17:01:12 <sridhar_ram> #startmeeting tacker 17:01:17 <openstack> Meeting started Tue Apr 19 17:01:12 2016 UTC and is due to finish in 60 minutes. The chair is sridhar_ram. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:20 <openstack> The meeting name has been set to 'tacker' 17:01:27 <sridhar_ram> #topic Roll Call 17:01:31 <bobh> o/ 17:01:35 <brucet> o/ 17:01:41 <dkushwaha> o/ 17:01:56 <janki91> o/ 17:02:29 <sridhar_ram> good morning, good afternoon and good evening - for all the folks around the globe! 17:02:38 <sripriya> o/ 17:03:05 <sridhar_ram> lets start.. 17:03:09 <sridhar_ram> #topic Agenda 17:03:34 <sridhar_ram> #link https://wiki.openstack.org/wiki/Meetings/Tacker#Meeting_Apr_19.2C_2016 17:03:42 <sridhar_ram> anything else beyond this ? 17:04:21 <sridhar_ram> #topic Annoucement 17:04:27 <sripriya> sridhar_ram: cherypicks to stable/mitaka 17:04:46 <sridhar_ram> sripriya: sure, good one.. 17:04:56 <LouisF> hi 17:05:02 <sridhar_ram> LouisF: hi ! 17:05:38 <sridhar_ram> T-1 for Summit ! 17:06:06 <sridhar_ram> of course we won't have weekly meeting next week.. 17:06:27 <sridhar_ram> .. I also propose to cancel the weekly meeting immediately after the summit. 17:06:40 <sridhar_ram> .. on May 3rd. 17:06:55 <sripriya> +1 17:07:16 <sridhar_ram> sripriya: thanks, anyone else ? 17:07:29 <bobh> +1 17:08:14 <sridhar_ram> alright.. canceled. Next meeting will be on May 10th. 17:08:55 <dkushwaha> sridhar_ram, peoples who are unable to attend summit, can join discussions remotely ? 17:09:01 <sridhar_ram> #topic Newton Summit 17:09:35 <sridhar_ram> dkushwaha: yes, I'm planning to setup a google hangout or webex 17:09:55 <sridhar_ram> will update the etherpad with details by early next week... 17:10:24 <sridhar_ram> Summit Etherpad.. #link https://etherpad.openstack.org/p/tacker-newton-summit 17:10:25 <sripriya> sridhar_ram: dkushwaha: is this for the developer tracks? 17:10:40 <dkushwaha> sridhar_ram, It will be very helpful. 17:10:55 <dkushwaha> sripriya, yes 17:10:59 <janki91> sridhar_ram: dkushwaha: this is a very good idea 17:10:59 <sridhar_ram> sripriya: yes, some of the European folks where asking for this as well 17:11:28 <sridhar_ram> though, one caveat ... please keep in mind, this is best effort commitment... 17:12:09 <sridhar_ram> design summits are inherently interactive .. we can't, for e.g., repeat the question that was asked in the floor 17:12:35 <sripriya> yes,etherpad chats should be best used :-) 17:13:26 <sridhar_ram> sripriya: yes, we need some discipline in handling a "formal" design summit session 17:13:50 <sridhar_ram> above summit etherpad is the main landing page.. 17:13:55 <s3wong> sorry, late 17:14:12 <sridhar_ram> I anticipate individual subject specific etherpad as needed.. 17:14:19 <sripriya> sridhar_ram: should we nominate ourselves to capture meeting minutes on etherpad for each of the tracks 17:14:33 <sridhar_ram> bobh: you are going to put on together for everything related to TOSCA ? 17:14:48 <bobh> sridhar_ram: yes 17:15:25 <sridhar_ram> sripriya: I've marked some folks as leads ... so someone other than them should help to capture the minutes / discussion in the etherpad 17:15:46 <sripriya> sridhar_ram: yeah 17:15:49 <sridhar_ram> trozet: s3wong: please plan to lead the Tacker VNFFG / SFC discussion 17:16:06 <s3wong> sridhar_ram: sure 17:16:35 <s3wong> doesn't look like a lot of folks would be at the summit Friday PM :-) 17:16:47 <sridhar_ram> I've also marked Friday morning contributor meetup for Tacker Architecture evolution 17:17:09 <sridhar_ram> s3wong: we decided to move the Friday meetup to AM 17:17:35 <sridhar_ram> brucet: will you be available on Friday AM to join this discussion ? 17:17:41 <brucet> Yes 17:18:12 <brucet> My flightbout is PM Friday 17:18:33 <s3wong> brucet: seems to be the case for everybody :-) 17:18:50 <sridhar_ram> among other things, I'd like to discuss / drive some clarity on Tacker VNFM / NFVO separation .. 17:20:14 <sridhar_ram> bobh: for TOSCA here are the main things I've in mind... (a) move tacker nfv definitions up into tosca-parser (b) discussion supporting VNFC / SoftwareComponent, (c) CSAR support 17:20:27 <bobh> sridhar_ram: ok 17:20:41 <sridhar_ram> bobh: (c) based on our previous chat.. anything else in this area ? 17:21:08 <bobh> sridhar_ram: I don't think so - there is new support for Policies and Groups coming into tosca-parser that we can leverage 17:22:12 <sridhar_ram> bobh: Sure. I also see few things are cross-track - VNF Scaling and recent TOSCA Scaling effort.. I'd like to discuss this as well 17:22:26 <bobh> sridhar_ram: yep 17:22:56 <sridhar_ram> next on refactoring / technical debt... 17:23:41 <sridhar_ram> I'd like to hear all of your top of your mind technical debt that you would like to have it fixed... 17:23:53 <sripriya> sridhar_ram: heat driver 17:24:16 <sridhar_ram> please rattle out as it comes to your mind... 17:24:38 <sridhar_ram> sripriya: couldn't agree more... 17:25:10 <s3wong> sridhar_ram: long time ago we talked about moving the ping/http monitoring driver out of server for scalability reason, is it time to finally install agents? 17:25:11 <sripriya> sridhar_ram: monitor file refactoring 17:25:26 <sripriya> s3wong: :-) jinxed! 17:25:35 <sridhar_ram> indeed :) 17:25:43 <s3wong> sripriya: :-) 17:25:59 <bobh> s3wong: agents or just external threads? 17:26:11 <sridhar_ram> s3wong: that is something we can discuss as part of arch-evolution.. 17:26:19 <s3wong> bobh: definitely something outside of where Tacker server is running 17:26:44 <s3wong> bobh: at least IMHO :-) 17:26:45 <sripriya> sridhar_ram: in terms of monitoring referrring, i was referring to removing the action policies out of monitor file 17:26:55 <sridhar_ram> this goes to the previous comment where there is now one monolithic server.. 17:26:58 <sripriya> * refactoring 17:27:20 <bobh> s3wong: I agree, issue is always the details 17:28:28 <sridhar_ram> a tacker-vnfm-mon-agent taking through an rpc bus back to tacker-vnfm-server is something we should consider 17:28:54 <sridhar_ram> that design pattern is quite common on various openstack services 17:29:28 <s3wong> sridhar_ram: agreed --- particularly, if different VNFs have different monitoring need and policies, funneling all of that into a single server is a bit much 17:30:32 <sridhar_ram> one challenge, some class of monitoring driver doesn't belong in the tacker-vnfm code base.. instead it could come in the CSAR bundle 17:31:03 <sridhar_ram> in fact, I envision a model where mon drivers are actually registered with tacker when things come up... 17:31:08 <brucet> I actually think you should think about packaging when considering this problem. 17:31:42 <brucet> The monitoring function may be specific to a VNFM, so it should be possible to package the monitoring function with the VNFM. 17:31:52 <sripriya> sridhar_ram: does that involve an external monitoring agent? 17:32:20 <sridhar_ram> brucet: true some mon drivers are generic like ping, http-ping, perhaps bfd.. 17:32:37 <sridhar_ram> but some might be quite custom to a specific VNF... 17:32:45 <s3wong> brucet: in fact, it is even possible that some monitoring is specific to an VNF 17:32:52 <brucet> Exactly 17:33:09 <sridhar_ram> today we have to have a entry point in setup.cfg *before* tacker server start.. that's bad 17:33:21 <sripriya> +1 totally agree 17:34:52 <sridhar_ram> some of these can be fixed in refactoring cycles.. other might go into arch evolution .. 17:35:32 <sridhar_ram> I'd request your help to break down each such idea into smaller "units of work" .. 17:36:02 <sridhar_ram> . without that we would only be talking about this forever :) 17:36:45 <sridhar_ram> I think we have primed our brain to think in this area of refactoring now... 17:36:48 <s3wong> sridhar_ram: true :-) 17:37:07 <sridhar_ram> please bring your ideas for some nice debates next week 17:37:27 <sripriya> sridhar_ram: we can have a separate etherpad for monitoring framework discussion to breakdown these ideas? 17:37:27 <sridhar_ram> anything else related to newton summit ? 17:37:35 <dkushwaha> sridhar_ram, isn't good to lock bugs for all, for better tracking? 17:37:51 <sridhar_ram> sripriya: good idea... 17:39:40 <sridhar_ram> sripriya: created a dummy page for now.. #link https://etherpad.openstack.org/p/tacker-newton-monitoring-refactor 17:40:11 <sripriya> sridhar_ram: i was just creating one, cool thanks 17:40:16 <sridhar_ram> dkushwaha: we can consider creating RFE bugs once we decide what to do and break down into individual chunks 17:40:50 <sridhar_ram> breaking down will help for new developers joining tacker community as well...! 17:41:31 <sridhar_ram> Also don't forget to attend many Tacker related sessions... 17:42:01 <sridhar_ram> there is at least 3 talk session, one birds of feature and a hands on lab ! 17:42:04 <sridhar_ram> should be fun 17:42:11 <sridhar_ram> moving on... 17:42:16 <sridhar_ram> #topic VNF Scaling 17:42:22 <sridhar_ram> just a quick heads up... 17:43:06 <brucet> OK. My head is up. 17:43:17 <sridhar_ram> a contributor named Tung Doan has submitted a blueprint and some WIP code that looks promising :) 17:43:41 <sridhar_ram> brucet: yeah, thought you will be interested 17:43:49 <s3wong> sridhar_ram: for auto-scaling? 17:44:04 <sridhar_ram> #link https://review.openstack.org/#/c/306562/ 17:44:08 <sridhar_ram> s3wong: yes.. 17:44:32 <sripriya> and https://review.openstack.org/#/c/306256/ 17:44:37 <sridhar_ram> any chance Tung Doan here in the meeting ? 17:45:16 <brucet> From a quick look, this makes perfect sense 17:45:21 <sridhar_ram> glad to see the proposal ! 17:45:26 <sridhar_ram> brucet: yep! 17:45:31 <brucet> I think unrelated to dcaling though 17:46:01 <s3wong> just as we talked about monitoring driver... a spec is filed to integrate Ceilometer... :-) 17:46:15 <sridhar_ram> brucet: if you see the examples some of the action routine for cpu_util is scale-up... 17:46:18 <brucet> Exactly. And makes sense 17:46:33 <brucet> OK. I'll look further 17:46:34 <sripriya> s3wong: actually it was registered quite sometime back 17:47:14 <s3wong> sripriya: four days ago... :-) 17:47:28 <sridhar_ram> s3wong: exactly.. this has multiple touch points. one is the TOSCA parser work on Scaling.. 17:47:46 <sripriya> https://blueprints.launchpad.net/tacker/+spec/alarm-based-monitoring-driver 17:48:23 <s3wong> sripriya: yeah, that one is about a month... and no approver... 17:48:41 <sridhar_ram> lets converge on this spec and help to shape it up... 17:49:03 <sridhar_ram> I'd like to get manual scaling "designed" in as well, if not implement.. 17:49:32 <sridhar_ram> they (auto-scaling & manual scaling) need to gel well in the design and in the TOSCA templates 17:49:33 <brucet> I had a discussion a while back about manual vs. automatic sclaing 17:49:53 <brucet> I think the conclusion we came to is that a single implementation can easily do both 17:50:02 <bobh> manual scaling could simply invoke the same trigger as automatic scaling - just do it manually 17:50:07 <brucet> I think discussion was via email 17:50:15 <brucet> Exactly 17:50:20 <sripriya> sridhar_ram: the scaling model needs to be generic enough to adapt to both auto and manual scaling ... 17:50:21 <sridhar_ram> brucet: yeah, it will be good to bring it into gerrit 17:50:30 <sridhar_ram> sripriya: agree.. 17:50:46 <brucet> OK. I'll find it in my email and post. 17:51:17 <sridhar_ram> manual scaling could be "tacker vnf-update --vnf foo --vdu VDU2 --action scale-up 17:51:24 <brucet> sridhar_ram: Please tell me where to post it 17:51:46 <sripriya> sridhar_ram: should this be scale up or scale out? 17:51:47 <sridhar_ram> brucet: lets use the gerrit link https://review.openstack.org/#/c/306562/ 17:52:08 <sridhar_ram> sripriya: oops, NO, NO scale-up / down... 17:52:14 <brucet> Hmmm.... I think that's different 17:52:17 <sridhar_ram> only scale-out and scale-in 17:52:40 <sridhar_ram> we can discuss more on this next week... 17:52:42 <brucet> There was a blueprint done for scaling a while ago. 17:52:55 <brucet> I'll find it and update based on email thread 17:52:55 <sridhar_ram> brucet: yes, one from prashantD_ ... 17:52:58 <sripriya> so the action should be --action scale-out? 17:53:04 <brucet> Think so 17:53:14 <s3wong> yeah, auto-scaling was an item we penciled into roadmap since we came back from Liberty (Vancouver) summit 17:53:19 <sridhar_ram> brucet: .. it will be good to incorporate the ideas into this new one 17:53:28 <sridhar_ram> sripriya: yes... 17:53:56 <sridhar_ram> s3wong: yep, we have punted this enough... we need to get it one way or the other in Newton 17:54:14 <sridhar_ram> folks - lets park the scaling discussion here... lets continue next week 17:54:18 <sridhar_ram> moving on... 17:54:31 <sridhar_ram> #topic stable/mitaka cherrypick 17:54:45 <sridhar_ram> sripriya: ^ 17:55:34 <sripriya> sridhar_ram: yeah, just wanted to remind tackers to cherrypick their commits to mitaka if they think that needs to be added to stable release 17:56:10 <sridhar_ram> sripriya: do we need original owners to do cherrypick or others can do ? 17:56:26 <sripriya> this can include regressions/rfes/improvements if you think it needs to be merged to mitaka release, please do so accordingly 17:56:57 <sripriya> sridhar_ram: i guess owners can take it up because there may be merge conflicts to handle 17:57:08 <sripriya> and i assume they would know it better 17:57:24 <sridhar_ram> okay.. 17:57:54 <sridhar_ram> devs - please look at your patchsets that merged in the last 3 weeks and do a cherrypick to stable/mitaka 17:58:15 <sridhar_ram> dkushwaha: you might have a sizable number ! 17:58:18 <janki91> sridhar_ram: sripriya: just clicking on cherrypick button will work? 17:58:23 <sripriya> it is also important to merge important fixes this week, as tacker hands on session team may use our mitaka branch to demo VNFM 17:58:33 <sripriya> thanks everyone 17:58:35 <sridhar_ram> janki91: yep, most of the time that would work.. 17:58:54 <dkushwaha> sridhar_ram, yep, looking into that 17:59:26 <janki91> I have made changes in the latest tacker release cloning master from github 17:59:38 <janki91> does it reflect mitaka? 17:59:44 <sridhar_ram> bobh: would you be able to wrap up https://review.openstack.org/#/c/299737/ ? 17:59:55 * sridhar_ram notes time is up! 18:00:12 <sridhar_ram> bobh: we can take it up in the #tacker channel 18:00:16 <sripriya> janki91: nope , you need to cherrypick that to mitaka 18:00:23 <sripriya> here are the guidelines 18:00:24 <sripriya> http://docs.openstack.org/project-team-guide/stable-branches.html 18:00:29 <sridhar_ram> Folks - have a safe trip to Austin 18:00:36 <sridhar_ram> see you all next week 18:00:39 <sridhar_ram> #endmeeting