04:01:04 <ekcs> #startmeeting congressteammeeting
04:01:04 <openstack> Meeting started Fri Sep 28 04:01:04 2018 UTC and is due to finish in 60 minutes.  The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot.
04:01:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
04:01:08 <openstack> The meeting name has been set to 'congressteammeeting'
04:02:11 <ekcs> hi all! topics kept here as usual. https://etherpad.openstack.org/p/congress-meeting-topics feel free to add your topics!
04:02:35 <akhil_jain> ekcs: Hi!
04:03:22 <ekcs> hi akhil_jain !
04:03:27 <akhil_jain> I would like to introduce dkushwaha: PTL of tacker and mbindlish: active contributor in QA
04:03:29 <dkushwaha> o/
04:03:36 <mbindlish> hi
04:03:51 <ekcs> hi dkushwaha !
04:03:53 <ekcs> hi mbindlish !
04:04:13 <ekcs> big thanks for joining us at this time.
04:05:26 <mbindlish> :)
04:06:51 <akhil_jain> ekcs:  So as discussed they will help us understand tacker-monasca-congress usecase
04:07:09 <dkushwaha> Thanks akhil_jain ekcs
04:07:20 <dkushwaha> Let me explain our use case.
04:07:36 <ekcs> that’s awesome. thanks akhil_jain ! yes definitely excited to learn more and see where I can help.
04:07:52 <dkushwaha> ekcs, please see https://docs.google.com/presentation/d/1C1GEoSC34Yg6rNCD9MVUkhg7s4-9WMI6c7NK9FVnxwY/edit?usp=sharing
04:08:24 <dkushwaha> As per the OPNFV Doctor fault management framework
04:08:49 <ekcs> got it!
04:08:53 <dkushwaha> we are using congress as inspector, monasca as monitor, and Tacker as VNF manager
04:09:37 <dkushwaha> For automatic corrective actions depending on the monitor data evaluation Tacker execution driver needs to be supported in congress.
04:10:23 <ekcs> that makes sense.
04:10:23 <dkushwaha> which will poll VNF data from Tacker and can execute policy
04:11:27 <ekcs> I see.
04:12:10 <ekcs> So here Congress’s role is to poll tacker data, combine with monasca webhook data, then send something back to tacker is the form of executions?
04:12:13 <dkushwaha> Although, on Tacker side, we have one gap. We needs to have an interface which can list VNFs per comupte node(which is currently missing in Tacker side)
04:12:40 <dkushwaha> ekcs, yes
04:14:05 <ekcs> got it.
04:14:31 <ekcs> so just to help me get a more concrete sense, what’s a typical action congress may execute on tacker?
04:14:35 <ekcs> in these scenarios?
04:16:24 <dkushwaha> ekcs, the corrective action can be to fatch, VNFd id from Tacker VNF list, and re-spawn VNF.
04:17:47 <dkushwaha> this was the one of action we looked for our PoC, but i think that can be explore for more
04:18:55 <ekcs> got it that makes sense! again just for my context, is there a specific tacker api call or a series of api calls that accomplish that?
04:19:20 <ekcs> i’m looking through the tacker api reference here https://docs.openstack.org/tacker/ocata/devref/mano_api.html
04:19:47 <ekcs> not most current though.
04:20:06 <ekcs> is this a better place to look? https://developer.openstack.org/api-ref/nfv-orchestration/v1/index.html
04:22:04 <dkushwaha> ekcs, to re-spawn VNF we can use POST /v1.0/vnfs
04:22:45 <dkushwaha> which is nothing, but create the VNF with specified VNFD id
04:23:22 <ekcs> got it!
04:24:15 <dkushwaha> ekcs, and later improvement, more actions can be scaling etc.. too
04:24:35 <dkushwaha> *VNF Scaling
04:24:57 <ekcs> I see. That’s great.
04:26:54 <ekcs> that’s cool. I’m definitely excited about this work. Seems that with some work, all the pieces would fit great together.
04:27:54 <ekcs> dkushwaha: let me know if you have more to say before I continue with questions/response =)
04:28:33 <dkushwaha> ekcs, i think, i am done from my side.
04:29:06 <ekcs> got it thanks a lot for the clear overview dkushwaha !
04:29:27 <ekcs> akhil_jain and mbindlish do you have anything to add?
04:30:09 <mbindlish> ekcs: I think dkushwaha explained completely!
04:30:24 <akhil_jain> ekcs: None from my side
04:31:51 <ekcs> cool cool. So yea like I said I think it’s a great vision overall. I wonder what the work items would be on Congress side.
04:32:09 <ekcs> monasca webhook integration and tacker data source driver obviously.
04:32:44 <ekcs> I wonder if there are further requirements for the use case.
04:33:43 <ekcs> for example, if congress calls the api to respawn VNF but it fails, what would we like to happen?
04:36:30 <dkushwaha> ekcs, I think for now we can just retry for re-spawn, but further improvement definitely will be a great addition
04:36:51 <ekcs> dkushwaha: got it!
04:37:26 <ekcs> on another note, akhil_jain and I were wonderingi if there may be a example policy for the use case.
04:38:23 <ekcs> in particular, it’d help to understand whether the splitting of one monasca webhook update into multiple transactions would cause problems or not.
04:39:10 <dkushwaha> mbindlish, do you have example policy right now? otherwise we can share it later
04:39:29 <ekcs> I understand that not all the schemas are fixed yet, but even a rough sketch would be helpful.
04:40:46 <mbindlish> Currently not having for respawn, but for example: deleting VNF
04:40:54 <mbindlish> openstack congress policy rule create --name delete_vnf classification 'execute[tacker:delete_vnf("<VNF_ID>")]:- monasca:alarm_notification(state="ALARM")'
04:41:12 <mbindlish> a rough idea to understand
04:41:44 <ekcs> got it!
04:42:40 <ekcs> great. well if there’s more sample available to share in the future I’d love to see it.
04:43:08 <ekcs> is there anything specific I can help with?
04:44:12 <ekcs> on the rules, I’m down to help create them together if there’s ever any difficulty.
04:44:51 <dkushwaha> ekcs, is there some patch already in fly for the same?
04:45:35 <akhil_jain> dkushwaha: i have started working on it https://review.openstack.org/#/c/604275/
04:45:42 <ekcs> yes. akhil_jain has a WIP patch for tacker data source driver. a version of monasca webhook integration is already in rocky.
04:46:42 <dkushwaha> great, I thinks we can further discuss on patch review
04:46:46 <akhil_jain> but maybe for deleting and respawning VNF I will have to look more into it
04:47:04 <ekcs> perfect.
04:47:12 <ekcs> one more question: roughly how many VMs and how many VNFs are we envisioning in a setup?
04:47:21 <ekcs> like 10s or 100s or 1000s?
04:48:13 <ekcs> or we can defer till later.
04:49:34 <dkushwaha> during our PoC, we used around 10s for this integration. mbindlish, correct me if i m wrong.
04:50:22 <mbindlish> Yes, correct but I think there were no limit defined
04:50:52 <ekcs> got it. what’s a typical scale envisioned in production?
04:51:46 <dkushwaha> ekcs, not sure, but i think, we can defer limit part for now
04:52:31 <ekcs> cool. not looking for limit per se, but hoping for a general sense to understand if we’ll run into scaling issues. but yes definitely can be discussed in future.
04:52:47 <ekcs> well is there anything else to discuss on the topic for today?
04:53:41 <dkushwaha> ekcs, from my side, nothing for now
04:54:02 <mbindlish> nothing from my side
04:54:29 <ekcs> got it! I think we can end then unless akhil_jain there’s something you want to discuss?
04:54:54 <ekcs> thanks again so much dkushwaha and mbindlish for stopping by for the very helpful discussion.
04:55:22 <akhil_jain> ekcs: nothing for now. got direction to work on from this discussion
04:55:25 <dkushwaha> ekcs, thanks
04:55:36 <akhil_jain> Thanks dkushwaha and mbindlish
04:55:39 <mbindlish> ekcs: dkushwaha akhil_jain thanks
04:55:58 <ekcs> awesomeness. looking forward to continuing to work together!
04:56:03 <ekcs> let’s end meeting then.
04:56:06 <ekcs> happy friday all!
04:56:11 <dkushwaha> Thanks folks :)
04:56:16 <mbindlish> :)
04:56:21 <akhil_jain> Happy friday!
04:56:31 <ekcs> bye all =)
04:56:33 <ekcs> #endmeeting