16:02:58 <sridhar_ram> #startmeeting tacker
16:02:58 <openstack> Meeting started Thu Sep 10 16:02:58 2015 UTC and is due to finish in 60 minutes.  The chair is sridhar_ram. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:03:02 <openstack> The meeting name has been set to 'tacker'
16:03:14 <sridhar_ram> #chair bobh sripriya
16:03:15 <openstack> Current chairs: bobh sridhar_ram sripriya
16:03:40 <sridhar_ram> Agenda - #link  https://wiki.openstack.org/wiki/Meetings/Tacker#Meeting_Sep_10.2C_2015
16:04:13 <sridhar_ram> #topic Announcements
16:04:59 <sridhar_ram> Tacker Liberty Deadline is Sept 25th - #link https://wiki.openstack.org/wiki/Tacker/Releases
16:05:13 <sridhar_ram> just two weeks away
16:05:57 <sridhar_ram> New Tacker development process is now merged and official - #link https://review.openstack.org/#/c/219012/
16:06:37 <sridhar_ram> please follow the process in your contribution efforts, for any clarifications please reachout to the core team
16:07:06 <sridhar_ram> Experimental features proposal for Tacker is announced in the ML - #link http://lists.openstack.org/pipermail/openstack-dev/2015-September/074015.html
16:07:25 <sridhar_ram> comments / suggestion are welcome ...
16:08:07 <sridhar_ram> as part of this I'll open mitaka dir in tacker-specs repo to start working in post-Liberty specs (like sfc, nvfo, ...)
16:08:41 <sridhar_ram> any other general heads up / thoughts ?
16:09:40 <sridhar_ram> #topic Liberty Release Status
16:09:57 <sridhar_ram> if you've noticed we have a lose agenda today :)
16:10:23 <sridhar_ram> lets go around the table to see where we are w.r.t Liberty release - the deadline is approaching fast
16:10:55 <sridhar_ram> sripriya: few words about MANO API ?
16:11:47 <sripriya> sridhar_ram: update on API, it is ready for review, it had some issues with jenkins not picking it up, i'm resolving that and it should be good to go
16:12:09 <sripriya> sridhar_ram: plugin unit tests are added, so we have something to quick start for py27
16:12:47 <sripriya> sridhar_ram: however tox would pick up all test cases that are under unit and error out crazily as none of them are functional
16:12:47 <sridhar_ram> sripriya: this is awesome, we finally have py27 unit-tests for tacker
16:13:38 <sripriya> sridhar_ram: we need to take up cleaning task of unit tests dir. and retail only whatever is relevant
16:13:41 <sridhar_ram> sripriya: we can remove / move old test out of the way so that tox doesn't pick up
16:14:07 <sripriya> sridhar_ram: yes, i plan to do that as a seperate activity
16:14:48 <sridhar_ram> sridhar_ram: okay, let me know if you need help. I've few other things to cleanup as part of my Part-2 of my original cleanup patchsets
16:15:13 <sripriya> just tested the tox against the new dir. 'vm' I introduced and it worked fine
16:15:27 <sripriya> since my unit tests does not have any dependencies
16:15:42 <sridhar_ram> sripriya: for now, can't you set the OS_TEST_PATH to your specific dir ?
16:15:52 <sripriya> i can do
16:16:07 <sripriya> but that is not what i want to push upstream
16:16:37 <sripriya> it is basically updating the .testr.conf file, if you want me to push that change to point it to the dir., i will include that change
16:17:43 <sridhar_ram> hmm... that might trip what Santosh is introducing for scenario tests ?
16:17:59 <sripriya> sridhar_ram: no
16:18:18 <sripriya> sridhar_ram: functionla tests has its path setup in tox.ini itself
16:18:23 <sripriya> *functional
16:18:29 <sridhar_ram> okay .. so we can keep -e py27 and -e functional points to separate paths.
16:18:37 <sripriya> sridhar_ram: yes
16:19:11 <sridhar_ram> btw - we also need the doc.openstack.org enabled for Tacker and have it point to MANO API..
16:19:49 <sridhar_ram> please plan that for Liberty Release
16:19:57 <sripriya> that will mean working with the docs team right?
16:20:36 <sripriya> sridhar_ram: i dont know if they allow for stackforge projects on docs.openstack.org
16:20:44 <sridhar_ram> no, I can help to get the project-config to do doc-pubs... from your part you need to write up the doc
16:20:53 <sridhar_ram> ah.. that's a good point..
16:21:12 <sripriya> sridhar_ram: they allow only governance projects
16:21:22 <sridhar_ram> either case we will fold into openstack/ soon .. I'll follow up w/ infra team on that
16:21:44 <sripriya> sridhar_Ram: alright, meanwhile i will add the doc-ref in tacker dir.
16:21:56 <sridhar_ram> cool..
16:22:10 <sridhar_ram> sripriya: anything else on MANO API ?
16:23:08 <sripriya> sridhar_ram: that is it, i request everyone to do some testing of patchsets and provide any feedback including unit tests
16:23:39 <sridhar_ram> sounds good.. I'll give it a spin for sure
16:24:07 <sridhar_ram> bobh: any update on the health-monitoring spec and its progress ?
16:24:38 <lsp42> sridhar_ram, bobh: Is this to take us beyond just ping?
16:24:41 <bobh> sridhar_ram: Looking for feedback on the latest spec - a couple of outstanding questions
16:25:03 <lsp42> sridhar_ram: Where is that spec?
16:25:05 <bobh> lsp42: Yes - it is going to be modular so you can supply your own monitor
16:25:33 <lsp42> bobh: Sorry, me being impatient! I would like to read this spec
16:25:45 <bobh> #link https://review.openstack.org/#/c/202126/
16:26:20 <bobh> At this point I think it will mostly be copied from the mgmt_driver design with the appropriate changes for monitoring
16:26:35 <sridhar_ram> bobh: towards lsp42's point - are you planning to keep http-ping in scope for this effort ?
16:26:50 <bobh> sridhar_ram: Yes - that is going to be the reference implementation
16:27:25 <bobh> sridhar_ram: The basic design is a copy of the existing mgmt_driver
16:28:06 <sridhar_ram> bobh: sure, but are you planning to introduce http-alive as one of the pre-supplied mon-driver ?
16:28:31 <bobh> sridhar_ram: I'm hoping to - ping and http-ping should both be pretty simple to implement
16:28:43 <sridhar_ram> bobh: that's cool..
16:28:57 <bobh> sridhar_ram: The challenge is getting the whole driver mechanism plumbed into the server
16:29:14 <sridhar_ram> so at the end of this effort - we will have icmp-ping, http-ping and an ability to introduce a custom driver
16:29:19 <bobh> sridhar_ram: which is why I'm going to copy mgmt_driver as much as possible
16:29:26 <sridhar_ram> bobh: totally agree, this will be a non-trivial effort
16:29:43 <sridhar_ram> bobh: sure, that would be good start..
16:30:07 <bobh> sridhar_ram: So if anyone has feedback on the spec please let me know ASAP - coding has started :-)
16:30:22 <sridhar_ram> sure, here is one :)
16:30:40 <sridhar_ram> we also need to bring boot-wait that is currently in tacker.conf into something being VNF / VDU specific
16:30:53 <s3wong> sorry, way late --- still need time to get used to the new commute
16:30:57 <sridhar_ram> will leave these thoughts in the review..
16:31:09 <bobh> that may apply to both mgmt and monitor
16:31:32 <sridhar_ram> s3wong: morning, we are just going around the table to get liberty status ...
16:31:37 <sridhar_ram> bobh: agree
16:32:02 <sripriya> bobh: regarding driver mechanism plumbing in server i can help you there, invested some time in that
16:32:14 <bobh> sripriya: Thanks!
16:32:50 <sridhar_ram> bobh: btw, on the bigger context, there is renewed interest in Tacker's health-monitoring
16:33:46 <sridhar_ram> few potential tacker user / operators reached out .. and their immediate needs are all around health-mon
16:33:49 <bobh> sridhar_ram: This should provide a generic framework for monitoring and health and scaling can take advantage of that
16:34:24 <sridhar_ram> bobh: sounds good
16:34:34 <sridhar_ram> anything else on health-mon ?
16:34:43 <bobh> sridhar_ram: Ideally I would like to allow the VNF to provide it's own monitoring plugin without needing to install on the server.  Future...
16:35:31 <sridhar_ram> well... currently we are hampered by the need to have an entry-point in setup.cfg for every "driver"
16:36:09 <lsp42> maybe we need to add the ability to upload a monitoring driver via the API like we do with VDUs? Let it auto-configure
16:36:14 <sridhar_ram> we can loosen this in the next iteration.. it seems you already have quite a bit in your plate for next few weeks
16:36:20 <bobh> sridhar_ram: Definitely
16:37:26 <sridhar_ram> lsp42: interesting - this exact suggestion came in our recent mid-cycle discussion (Xin who suggested this ? )...
16:38:41 <sridhar_ram> lsp42: introducing "code" during runtime is a privileged action and it has many security implication as well...
16:38:59 <lsp42> ok
16:39:11 <lsp42> so maybe we need abstract drivers which then get fed directives
16:39:20 <sridhar_ram> typically - any code is introduced by a conscious action of the deployer
16:39:33 <bobh> lsp42: A well-defined monitoring interface would help a lot
16:39:46 <sridhar_ram> lsp42: exactly, that is what we are shooting for..
16:39:54 <bobh> between the VNFM and VNF
16:40:04 <lsp42> sure
16:41:12 <sridhar_ram> anything else on health-mon ?
16:41:59 <sridhar_ram> I don't see prashant online ..
16:42:07 <sridhar_ram> for an update on auto-scaling ..
16:43:04 <sridhar_ram> I can give a quick update on SFC .. though it is not a liberty target..
16:43:40 <sridhar_ram> As described in the experimental feature .. we would like to bring in Tim's SFC efforts into Tacker
16:44:23 <sridhar_ram> some of us attended last couple of opnfv-sfc meeting to answer any Tacker related question..
16:44:48 <sripriya> sridhar_ram: may be you want to invite trozet in the next tacker meeting?
16:45:04 <sridhar_ram> here is a slide deck that shows our thoughts on how we plan to introduce SFC #link https://docs.google.com/presentation/d/18AGaiysVgHOd_fIHVpObMO7zUkMjJGOQ98CUwkxU1xo/edit#slide=id.p4
16:45:58 <sridhar_ram> sripriya: sure, I plan to .. Tim is busy with some OPNFV deliverables this week and he is planning to get back to SFC next week.
16:46:13 <sripriya> sridhar_ram: cool
16:47:31 <sridhar_ram> any comments on the  interested in Tacker + SFC slide above is welcome!
16:47:56 * sridhar_ram bad edit, argh
16:48:22 <sridhar_ram> any questions on SFC ?
16:49:27 <s3wong> sridhar_ram: I missed the OPNFV SFC meeting yesterday morning --- was there any Tacker related discussion there?
16:50:12 <sridhar_ram> s3wong: yes, I prepared the above slide deck for that meeting, but didn't get time to present it...
16:51:15 <sridhar_ram> but Tacker was discussed quite a bit to full fill their need VNFM needs ...
16:52:00 <lsp42> was there any mention of Tacker being the VNFO once SFC is in place?
16:52:01 <sridhar_ram> I suggested them to use Tacker as the api-server / wsgi server (along the lines Tim.R explored) for SFC API instead of inventing something adhoc
16:52:32 <sridhar_ram> lsp42: absolutely, that's where we are heading ...
16:52:45 <sridhar_ram> if you read the ML email on experimental features - #link http://lists.openstack.org/pipermail/openstack-dev/2015-September/074015.html
16:52:46 <lsp42> sorry - still trying to play catchup as well as reading code!
16:53:21 <sridhar_ram> ... we are building all the puzzle pieces towards an NFVO !
16:53:26 <sridhar_ram> lsp42: no problem
16:54:03 <lsp42> sridhar_ram: Thanks. What was their response to using Tacker as the api-server for SFC?
16:54:41 <sridhar_ram> They are quite supportive but they also need something really quick.. their B-release testing is not too far
16:54:58 <sridhar_ram> that's why we need to fast-track Tacker SFC efforts!
16:55:24 <s3wong> sridhar_ram: yes, I attended last week's meeting --- and it seems like fitting us into their B-release timeframe is going to be difficult
16:56:02 <s3wong> despite trozet's great effort on the coding front
16:57:25 <sridhar_ram> s3wong: I think are are other bigger road-blocks in openstack - ODL-SFC interaction - on how nova / neutron created the vnics in the ovs bridge and around  vxlan tunnel terminations...
16:58:29 <s3wong> sridhar_ram: yes, there was some discussion on that also last week
16:58:32 <sridhar_ram> what I mean is we will have a window to get this done while other things (around OVS-NSH support, vnice / bridge inters) are still falling in place
16:58:52 <sridhar_ram> quick time checkk - 2min mark!
16:59:06 <sridhar_ram> we need to wrap up soon...
16:59:06 <s3wong> sridhar_ram: OVS support of NSH would be masked from us anyway, it is more ODL's problem
16:59:28 <sridhar_ram> s3wong: but it is required for meaningful B-release
16:59:30 <lsp42> this could be a huge win
17:00:04 <s3wong> sridhar_ram: technically it is required for ODL Lithium for SFC, but the SFC team decided to maintain their own fork...
17:00:22 <lsp42> hrmm
17:00:25 <sridhar_ram> lsp42: It sure will, but we need to keep an open mind on sfc and execute well to enable them (opnfc-sfc) to succeed :)
17:00:39 <sridhar_ram> times up folks...
17:00:47 <sridhar_ram> talk to all next week
17:00:52 <s3wong> bye
17:00:53 <sridhar_ram> thanks for attending!
17:00:57 <lsp42> bye
17:00:58 <sridhar_ram> #endmeeting