14:06:31 <edwarnicke> #startmeeting OFplugin Daily 14:06:31 <odl_meetbot> Meeting started Thu Aug 7 14:06:31 2014 UTC. The chair is edwarnicke. Information about MeetBot at http://ci.openstack.org/meetbot.html. 14:06:31 <odl_meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:06:31 <odl_meetbot> The meeting name has been set to 'ofplugin_daily' 14:06:38 <edwarnicke> #info edwarnicke 14:06:39 <prasanna> #info prasanna 14:06:42 <kamal_rameshan> #info kamal_rameshan 14:06:44 <jmedved> #info jmedved 14:06:45 <oflibMichal> #info oflibMichal 14:06:45 <avishnoi> #info avishnoi 14:06:55 <hideyuki> #info Hideyuki 14:07:02 <edwarnicke> hideyuki: Welcome! :) 14:07:10 <hideyuki> edwarnicke: Thank you :) 14:07:49 <edwarnicke> Apologies... I am going to wing it a bit, because abhijitkumbhare was tracking things... and he's stuck in traffic 14:07:51 <edwarnicke> *but* 14:07:56 <edwarnicke> We have trello :) 14:08:09 <edwarnicke> #link https://trello.com/b/TtnyEJZr/helium-release - OFplugin Trello 14:09:43 <prasanna> #info moved testing gerrit@9393 to done 14:10:14 <edwarnicke> prasanna: Could folks tell us their results here of testing gerrit9393 14:10:55 <v_demcak> #info vdemcak 14:11:08 <edwarnicke> #link https://git.opendaylight.org/gerrit/#/c/9393/ 14:11:10 <jmedved> i have not had the opportunity to test it 14:11:13 <edwarnicke> So... who has tested https://git.opendaylight.org/gerrit/#/c/9393/ ? 14:11:21 <edwarnicke> And what is the result of their testing? 14:11:22 <prasanna> edwarnicke : yes, but my testing was completed , we will hear from others. 14:11:31 <prasanna> I have teste 14:11:43 <edwarnicke> prasanna: Thank you :) 14:12:20 <prasanna> I have verified 3 normal match calls and the NXM rest call(mentioned in wiki).....the switch was getting programmed correctly 14:12:34 <prasanna> But Stats for NXM may be the a issue 14:12:58 <prasanna> Not seeing NXM stats in postman 14:13:01 <edwarnicke> prasanna: Did you try non-extension stuff? 14:13:25 <prasanna> yes edwarnicke that is what is "3 normal match calls" 14:13:49 <prasanna> thats what I meant.....not clear in my earlier statement 14:13:49 <edwarnicke> And you see the normal flows in stats? 14:14:11 <prasanna> Yes normal flows are showing up in stats 14:15:05 <edwarnicke> jmedved: Did you test 9393 ? 14:15:14 <jmedved> no 14:15:30 <jmedved> i only got to test 8174 14:15:48 <jmedved> (which we need to talk about) 14:15:48 <edwarnicke> abhijitkumbhare: Welcome, please #info in 14:15:50 <abhijitkumbhare> #info abhijitkumbhare here 14:15:52 <avishnoi> i also tested 8174 14:16:03 <edwarnicke> Shall we start with 9393? 14:16:13 <abhijitkumbhare> Yes 14:16:14 <edwarnicke> #topic https://git.opendaylight.org/gerrit/#/c/9393/ 14:16:50 <edwarnicke> abhijitkumbhare: to summarize: prasanna has tested 9393, normal flows continue to work, nxm flows are having a stats issue. 14:17:04 <abhijitkumbhare> OK 14:17:25 <abhijitkumbhare> So - what kind of stats issue? 14:17:34 <edwarnicke> prasanna: Could you explain 14:17:47 <prasanna> abhijitkumbhare not able to see the NXM flows in rest 14:18:07 <abhijitkumbhare> In the operational side? 14:18:12 <prasanna> yes 14:18:43 <prasanna> not able to capture the wireshark....because the machine had some issues with tcpdump capture... 14:18:54 <edwarnicke> prasanna: Do you see the flows on the switch? 14:18:58 <prasanna> Yes 14:19:25 <edwarnicke> OK guys... shall we merge 9393? 14:19:41 <abhijitkumbhare> Yes 14:20:04 <avishnoi> +1 14:20:05 <abhijitkumbhare> We just let know we have this issue that is being worked on 14:20:49 <prasanna> +1 14:20:50 <abhijitkumbhare> Who will work on it? 14:21:00 <abhijitkumbhare> The stats issue? 14:21:17 <prasanna> abhijitkumbhare we will just confirm with Rob once 14:21:22 <edwarnicke> abhijitkumbhare: Agreed... who will create the bug and the trello? 14:21:24 <prasanna> regarding Stats 14:21:25 <abhijitkumbhare> OK 14:21:41 <abhijitkumbhare> prasanna will create the bug? 14:22:13 <prasanna> Since Madhu was also was reporting the stats issue , which I feel is same as what we are seeing 14:22:18 <edwarnicke> #action prasanna to file a bug for stats issue with 9393 14:22:20 <prasanna> Yes will create the bug 14:22:46 <edwarnicke> abhijitkumbhare: Will you create the trello? 14:22:50 <abhijitkumbhare> yes 14:23:30 <edwarnicke> #action abhijitkumbhare to create a trello for 9393 14:23:35 <abhijitkumbhare> I will update the Trello with the bug ID when prasanna has filed it 14:23:42 <edwarnicke> OK, who is going to work on the bug? 14:23:55 <abhijitkumbhare> if its a stats bug :-) 14:24:02 <edwarnicke> abhijitkumbhare: Sure 14:24:07 <edwarnicke> But who's going to work on it? 14:24:31 <abhijitkumbhare> do you have any cycles avishnoi ? probably next week? 14:24:32 <avishnoi> I can, but i will be able to do it after next tuesday 14:24:59 <avishnoi> yes, after next tuesday i can pick it 14:24:59 <jmedved> there is another stats bug - 1499. maybe they could be combined. not sure 14:25:10 <jmedved> work really needed on the stats 14:26:23 <avishnoi> I think vaclav is looking at 1499 14:26:36 <v_demcak> 1499 is fixed in 9788 14:26:43 <abhijitkumbhare> good 14:27:38 <prasanna> #info https://bugs.opendaylight.org/show_bug.cgi?id=1501 : created bug for NXM stats issue 14:27:55 <abhijitkumbhare> I will link it prasanna 14:28:17 <jmedved> v_demcak: excellent! :-) 14:30:44 <jmedved> can we move onto 8174? 14:30:55 <abhijitkumbhare> Have created the Trello for bug 1501 14:31:06 <abhijitkumbhare> yes jmedved 14:31:46 <jmedved> bottom line, as i described in the email: major stability improvement. IMO we can merge. did anybody test it? 14:31:54 <jmedved> avishnoi: you said you did? 14:31:59 <jmedved> what’s your resutls? 14:32:46 <v_demcak> :-) 1491 14:33:03 <v_demcak> sorry 14:33:04 <avishnoi> jmedved: edwarnicke : yes i tested it as well 14:33:22 <jmedved> did it work for you? 14:33:31 <avishnoi> i tested pingall scenario and verified that all the flows stats are there 14:33:44 <jmedved> great 14:33:45 <edwarnicke> Apologies 14:33:47 <avishnoi> i tested with rest conf as well 14:33:51 <avishnoi> it was positive 14:34:10 <avishnoi> and i tested it with the case for which we actually were using the comparators 14:34:15 <jmedved> i loaded with 10k flows, which pretty much killed the controller before, and it kept going. all flows got programmed 14:34:25 <jmedved> i +1 the merge 14:34:31 <jmedved> but 14:34:43 <avishnoi> and that also worked as expected. I have few question, which i discussed with vaclav and things look good to me 14:34:46 <jmedved> we still have more work to do on the overall performance & scale 14:34:48 <avishnoi> +1 for merge 14:34:53 <edwarnicke> Did we get a #action assigned for someone to work on but 1501 14:34:54 <edwarnicke> bug 14:36:59 <jmedved> if i see it correctly avishnoi said he will work on it after next tuesday. 14:37:31 <edwarnicke> #action avishnoi to work on https://bugs.opendaylight.org/show_bug.cgi?id=1501 for Tuesday 14:37:31 <abhijitkumbhare> I assigned 1501 to avishnoi 14:37:43 <edwarnicke> abhijitkumbhare: Can I just assign you getting the Trello's for these things? 14:38:00 <avishnoi> thanks 14:38:01 <abhijitkumbhare> I already added trello 14:38:01 <abhijitkumbhare> :-) 14:38:01 <edwarnicke> :) 14:38:07 <edwarnicke> Did we resolve stuff on 8174? 14:38:16 <jmedved> ewarnicke: not 100% 14:38:26 <jmedved> we agreed that it’s good to be merged 14:38:49 <jmedved> and that it improves stability in a major way 14:39:13 <edwarnicke> Who is reviewing and merging 9393 and 8174 ? 14:39:19 <jmedved> we still need more performance improvements. which i’d like to discuss now, because they may require some code reorg 14:39:23 <edwarnicke> Note: 9393 has a companion in integration that needs merge too 14:39:38 <edwarnicke> jmedved: Happy to, but lets get the actions for existing stuff pinned on people first 14:40:14 <jmedved> ewarnicke: ok, +1 on 8174 merge. AI should be on you or any other controller committer 14:40:53 <edwarnicke> #action edwarnicke to review (and unless there is a bad late breaking issue) merge https://git.opendaylight.org/gerrit/#/c/8174/ 14:40:55 <abhijitkumbhare> It will be good to get 9393 merged before the TSC meeting in a 2-3 hours - I have back to back meetings till then to finish the review and merge 14:41:22 <edwarnicke> I can do the merge for both 9393 and its companion integration gerrit 14:41:31 <abhijitkumbhare> great 14:41:32 <prasanna> rebased 9393 will merge it 14:41:34 <avishnoi> edwarnicke: i already reviewed statistics manager code and robert/tony already did +1 to this 14:41:53 <edwarnicke> #action edwarnicke to review (and unless there is a bad late breaking issue) https://git.opendaylight.org/gerrit/#/c/9393/ and https://git.opendaylight.org/gerrit/#/c/9614/ 14:43:09 <abhijitkumbhare> if prasanna manages to review/merge 9393 before edwarnicke does - that's also fine :-) 14:43:38 <prasanna> abhijitkumbhare I think 9614 needs to be merged by ed 14:43:52 <abhijitkumbhare> OK 14:44:15 <edwarnicke> prasanna: Lets coordinate on these 14:44:29 <edwarnicke> prasanna: Does 9393 require a manual rebase? 14:44:43 <prasanna> Sure edwarnicke no I was able to rebase from gerrit itself 14:44:48 <prasanna> Build is started 14:44:53 <edwarnicke> prasanna: Perfect 14:45:02 <edwarnicke> prasanna: Let me know when you are ready to merge it :) 14:45:14 <edwarnicke> jmedved: You wanted to change topic 14:45:14 <edwarnicke> ? 14:45:28 <jmedved> ewarnicke: yes please 14:45:40 <jmedved> performance (my favorit ;-) ) 14:45:41 <edwarnicke> jmedved: What topic? 14:45:47 <edwarnicke> #topic performance 14:46:06 <abhijitkumbhare> we should keep at least a few minutes for hideyuki's issue 14:46:23 <abhijitkumbhare> unless it has been discussed 14:46:23 <jmedved> we have two performance issues: 14:46:29 <jmedved> 1. flow programming rate 14:46:37 <jmedved> 2. stats collection 14:47:00 <edwarnicke> hideyuki: Do you have a bug for action order? 14:47:17 <edwarnicke> jmedved: Do we have performance issues on topology (asking, don't know) 14:47:18 <jmedved> Re #1: when mininet is connected, the flow prgramming rate i observed is about 7 (seven) flows per second 14:47:40 <hideyuki> edwarnicke: yes. https://bugs.opendaylight.org/show_bug.cgi?id=1338 14:47:48 <jmedved> with quite high CPU utilization 14:48:42 <prasanna> jmedved did we profile ? 14:48:58 <jmedved> whne mininet is not connected, the flow programming rate is ~350 flows/sec. not great, but probably not that difficult to improve, since most of the time is spent in restconf (tony is doing new codecs which will fix that) and only one core is utilized 14:49:02 <jmedved> yes 14:49:18 <jmedved> i will send out the profiler output for both cases 14:49:49 <prasanna> jmedved that would be gr8 to identify the time spent in different modules ...we know netty for sure...:) 14:50:40 <prasanna> jmedved Thanks will wait for the mail 14:50:46 <edwarnicke> hideyuki: I have created a trello card for it https://trello.com/b/TtnyEJZr/helium-release 14:50:48 <jmedved> but for the case where mininet is connected, the rate is outrageously low, which we need to improve. my first question is: do we program the flows from FRM into the plugin one by one synchronously? i.e waiting for the programming of a previous flow to finish before we start programming the next flow? 14:51:10 <jmedved> who knows about the FRM? 14:51:55 <prasanna> No I do think so ... 14:52:11 <jmedved> ok - so somebody needs to go in and analyze 14:52:12 <kamal_rameshan> jmedved are the flows written via restconf 14:52:14 <kamal_rameshan> ? 14:52:18 <jmedved> yes 14:52:46 <kamal_rameshan> then i dont think so that they are queued 14:52:58 <jmedved> as i said, i did 2 runs: one with mininet connected, one where mininet was NOT connected, and the flows were only written into the data store 14:53:19 <jmedved> with mininet connected, the programming rate is 7 flows/sec 14:53:28 <abhijitkumbhare> edwarnicke & hideyuki - is this different Trello card from the one for "Fix behavior of SET_VLAN_ID for openflow 1.3 switches" with bug https://bugs.opendaylight.org/show_bug.cgi?id=1421 ? 14:53:30 <edwarnicke> #nexttopic (ie, we aren't switching yet... but when we do): Action ordering bug... getting it assigned to someone 14:53:33 <jmedved> with data store only, the prgramming rate is 350 flows/sec 14:53:55 <jmedved> yep - who can work wuth me on this? needs an FRM expert 14:53:58 <prasanna> jmedved "and the flows were only written into the data store" does it mean the call get terminated in MD-SAL with a error no switch registered? 14:54:16 <edwarnicke> OK guys... we are close on time 14:54:24 <jmedved> prasanna: i don’t think there is an error 14:54:26 <edwarnicke> What are the concrete actions for tomorrow on performanace? 14:54:29 <jmedved> but yes, that’s the gist of it 14:54:30 <edwarnicke> And who is taking them? 14:54:54 <jmedved> ewarnicke: i will send out the profiler outputs for both runs 14:55:11 <jmedved> but, as is said, we need an FRM expert to start looking at this 14:55:16 <jmedved> who is that? 14:56:13 <prasanna> I can work with jmedved but I am not in office next week 14:56:24 <edwarnicke> prasanna: Mind if I push just a commit message change to 9393... I don't think merging something with DO NOT MERGE is good 14:56:40 <prasanna> edwarnicke will change it thanks 14:57:21 <edwarnicke> prasanna: Cool thanks :) 14:58:02 <hideyuki> abhijitkumbhare: "Fix behavior of SET_VLAN_ID for openflow 1.3 switches" and "Bug 1421" are different problems. 14:58:12 <abhijitkumbhare> We are running out of time - and we have not discussed hideyuki's issues yet 14:58:16 <abhijitkumbhare> OK 14:58:34 <edwarnicke> #topic https://bugs.opendaylight.org/show_bug.cgi?id=1421 14:58:40 <edwarnicke> Who can work on this one 14:59:01 <edwarnicke> there is a fix here https://git.opendaylight.org/gerrit/#/c/9214/ which may even work out of the box 14:59:06 <edwarnicke> But someone needs to test it 14:59:11 <kamal_rameshan> i can take it 14:59:44 <kamal_rameshan> if somebody else is not already working on it 14:59:51 <edwarnicke> #action kamal_rameshan takes https://bugs.opendaylight.org/show_bug.cgi?id=1421 with candidate fix at https://git.opendaylight.org/gerrit/#/c/9214/ 15:00:17 <edwarnicke> kamal_rameshan: I added you to the trello 15:00:41 <kamal_rameshan> edwarnicke thanks 15:00:41 <hideyuki> Gerrit 9214 is not for Bug 1421, it is for Bug 1338, right? 15:01:14 <edwarnicke> hideyuki: Did I get the bugs wrong? 15:01:21 <edwarnicke> I had asked about the action order one 15:01:39 <edwarnicke> hideyuki: You are right, 1421 is the wrong bug for action order 15:01:57 <hideyuki> edwarnicke: thank you. 15:02:44 <edwarnicke> I moved it to OFplugin and assigned it to Kamal 15:02:51 <edwarnicke> Now to fix the trello 15:03:28 <edwarnicke> Someone already fixed the trello :) 15:03:36 <edwarnicke> #undo 15:03:36 <odl_meetbot> Removing item from minutes: <MeetBot.ircmeeting.items.Action object at 0x272e5d0> 15:04:07 <edwarnicke> #action kamal_rameshan takes https://bugs.opendaylight.org/show_bug.cgi?id=1338 with candidate fix at https://git.opendaylight.org/gerrit/#/c/9214/ 15:04:07 <kamal_rameshan> hideyuki: as i understand there are 2 issues: 1. action-order-issue (1338) and 2. push-vlan-set-field issue with 1.3 and 1.0 15:04:14 <kamal_rameshan> am i correct? 15:04:34 <hideyuki> And, Bug 1491. 15:04:39 <edwarnicke> hideyuki: I am not dismissing the push_vlan... I can just get someone assigned and working on order easier :) 15:04:39 <hideyuki> There are three issues. 15:06:47 <edwarnicke> hideyuki: What is the third? 15:06:54 <kamal_rameshan> 1338, 1421 and 1491? 15:06:59 <hideyuki> Three issues: 1. action-order-issue (1338), 2. push-vlan-set-field issue with 1.3 and 1.0 (1421), and 3. SET_TP_SRC/SET_TP/DST issue (1491) 15:07:02 <hideyuki> kamal_rameshan: yes. 15:07:17 <hideyuki> https://bugs.opendaylight.org/show_bug.cgi?id=1491 15:09:04 <kamal_rameshan> if i am not wrong they all seem to be related....if yes then this has to be viewed hoslitically and we can come with a solution 15:10:38 <hideyuki> kamal_rameshan: Yes, these are related. 15:12:31 <kamal_rameshan> hideyuki: thanks... 15:15:32 <abhijitkumbhare> So kamal_rameshan - are you looking at all three? 15:18:17 <kamal_rameshan> abhijitkumbhare i will start analyzing all three... and see there is a solution, since they seem to be related. 15:18:25 <abhijitkumbhare> OK 15:19:10 <kamal_rameshan> has the meeting ended? 15:19:43 <hideyuki> kamal_rameshan: Thank you! 15:21:00 <abhijitkumbhare> Yes - edwarnicke can you do #endmeeting? 15:21:06 <abhijitkumbhare> I will try as well 15:21:12 <abhijitkumbhare> #endmeeting