09:01:12 <gampel> #startmeeting dragonflow 09:01:13 <openstack> Meeting started Mon Mar 14 09:01:12 2016 UTC and is due to finish in 60 minutes. The chair is gampel. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:01:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:01:16 <openstack> The meeting name has been set to 'dragonflow' 09:01:32 <gampel> hi everyone who is here ? 09:01:36 <raofei> hi everyone 09:01:38 <oanson> Hi. 09:02:01 <gampel> #info oanson raofei in the meeting 09:02:26 <gampel> gal will not be able to join us today :-( 09:02:48 <gampel> is there anyone else here today ? 09:03:32 <gampel> yuli , duankebo ? 09:03:35 <raofei> oh, hope there is no impact. 09:03:45 <yuli_s> i am here 09:04:12 <DuanKebo> HI 09:04:19 <gampel> I think that we will mange we will cover with him his part latter 09:04:23 <gampel> welcome 09:04:39 <gampel> #info DuanKebo, yuli_s in the meeting 09:05:25 <gampel> are we waiting for some one else , is nick-ma joining ? 09:05:47 <raofei> Hi Duankebo, does Yuanwei need to join the meeting for sg feature? 09:06:05 <nick-ma> i'm here. 09:06:20 <gampel> welcome 09:06:32 <gampel> #info nick-ma in the meeting 09:06:38 <DuanKebo> Yuawei is answering the pone. 09:06:43 <gampel> #topic Bugs 09:06:49 <gampel> #topic Bugs 09:07:03 <gampel> @omer and @yuli can you please update on this topic 09:07:13 <oanson> Yes 09:07:15 <yuli_s> Sure 09:07:17 <oanson> We went over the bugs. 09:07:30 <oanson> We have given each bug a priority. 09:07:48 <yuli_s> Nad closed some fixed ones 09:07:59 <oanson> Obviously if someone thinks a bug is more or less important than what we said he should let @yuli_s know 09:08:23 <raofei> ok 09:08:25 <oanson> Most bugs have assignees, but if anyone wants to volunteer to solve a bug, feel free :) 09:08:26 <nick-ma> ok. 09:08:41 <gampel> are there any important bugs that are not assign ? 09:08:50 <oanson> There are a few bugs for which there are no assignees, and they're set to high. 09:09:07 <yuli_s> oanson: can you paste the list pls 09:09:25 <oanson> Specifically, #1495432 #1517049 and #1527220 need investigation, and owners to investigate them 09:10:03 <gampel> this bugs are open for grab any takers 09:10:03 <oanson> nick-ma, #1529326 and #1529812 are related to db consistency. Is there one bug that unifies all these issues? Since you are rewriting it according to a new spec? 09:10:16 <nick-ma> yes. 09:10:40 <nick-ma> the spec is to fix them all. 09:10:51 <DuanKebo> Zhu weiqiong can work in some bugs. 09:11:03 <yuli_s> we have api performance degradation bug , #1556807 09:11:29 <gampel> ok thats great please look at the bugs list 09:12:10 <gampel> #action everyone go over the high priority bugs and try to take what he can 09:12:18 <oanson> Bug 1548029 also needs an owner 09:12:18 <openstack> bug 1548029 in DragonFlow "Deadlock when allocating IP" [High,New] https://launchpad.net/bugs/1548029 09:13:03 <gampel> Yuli: you are taking #1556807 09:13:47 <gampel> Everyone please go over the bugs 09:13:53 <nick-ma> ok 09:13:56 <oanson> I think that's it for High importance bugs. 09:14:11 <oanson> If anyone sees a bug assigned to someone else, don't feel shy to take over :) 09:14:18 <gampel> is there any additional issues with any of the BUGs ? 09:14:34 <raofei> ok:) 09:14:48 <gampel> #topic DB Consistency 09:14:50 <yuli_s> gampel: I will dig deeper in #1556807 09:14:59 <gampel> #link DB Consistency https://review.openstack.org/#/c/282290/ 09:15:18 <gampel> nick-ma: do you what to update the stage is yours 09:15:34 <nick-ma> I updated the spec and code to use a simple lock to fix the race. 09:16:03 <nick-ma> I also removed the versioned object design, because I think it is not for the current stage and need a spec to discuss about it. 09:16:14 <gampel> ok so we do not do the version per resource in the first phase 09:16:36 <gampel> so you think that we should move this to the n cycle 09:16:37 <nick-ma> yes, we can iterate based upon the current implementation. 09:16:54 <gampel> Ok I agree we do not have much time 09:16:58 <nick-ma> for db consistency, it is not mandatory. 09:17:26 <gampel> we need to look why the rally test are failing if it is related 09:18:18 <nick-ma> I checked the rally. it is about deleting default sg. 09:18:34 <nick-ma> which is really weird. 09:18:54 <gampel> Ok so please everyone review the db consistency path I think it is high priority 09:19:08 <gampel> lets try to merge it as soon as possible 09:19:12 <nick-ma> yes, I tested it using rally in my lab, it is working. 09:19:25 <nick-ma> I will figure out the gate failure. 09:19:33 <gampel> yuli_s: can you run nick-ma patch on your local rally test 09:19:43 <yuli_s> sure 09:20:19 <raofei> ok, will do 09:20:20 <gampel> #action yuli_s test the DB consistency path with his local scripts (the control path testing ) 09:20:20 <yuli_s> I will talki with nick-ma about his tests 09:20:31 <nick-ma> ok. 09:20:45 <gampel> nic-ma: any other issues with this topic ? 09:21:05 <nick-ma> no. 09:21:11 <gampel> #topic Publish-Subscribe 09:21:32 <gampel> we are missing reliability and service discovery table 09:21:54 <gampel> omer: will you be able to take the service discovery table 09:22:02 <oanson> Sure 09:22:20 <gampel> reliability is dependent on the resource version 09:22:49 <gampel> currently we do not send the value but get it fro the controller just send notification of a change 09:22:50 <raofei> Can you please explain what's the reliability table? 09:23:04 <gampel> it is the service discovery table 09:23:26 <raofei> OK. 09:23:35 <gampel> currently pubsub subscriber take the publisher IPs for the config file 09:23:54 <gampel> this is realted only to non broker solution like zmq 09:24:25 <gampel> I think that the reliability will be delayed to n cycle :-( 09:25:10 <gampel> #link pubsub spec https://review.openstack.org/#/c/263733/ 09:25:25 <oanson> I can try taking that as well - at least verifying that no messages are missed from publishers 09:25:29 <gampel> i will rebase it today please everyone review 09:25:35 <oanson> (i.e. publisher UUID + message ID) 09:25:39 <nick-ma> ok 09:25:48 <gampel> #action omer do the service discovery for the publishers 09:26:03 <gampel> #topic Redis Driver 09:26:17 <gampel> #link redis https://review.openstack.org/#/c/274340/ 09:26:37 <gampel> @feipeng is here 09:26:59 <gampel> DuanKebo: can you update ? 09:27:07 <DuanKebo> OK 09:27:38 <DuanKebo> the code is completed. He is adding some ut 09:27:45 <DuanKebo> and fullstack test code. 09:28:00 <gampel> it is missing the devstack plugin update for redis 09:28:05 <nick-ma> do not forget to add devstack script, that greatly helps deployment. 09:28:15 <gampel> :) 09:28:39 <DuanKebo> Yes. 09:28:48 <gampel> fullstack is currently configured with zmq 09:29:23 <gampel> we need to consider adding additional CI or do unit testing for the redis pubsub 09:29:27 <DuanKebo> For the code, he has fixed all the comments. 09:29:49 <gampel> ok does he want to have the devstack in the same patch ? 09:30:10 <gampel> if not this patch need to be dependent on it 09:31:21 <gampel> #action feipeng i add devstack and testing to the redis patch 09:31:53 <gampel> lets try to merge it this week 09:32:14 <gampel> let us know if you need any help with the devstack plugin 09:33:01 <gampel> #topic Selective Proactive and OVSDB Monitor 09:33:11 <gampel> DuanKebo: the stage is yours 09:33:20 <gampel> #link Selective Proactive https://review.openstack.org/#/c/282945/ 09:33:30 <gampel> #link OVSDB Monitor https://review.openstack.org/#/c/281687/ 09:33:51 <DuanKebo> there are some new review comments 09:34:06 <DuanKebo> Hujie will renew the code 09:34:31 <gampel> I will go over it again today i think this is very high priority to merge and test 09:35:07 <DuanKebo> Yes. 09:35:26 <DuanKebo> we are building the fullstack envirment 09:35:38 <gampel> are there any open issues , as i wrote i think that local_interface or just interface are better then linterface for the table name 09:36:04 <gampel> when do you think you will have it ready ? 09:37:17 <gampel> DuanKebo: any open issues or you think we could merge this week 09:38:47 <DuanKebo> We need to merge 09:38:57 <DuanKebo> ovsdb monitor, redis 09:39:06 <DuanKebo> and selective topo 09:39:11 <gampel> Ok lets everyone try to review this two patches as soon as possible 09:39:15 <DuanKebo> asap. 09:39:27 <gampel> #topic Distributed DNAT 09:39:31 <DuanKebo> yes. 09:39:37 <gampel> #link D DNAT https://review.openstack.org/#/c/289148/ 09:39:45 <nick-ma> ok 09:39:52 <raofei> please everyone review it. 09:39:52 <gampel> raofei: the stage is yours 09:40:05 <gampel> any issues ? 09:40:07 <raofei> The 1st stage, distributed-DNAT only support VXLAN network. The next stage it will support VLAN network. 09:40:24 <raofei> no, just some merge conflict. 09:40:38 <gampel> it is any tunneling right gre vxlan ... 09:41:30 <nick-ma> how about geneve? 09:41:31 <raofei> yes. 09:41:45 <nick-ma> it is the default in df. 09:41:53 <gampel> again we need every one to review as much as he can not only his patches 09:41:54 <raofei> it can support. I think all the overlay tunnel 09:42:02 <nick-ma> ok 09:42:09 <raofei> current DNAApp doesn't support send arp request periodically 09:42:30 <raofei> I will optimize this point. 09:42:34 <gampel> for the pub IP , will you add this 09:42:40 <gampel> great 09:42:51 <raofei> yes 09:43:07 <gampel> #action raofei add another patch for sending periodic arp 09:43:32 <gampel> any other issues on D DNAT 09:44:08 <raofei> current there is no issue. please test and review the code. 09:44:14 <raofei> help to find issues. 09:44:24 <gampel> #topic Security groups and port security 09:44:34 <gampel> #link https://review.openstack.org/#/c/280538/ 09:44:41 <gampel> who can update ? 09:45:39 <gampel> raofei ? DuanKebo ? 09:45:50 <DuanKebo> Dingbo has submit 09:45:54 <DuanKebo> part of the code 09:46:10 <raofei> yes, Dingbo will complete this feature this week. 09:46:13 <DuanKebo> and he well sumbit rest of code later. 09:46:33 <gampel> do you have an estimate when ? 09:47:14 <DuanKebo> I think the week 09:47:31 <raofei> Can you please verify gample? 09:47:52 <raofei> @Dingbo ,Can you please verify gample? 09:48:10 <dingboopt> ok 09:48:20 <gampel> hi welcome 09:48:27 <nick-ma> hi 09:48:30 <gampel> can you please update on SG 09:48:47 <dingboopt> I submit code for review 09:49:05 <dingboopt> the test code is being writted 09:49:22 <dingboopt> I will update it soon 09:49:27 <gampel> there are few comment all ready and me and Omer will go over it again today 09:49:42 <dingboopt> thanks 09:49:46 <gampel> every one please review this is a big patch 09:49:55 <dingboopt> I will take a look 09:50:01 <raofei> ok 09:50:19 <DuanKebo> yes, i will review it also. 09:50:26 <gampel> ok lets try to merge this one as soon as possible as well so we could start testing SG 09:50:40 <gampel> #topic testing and QA 09:50:48 <DuanKebo> dingboopt, please upload the rest code asap 09:51:13 <gampel> DuanKebo, raofei are you doing any scale and feature test on your side 09:51:59 <DuanKebo> yes 09:52:14 <DuanKebo> we are preparing a perf test envirnment. 09:52:38 <gampel> thats great , do you have any thing to share ? 09:52:58 <DuanKebo> not yet. 09:52:59 <gampel> did you talk with @shlomo about his testing 09:53:23 <raofei> I just did the the functionality test, the performance test depends on the test setup. 09:53:33 <DuanKebo> we are focus on control plane currently 09:53:35 <raofei> of course, some test tools will be used to do. 09:54:00 <gampel> this is good @yuli is working on it as well 09:54:13 <gampel> maybe you could work together on it 09:54:15 <gampel> yuli: do you what to share the control path testing you are writing 09:54:27 <yuli_s> sure 09:54:36 <DuanKebo> we will work togher with yuli 09:54:43 <yuli_s> the spec is here https://review.openstack.org/#/c/282873/ 09:54:50 <yuli_s> i started to do the tests 09:54:53 <DuanKebo> he is also woking on controll plane test. 09:55:05 <yuli_s> these are the results of my tests so far 09:55:13 <gampel> #action DuanKebo and yuli work together on the control path testing 09:55:21 <yuli_s> https://bugs.launchpad.net/dragonflow/+bug/1556807 09:55:21 <openstack> Launchpad bug 1556807 in DragonFlow "Performance degradation when multiple subnets added" [High,New] - Assigned to Yuli (stremovsky) 09:55:34 <yuli_s> https://bugs.launchpad.net/dragonflow/+bug/1556598 09:55:34 <openstack> Launchpad bug 1556598 in DragonFlow "ovs-vswitchd too many open files" [High,New] - Assigned to Yuli (stremovsky) 09:55:40 <gampel> #link control path testing patch https://review.openstack.org/#/c/282873/ 09:56:20 <gampel> we will increase the ulimit for the ovs vswitchd 09:56:48 <gampel> we do not have much time , any other issues ? 09:57:05 <gampel> lets try to collaborate on the testing 09:57:29 <raofei> ok 09:57:46 <gampel> any other comments issues suggestions ? 09:58:23 <gampel> if not thank you everyone very good job 09:58:37 <nick-ma> thanks 09:58:41 <gampel> I think the M release will be very good 09:59:11 <nick-ma> of course. bye. 09:59:19 <gampel> bye 09:59:23 <gampel> #endmeeting