05:30:15 #startmeeting tacker 05:30:16 Meeting started Wed Apr 19 05:30:15 2017 UTC and is due to finish in 60 minutes. The chair is gongysh. Information about MeetBot at http://wiki.debian.org/MeetBot. 05:30:17 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 05:30:19 The meeting name has been set to 'tacker' 05:30:32 gongysh, tbh hi 05:30:43 hi janki 05:30:43 #topic roll call 05:31:05 o/ 05:31:11 who are you guys here for tacker meeting? 05:31:18 hi, janki tbh 05:31:24 o/ 05:31:27 o/ 05:31:28 o/ 05:31:38 o/ 05:32:04 6 of us. 05:32:27 #topic BP 05:32:48 for the vim rechability monitor 05:32:57 I am trying to fix it via mistral 05:33:13 https://review.openstack.org/#/c/457140/ 05:33:24 it is still in progress. 05:33:39 gongysh: +1 for a mistral based approach ! 05:33:49 https://review.openstack.org/#/c/457843/ 05:34:28 since we will use mistral lots, I think we need to pull up mistral related codes into a top dir. 05:34:54 tung_doan, you said it is not good for separating vnfo and vnfm. I don't think so. 05:35:43 sridhar_ram, yes, I agree not to introduce more third party component, such as zookeeper if we can do on current components. 05:36:08 tung_doan: gongysh: what is the context for such a statement - nfvo & vnfm separation? 05:36:15 gongysh: ack! 05:36:40 tung_doan, do you have any thing to say about 'seprate nfvo and vnfm"? 05:36:53 gongysh: Mistral is one of openstack drivers, I just think we should consider where we should place mistral driver :) 05:37:45 tung_doan, mistral will be a part of tacker, not a driver. 05:38:02 we have vim driver, not workflow driver. 05:38:04 tung_doan: gongysh: we need to clearly differentiate what belong to VIM driver and what is tacker's functional dependency 05:38:22 mistral is tacker's functional dependency (just like keystone) 05:38:31 o/ 05:38:37 it has nothing to do with target VIM driver .. 05:38:49 sridhar_ram, yes, it is so at least until now. 05:38:57 currently the code is poorly organized .. hence the confusion 05:39:03 sridhar_ram: gongysh : ok. got it 05:39:35 i started an attempt to cleanup in https://review.openstack.org/#/c/450487/ 05:39:51 .. but need to find some time to finish that effort 05:40:21 this is good. 05:40:49 sridhar_ram, you should push a new patch at midnight. :) 05:41:05 gongysh: :) 05:41:25 tung_doan, what is your progress on mistral for vnf policy? 05:42:08 tung_doan, hi 05:42:29 gongysh: i still look into this feature: https://specs.openstack.org/openstack/mistral-specs/specs/newton/approved/event-notification-trigger.html 05:43:04 tung_doan, ok, if it is complex, you can put a spec on it. 05:43:18 gongysh: yes. thanks :) 05:43:42 also I am thinking about our customized 'mistral action' in tacker. 05:44:21 gongysh: for what purpose ? 05:44:27 I think one principle is: not use db connection in action code since the action code is running at mistral engine. 05:44:58 sridhar_ram, to use mistral, means we need to have our own actions. 05:45:22 sridhar_ram https://review.openstack.org/#/c/457140/2/tacker/mistral/mistralactions/pingaction.py 05:45:25 i get that, is it intended for scaling, respawn, etc? 05:46:05 sridhar_ram, https://specs.openstack.org/openstack/mistral-specs/specs/newton/approved/event-notification-trigger.html 05:46:26 tung_doan is looking into it to see if vnf policy can use this mistral feature. 05:46:44 okay, good.. i like the approach.. 05:46:53 gongysh: +1 05:47:08 ok, move to next topic 05:47:17 YanXing_an, hi 05:47:26 gongysh, hi 05:47:34 I have seen you have updated the barbican spec. 05:47:43 https://review.openstack.org/#/c/445543/ 05:48:23 I think it is almost done. we can get it merged soon. 05:48:55 since dharmendra kushwaha is not online, sridhar_ram, can you review it? 05:49:06 I can give my +2. 05:49:21 gongysh: will review.. looks close to me 05:49:32 During coding, i can update the merged spec later? 05:49:44 sridhar_ram, thanks 05:49:46 if something changes 05:50:11 YanXing_an, no need to update the spec, spec is just drafting overview. 05:50:40 we allow some gap between spec and implementation. 05:51:12 gongysh, got it. 05:51:25 storage BP 05:52:00 https://review.openstack.org/#/c/453442/ 05:52:14 is zhou zhihong online? 05:52:32 ZhouZhihong is on business, he will update the patch soon according the review comments 05:52:34 that spec is not updated last few days. 05:52:54 API spec 05:52:58 janki, 05:53:11 I have see your first draft framework 05:53:41 sorry, it is from diga 05:53:43 https://review.openstack.org/#/c/442887/ 05:53:55 it is very basic. 05:54:04 gongysh, Mine is containarization one 05:54:04 need more work. 05:54:18 janki, do you have a spec on it? 05:55:04 gongysh, https://review.openstack.org/#/c/397233/ 05:55:39 gongysh, need to update it. WIll do it this week 05:55:40 janki, wow, a lot of reviews. 05:55:48 janki, thanks 05:56:03 janki, can you move it into pike? 05:56:26 gongysh, ya. will give it best shot 05:56:34 when is the pike release? 05:56:45 janki: try reducing the scope .. 05:56:56 .. it will help to make it to pike 05:57:05 sridhar_ram, as in? 05:57:20 janki, p3 is on July 05:57:26 I am currently targetting just spawning vnf as container. no scaling or anything 05:57:28 24 05:57:46 gongysh, good amount of time in hand. I will do it 05:57:47 janki: make some assumptions - just VDU support using Heat 05:57:48 janki, but you should think about the scaling. 05:58:07 gongysh, scaling can come in second iteration 05:58:09 to avoid later large redo. 05:58:42 the design shd factoring in things like scaling but i think implementation should experiment with limited scope to make it 05:58:57 sridhar_ram, +1 05:59:29 sridhar_ram: agree. i hope to see this feature in pike 05:59:29 lets move to next topic 05:59:40 #topic bugs 06:00:04 I thing I ran into a bug which causes all functional test fail. 06:00:28 the trial fix is at https://review.openstack.org/#/c/457852/ 06:01:00 it seem devstack removs the 5000 port? 06:01:11 but I need to watch more. 06:01:38 do you guys have any other bugs to talk about? 06:02:14 vim-update fix from trinaths 06:02:37 yes, https://review.openstack.org/#/c/449956/ 06:02:43 trinaths it is a long story 06:02:52 gongysh: very long story 06:03:11 also, the BP, https://review.openstack.org/#/c/455188/ 06:03:13 trinaths: gongysh: let's write an ending to that story here :) 06:03:46 sridhar_ram: :) 06:04:06 sridhar_ram, please start with your ideas 06:04:18 trinaths: gongysh: bottomline - we need to support update of all attrs except auth_url 06:04:44 name/desc/is_default will result in local db level update / validation 06:05:04 anything more shd result in register-vim 06:05:48 sridhar_ram: then, will this bug also, redirects to vim-config.yaml validation. 06:06:40 trinaths: yes, it needs to .. it seems the above logic is poorly implemented and needs to be fixed. 06:07:03 it is better to fix all vim-update bugs as part of this 06:07:48 sridhar_ram, agree, we should make the vim-update works just as the api defined. 06:07:53 sridhar_ram: okay. Let have vim-update bugs detailed of their issues and be fixed. that way we can have certain scope of fixes and reviews. 06:08:32 trinaths: thanks.. didn't mean to do a scope-creep :) but this will help us all to knock this one off 06:08:49 sridhar_ram: agree. 06:09:30 sridhar_ram: we need to include deep_update modification to this commit ? 06:09:39 YanXing_an, seems jimmy is not here. 06:10:10 trinaths: i'm fine removing it as long as the original functional intent as i described is taken care 06:10:37 some vnffg bugs are still hanging there. 06:10:40 gongysh, jimmy is also on business for long time 06:10:59 gongysh: sridhar_ram: scaling and monitoring are not working now. The reason is tosca-parser was released last week 06:10:59 sridhar_ram: ok. we need to move all the changes from nfvo_db to nfvo_plugin.? 06:11:00 https://review.openstack.org/#/c/451827/ 06:11:24 trinaths: yes, let's consolidate validation in nfvo_plugin 06:11:40 sridhar_ram: okay. agree. 06:12:10 YanXing_an, could you have a look at this patch? 06:13:22 gongysh: i will look at this patch 06:13:31 ok, thanks 06:13:38 tbh, hi 06:13:48 about ns with vnffg bp 06:14:05 gongysh: sridhar_ram: this BP requires your review. https://review.openstack.org/#/c/455188/ 06:14:34 I haven't checked that gongysh assuming dkushuwaha looking into it 06:14:36 tbh, aren't you work with dharmendra kushwaha? 06:14:51 trinaths: i'm already on it .. please see the email thread in ML: http://lists.openstack.org/pipermail/openstack-dev/2017-April/115601.html 06:14:52 gongysh, not yet, I will look into it 06:15:15 jay has some suggestion .. that i latched on .. please see the email thread 06:15:34 sridhar_ram: yes. With the ML discussion, I have updated the BP. 06:16:02 trinaths: cool.. you are fast! will take a look 06:17:19 sridhar_ram: I'm observing the ML and doing the changes as appropriate. Once the BP is approved, I can start the coding. planning to target some command-plugins to P-3. 06:17:38 trinaths: sounds good .. 06:17:54 sridhar_ram, who is responding in the mail? 06:18:18 gongysh: what do you think about the 'openstack vnf' suggestion? It is a reasonable compromise IMO 06:18:21 who is Jay Pipes jaypipes at gmail.com ? 06:18:29 I need gongysh to bless me for https://review.openstack.org/#/c/455188/ 06:18:50 I am thinking if he knows what is vnf, what is nfv. 06:19:02 gongysh: Jay Pipes .. OpenStack veteran, ex-TC member... nova contributor 06:19:10 gongysh: https://launchpad.net/~jaypipes 06:19:18 sridhar_ram, he is not from nfv domain. 06:20:04 all guys know SDN & NFV, not SDN & VNF. 06:20:06 gongysh: he is quite knowledgable in NFV .. though we significantly disagree with him on few things .. in a healthy way :) 06:20:31 using vnf as prefix, it is narrowing our project's scope. 06:20:40 we are doing nfv, not vnf. 06:20:59 gongysh: the CLI keyword doesn't determine our project's scope 06:21:32 keep in mind the idea with OSC is to make openstack behave as "one product" using all the features coming from different projects 06:22:13 the keyword just determines the "resource" that is being created using tacker project .. 06:22:33 vnf, collection of vnfs, graph on top of vnfs, etc. 06:22:34 sridhar_ram, osc commands are what users are see what our tacker does. 06:23:14 all commands are 'vnf xxx', it seems we are doing vnf, not nfv. 06:23:29 gongysh: yes, for demo purposes .. but in real world deployments OSS/BSS will invoke Tacker API 06:24:23 we have to vote at next meeting 06:24:36 gongysh: it doesn't negate us in any way we are doing NFV by using vnf keyword 06:25:19 sridhar_ram, I even cannot know what is mean by 'vnf service' 06:25:51 sorry got late 06:26:25 all our tacker documents and industry standards is network service, not vnf service. 06:26:32 gongysh: it is just the restriction coming from being an openstack citizen 06:26:57 ok, let move it to tacker channel. 06:27:10 #topic open discussion 06:27:27 we have 3 mins. 06:27:31 what is the plan for tacker related activities in openstack boston summit? 06:27:47 i'm still tentative attending the summit :( 06:27:47 I have started pushing patches on api-framework branch, please take a look at it 06:27:54 sridhar_ram, we have a forum room booked for us. 06:28:02 I have changed the directory structure for API 06:28:15 who is planning to attend the summit? 06:28:21 me 06:28:55 I am not able to make it this time 06:28:59 anyone else ? 06:29:24 i guess rest is all waiting to go to australia ;-) 06:30:00 if just me is on the summit, I have to cancel the room. 06:30:34 diga, I will put my eyes on the api patches. 06:30:45 #endmeetings 06:30:49 gongysh: ok 06:30:49 #endmeeting