05:30:49 <sridhar_ram> #startmeeting tacker 05:30:50 <openstack> Meeting started Wed Jan 11 05:30:49 2017 UTC and is due to finish in 60 minutes. The chair is sridhar_ram. Information about MeetBot at http://wiki.debian.org/MeetBot. 05:30:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 05:30:53 <openstack> The meeting name has been set to 'tacker' 05:30:57 <sridhar_ram> #topic Roll Call 05:31:13 <diga> o/ 05:31:18 <sridhar_ram> who is here for tacker meeting? 05:31:23 <xuhaiwei_> o/ 05:31:23 <Yanxing_an> o/ 05:31:34 <dkushwaha> o/ 05:32:15 <sridhar_ram> howdy all! 05:32:23 <sridhar_ram> let's give a min before we start 05:33:14 <sripriya> o/ 05:33:28 <sridhar_ram> alright, let's start 05:33:42 <sridhar_ram> #topic Ocata Implementation Status 05:33:50 <sridhar_ram> #topic NSD 05:34:00 <sridhar_ram> this is the highest priority 05:34:13 <sridhar_ram> dkushwaha: any blocking issues? 05:34:25 <sridhar_ram> dkushwaha: what is pending to finish for NSD ? 05:35:18 <dkushwaha> sridhar_ram, not blocking, I am going through FT/UT parts now 05:35:54 <sridhar_ram> dkushwaha: okay, i just finished reviewing from my side.. 05:36:29 <sridhar_ram> dkushwaha: which other core member reviewing NSD ? 05:37:00 <dkushwaha> sridhar_ram, we got gongesh comments only before 05:37:25 <sridhar_ram> dkushwaha: okay, remind gongysh .. we can be the 2nd reviewer 05:37:32 <sridhar_ram> s/we/he/ 05:37:44 <dkushwaha> sridhar_ram, yes sure 05:37:55 <janki> o/, sorry late 05:38:12 <dkushwaha> sridhar_ram, it would be more batter, if more members can review it 05:38:22 <sridhar_ram> dkushwaha: once review comments are taken care, we can start merging NSD soon.. 05:38:35 <sridhar_ram> dkushwaha: actively ping diff core team members.. 05:38:53 <dkushwaha> sridhar_ram, yes, sure 05:39:22 <sridhar_ram> dkushwaha: thanks 05:39:42 <sridhar_ram> #topic VNF inline template 05:39:56 <sridhar_ram> janki: where do we stand on this? 05:40:11 <janki> sridhar_ram, cleint and server patchs are submitted for review 05:40:39 <dkushwaha> sridhar_ram, other than taker members, I also asked Mistral core, and also got comments frm his side. 05:40:57 <janki> gongysh has few comments on client side which conflict with what is mentioned in the spec. Need your thoughts on that too 05:41:06 <sridhar_ram> janki: we need to get the client / API part settled down soon... so focus on getting client patch https://review.openstack.org/#/c/414602/ 05:41:27 <sridhar_ram> gongysh is not here 05:41:39 <sridhar_ram> janki: do you mind briefly desc the issue ? 05:41:46 <janki> sridhar_ram, waiting for your opinion on the comments in client patch 05:42:26 <janki> sridhar_ram, sure. what he is proposing is to use --with-vnf-template option to show inline vnfds instaed of --all option 05:42:44 <janki> the spec says using --all. Not sure what to use 05:43:10 <sridhar_ram> this for vnfd-list cmd / api ? 05:43:25 <tbh> sorry for the late 05:43:29 <janki> sridhar_ram, yes. for vnfd-list --all or vnfd-list --with-vnf-template 05:43:31 <sridhar_ram> tbh: hi! 05:43:51 <tbh> sridhar_ram, hi 05:44:41 <janki> sridhar_ram, second, we talked about allowing user to enter both vnfd-id and vnfd-template with vnfd-template being precedent. His comment says to make vnfd-id and vnfd-template mutually exclusive. this again conflicts with what is mentioned in spec 05:45:18 <sridhar_ram> janki: this one is easy, if we can make it mutually exclusive that is alway better 05:45:41 <janki> sridhar_ram, ok. will make changes accordingly 05:45:50 <sridhar_ram> janki: we can edit the spec once the code merge, we usually don't do that.. but at times it is okay as an exception 05:46:17 <sridhar_ram> but cli doc will take care of the documentation 05:46:19 <janki> sridhar_ram, that solves my concern then :) 05:46:35 <sridhar_ram> janki: let's do the right thing :) 05:46:44 <janki> sridhar_ram, doc on client side too? or just the ref doc? 05:47:15 <sridhar_ram> janki: cli doc is autogenerated .. see http://docs.openstack.org/cli-reference/tacker.html 05:47:33 <janki> sridhar_ram, will have a look. 05:47:43 <sridhar_ram> regarding --with-vnf-template, i think gongysh comment is not to use "--all" 05:48:07 <janki> sridhar_ram, yes. he is suggesting to not use "--all" because it is used by opensatck client 05:49:04 <sridhar_ram> i personally don't like --with-vnf-template, i think we need a filter flag to indicate "inline" 05:49:15 <sridhar_ram> we can take it up in the gerrit review 05:49:42 <sridhar_ram> janki: another qq, will tacker vnfd-show <uuid> of an inline template work fine? 05:50:06 <janki> sridhar_ram, any variable used will be a filter flag. 05:50:31 <janki> sridhar_ram, yes. it should/will show. let me still check it on my system and get back on that 05:50:36 <sridhar_ram> janki: np 05:50:55 <janki> sridhar_ram, I also need your opinion on the change https://review.openstack.org/#/c/416215/4/tacker/api/v1/base.py 05:51:03 <janki> in this particular file. 05:51:45 <janki> so when vnfd-id is not specified and vnf-template is specified, vnfd-id will be None. THe above change bypasses validating UUID 05:51:55 <sridhar_ram> janki: i can look into it.. but i'm sure there are better qualified cores & member can provide some opinion as well :) 05:52:02 <janki> I have specified default vnfd-id to be None 05:52:32 <janki> sridhar_ram, yes ofcourse, everyone can have a look. 05:52:44 <sridhar_ram> tbh: sripriya: KanagarajM: please review this code.. it is fairly lightweight change.. 05:53:00 <s3wong> sorry --- just got home 05:53:12 <sridhar_ram> janki: on your side, please take care of the merge conflicts 05:53:18 <sridhar_ram> s3wong: howdy! 05:53:28 <sridhar_ram> let's move to the next item 05:53:29 <KanagarajM> sridhar_ram, sure. 05:53:36 <sridhar_ram> KanagarajM: thanks! 05:53:36 <janki> sridhar_ram, yes. I am waiting for few more review comments to make changes from my side 05:53:45 <sridhar_ram> #topic VNFC 05:54:03 <sridhar_ram> tbh: i left some comments in your patchset.. 05:54:36 <KanagarajM> sridhar_ram, should we schedule the corresponding bp for o3 on inline template ? 05:55:35 <sridhar_ram> KanagarajM: yes, https://blueprints.launchpad.net/tacker/+spec/vnf-inline-template is marked for o3 05:55:37 <tbh> sridhar_ram, I have seen the comments, 05:55:55 <sridhar_ram> KanagarajM: pretty much all BPs are now pointing to o3 :) 05:56:10 <tbh> sridhar_ram, but I have added abstract driver as just a placeholder for future drivers 05:56:19 <KanagarajM> sridhar_ram, yeah :) 05:56:41 <sridhar_ram> tbh: the summary is - can we make HOT SoftwareConfig has a inherent feature off openstack vim infra driver .. without any need to specify a vnfc-driver 05:57:02 <tbh> sridhar_ram, can I just keep it in the code, but remove "vnfc_driver" for now? 05:57:26 <sridhar_ram> tbh: tbh: not a good idea to leave dead code in the tree 05:57:41 <sridhar_ram> tbh: better to remove and reintroduce as needed... 05:58:03 <tbh> sridhar_ram, ok 05:58:40 <sridhar_ram> tbh: again, given this is a server side change we have some time... NSD is higher priority, fyi :) 05:58:57 <sridhar_ram> tbh: next, how to do gate tests for this feature? 05:59:15 <sridhar_ram> tbh: you need a special fedora image with heat agent ? 05:59:38 <tbh> sridhar_ram, yes, we need fedora image to be uploaded in glance 06:00:01 <sridhar_ram> KanagarajM: do you know how heat does gate tests for SoftwareConfig ? 06:00:08 <sridhar_ram> KanagarajM: any special image used ? 06:01:55 <sridhar_ram> tbh: there is a tool called diskimage builder .. http://docs.openstack.org/developer/diskimage-builder/ 06:02:30 <sridhar_ram> tbh: you can reach out to infra team on how to leverage this / other options to host a custom image for gate tests 06:02:49 <tbh> sridhar_ram, sure, will do that 06:03:13 <sridhar_ram> tbh: ping KanagarajM offline as they should have solved this problem in heat 06:03:39 <KanagarajM> sridhar_ram, i will have a look 06:03:48 <sridhar_ram> KanagarajM: oh, thanks! 06:04:23 <tbh> sridhar_ram, ure 06:04:25 <tbh> *sure 06:04:25 <sridhar_ram> moving on 06:04:46 <sridhar_ram> #topic Tacker-Senlin integration 06:05:07 <xuhaiwei_> hi, I am working on this sridhar_ram 06:05:44 <sridhar_ram> xuhaiwei_: KanagarajM: what is the latest in wrapping up the spec ? 06:05:54 <xuhaiwei_> the spec is reviewed by core members, I still have some comments need to feed back 06:06:45 <xuhaiwei_> I will modify the spec according to KanagarajM's comment 06:06:54 <sridhar_ram> xuhaiwei_: looks selective vdu support in senlin based scaling is the only outstanding this 06:07:03 <KanagarajM> sridhar_ram, i tried to ask some more details and i think xuhaiwei_ is working on it 06:07:16 <KanagarajM> xuhaiwei_, :) thanks. 06:07:20 <sridhar_ram> KanagarajM: xuhaiwei_: sounds good 06:07:38 <sridhar_ram> xuhaiwei_: anything else you want to bring it up on this topic ? 06:07:42 <xuhaiwei_> sridhar_ram, KanagarajM, just not sure what is selective vdu support means 06:08:49 <KanagarajM> xuhaiwei_, it helps user to scale give set of VDUs 1 and 2 from the all VUDs VDU 1,2,3... say for example. 06:08:56 <sridhar_ram> xuhaiwei_: currently all VDUs in the VNF gets scaled out / in .. 'selective VDU' is about scaling a subset of VDUs 06:10:18 <xuhaiwei_> sridhar_ram, KanagarajM, thanks for explaining, I will answer it in the spec 06:10:28 <sridhar_ram> xuhaiwei_: sounds good.. 06:10:34 <sridhar_ram> moving to the next topic 06:10:54 <KanagarajM> xuhaiwei_, yrw. 06:10:57 <sridhar_ram> #topic tosca-parser / heat-translator 0.7.0 06:11:39 <KanagarajM> sridhar_ram, i am closly following up with sahdev (HT ptl) and seems with current version, we can't support selective vdu, for which we are looking for HT. 06:12:33 <sridhar_ram> KanagarajM: thanks.. it was not clear whether this version would solve the problem.. 06:12:38 <KanagarajM> sridhar_ram, and sahdev assured to provide a reuired support in ocata so that tacker can provide support for selective vdu. but 0.7.0 does not have support 06:13:05 <sridhar_ram> KanagarajM: okay, it will be tight.. let's see 06:13:30 <tbh> sridhar_ram, and we latest tosca-parser to have NSD support 06:13:33 <KanagarajM> sridhar_ram, i have made it a high priority. hope we could make it. :) 06:13:36 <sridhar_ram> KanagarajM: thanks for following it up.. 06:13:46 <sridhar_ram> tbh: yep, that's great 06:14:18 <sridhar_ram> tbh: glad to have this nice cross-project collaboration between these two projects 06:14:39 <sridhar_ram> next topic.. 06:14:43 <tbh> sridhar_ram, yeah 06:14:51 <sridhar_ram> #topic API Framework 06:15:02 <sridhar_ram> diga: glad to see the spec is iterating nicely 06:15:12 <diga> sridhar_ram: :) Thanks 06:15:13 <sridhar_ram> diga: can you give an update? 06:15:17 <diga> sridhar_ram: yes 06:15:49 <diga> sridhar_ram: I think almost all the queries are answered in case of API framework selection & features support using new API framework 06:16:21 <KanagarajM> tbh, tacker requirements.txt has heat-translator>=0.4.0, you may need to set to right version based on your above comment. https://review.openstack.org/#/c/414937/16/requirements.txt 06:16:28 <diga> sridhar_ram: According to my study in API framework so far, we can start the work on falcon framework 06:17:11 <sridhar_ram> KanagarajM: good catch, post-NSD the version need to be bumped to 0.7.0 06:17:34 <diga> I had discussion with some community folks & infra team also, people suggests falcon over flask because falcon is 8x faster than flask 06:18:14 <diga> sridhar_ram: It is developed for cloud API only. 06:18:26 <sridhar_ram> diga: okay, for http-server, have you seen any other project using gunicorn ? 06:18:41 * sridhar_ram means any other openstack projects 06:19:06 <diga> sridhar_ram: in OpenSTack, most of the people uses, HTTP WSGI server like zaqar etc 06:20:03 <sridhar_ram> diga: that is for falcon, i was curious about gunicorn.. 06:20:15 <diga> sridhar_ram: ok 06:20:22 <sridhar_ram> diga: codesearch doesn't get much hits.. http://codesearch.openstack.org/?q=gunicorn&i=nope&files=requirements.txt&repos= 06:20:59 <sridhar_ram> diga: please investigate this portion (http-server to frontend falcon) 06:21:00 <diga> sridhar_ram: okay 06:21:29 <sridhar_ram> diga: i'm specifically interested to know what zaqar is using to frontend falcon 06:21:51 <diga> sridhar_ram: okay 06:22:00 <sridhar_ram> KanagarajM: btw, i agree w/ diga we should take up swagger work as a separate follow on 06:22:02 <diga> sridhar_ram: will talk to zaqar team 06:22:20 <sridhar_ram> KanagarajM: api framework is a *huge* effort 06:23:20 <KanagarajM> sridhar_ram, ok 06:23:22 <sridhar_ram> diga: another question, I'm assuming post falcon we need to work on API microversion support (this is a openstack wide goal) 06:23:25 <diga> sridhar_ram: just one question, are we using same tacker API server for tacker UI Now ? 06:23:35 <sridhar_ram> diga: something to keep in mind as you are exploring 06:24:05 <sridhar_ram> diga: yes, UI uses tackerclient library with uses the tacker API server 06:24:11 <diga> sridhar_ram: yes, current spec is written based on those efforts 06:24:46 <sridhar_ram> diga: wonderful work so far! 06:24:49 <KanagarajM> diga, i see, that monasca api also uses falcon and its one of the highly scalable service. so falcon would be a right choice as you discovered. thanks. 06:25:09 <diga> KanagarajM: :) YEs 06:25:13 <diga> sridhar_ram: Thanks :) 06:25:21 <sridhar_ram> diga: anything else? 06:25:27 <sridhar_ram> .. on this topic ? 06:25:53 <diga> sridhar_ram: then in this case, we can go with HTTP for API server like zaqar & monasca using now 06:26:29 <sridhar_ram> diga: yes, adoption is definitely a key factgor 06:26:40 <sridhar_ram> alright almost out of time 06:26:41 <KanagarajM> diga, i think you could start the impl in parallel, by that time we could merge the spec. 06:26:49 <sridhar_ram> KanagarajM: +1 06:26:50 <diga> sridhar_ram: I will discuss on thi topic with zaqar & monasca team today. 06:26:58 <sridhar_ram> diga: sounds good 06:27:01 <diga> KanagarajM: sridhar_ram : Sure 06:27:12 <sridhar_ram> #topic Open Discussion 06:27:26 <Yanxing_an> I have a few questions 06:27:34 <sridhar_ram> Just 13 calendar days left until feature freeze / client library release 06:27:34 <Yanxing_an> vnffg does not support EVENT , is it in plan? 06:27:43 <sridhar_ram> please get the reviews going 06:28:03 <Yanxing_an> ok 06:28:34 <diga> sridhar_ram: are we planning mid-cycle meetup in Feb ? 06:28:35 <sridhar_ram> Yanxing_an: good catch, those two feature landed in parallel .. it should've been fixed in Ocata 06:28:57 <gongysh> Yanxing_an, there is no design to not use event. so you can fix it if you have time. 06:29:14 <sridhar_ram> Yanxing_an: I think we still have room to do it .. 06:29:38 <sridhar_ram> Yanxing_an: btw, welcome to the project! thanks for the reviews 06:29:45 <sridhar_ram> Folks - we are out of time 06:29:45 <Yanxing_an> ok, i will to fix it 06:29:51 <Yanxing_an> tackerclient not support vnf-trigger command, is it need to add? 06:30:18 <sridhar_ram> Yanxing_an: let's take it offline in the #tacker channel 06:30:23 <sridhar_ram> bye everyone 06:30:25 <Yanxing_an> ok 06:30:28 <sridhar_ram> #endmeeting