...@gmail.com> wrote: >>> >>>>> >>> >>>>> From today’s l2switch call, I got action point to run system test >>> on Li patch [1]: >>> >>>>> >>> >>>>> >>> https://jenkins.opendaylight.org/releng/view/l2switch/job/l2switch-csit-1node-switch-only-boron/291/robot/report/log.html >>> >>>>> >>> >>>>> From the result above we can observe following issues: >>> >>>>> >>> >>>>> - Flow push: There is no LLDP flows being pushed. Also flood flows >>> seem to be in config space because operational shows flow ID. >>> >>>>> - Address Tracker: All host IP addresses are seeing in all nodes, >>> that is not right, only node connected to host should show the host. >>> >>>>> - Host Tracker: After remove/add OpenFlow port (ovs-vsctl >>> add-port), no host is learned anymore through that port. >>> >>>>> - Loop Remover: Not working, using a loop topology, even when we >>> observe some ports are in DISCARD mode, mininet sees DUP packets. >>> >>>>> >>> >>>>> BR/Luis >>> >>>>> >>> >>>>> [1] https://git.opendaylight.org/gerrit/#/c/33303/ >>> >>>>> >>> >>>>> >>> >>>> >>> >>>> >>> >>>> _______________________________________________ >>> >>>> L2switch-dev mailing list >>> >>>> l2switch-...@lists.opendaylight.org >>> >>>> https://lists.opendaylight.org/mailman/listinfo/l2switch-dev >>> >>>> >>> >>>> >>> >>> >>> >>> >>> >> >>> >> >>> > >>> >> >> >> > >
_______________________________________________ openflowplugin-dev mailing list openflowplugin-dev@lists.opendaylight.org https://lists.opendaylight.org/mailman/listinfo/openflowplugin-dev
