14:01:40 <davidsha> #startmeeting network_common_flow_classifier
14:01:41 <openstack> Meeting started Tue Oct  3 14:01:40 2017 UTC and is due to finish in 60 minutes.  The chair is davidsha. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:42 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:45 <openstack> The meeting name has been set to 'network_common_flow_classifier'
14:01:48 <davidsha> Hi everyone
14:01:51 <igordc> hi davidsha
14:02:14 <bcafarel> hello
14:02:35 <davidsha> I'll wait 2 mins for more to join.
14:02:50 <igordc> what's up bcafarel
14:03:37 <bcafarel> having fun with zuulv3 fallout mostly :)
14:04:05 <igordc> bcafarel: :(
14:04:07 <davidsha> As long as your having fun :P
14:05:09 <davidsha> I guess we should start
14:05:30 <davidsha> #topic CCF v0 - Update
14:05:44 <bcafarel> reedip will probably pop in (now that fwaas meeting was moved)
14:06:39 <davidsha> I've most of the proposed changes implemented, just had a few issues with my dev environment that have help me up.
14:06:48 <davidsha> held*
14:07:41 <davidsha> I should have at least the database migrations patch and the models patch up by Friday.
14:08:09 <davidsha> Are there any questions for the patches?
14:08:51 <igordc> davidsha: when is v0 ready to merge?
14:09:13 <igordc> davidsha: should existing code be merged as is and then smaller patches submitted to fix the multiple parts towards v1?
14:09:41 <davidsha> igordc: When there is a consensus on it being ready to merge is the simplest answer.
14:09:51 <bcafarel> :)
14:10:02 <davidsha> igordc: That could be an issue with database migrations
14:10:46 <davidsha> We don't want 2 or 3 migration contractions and expansions because we pushed it in too early
14:11:28 <davidsha> Butt sooner rather than later is prefered, especially when people need to experiment for PoCs
14:11:36 <igordc> davidsha: you need the migrations when releasing v1
14:11:43 <bcafarel> can't we allow updating the migrations until v1 is there?
14:11:46 <igordc> davidsha: there is no release for v0... migrations can be adapted
14:12:07 <davidsha> That's true, fair enough.
14:12:54 <davidsha> If there are not massive problems with next next patch will we merge it then and refine it as we progress?
14:13:30 <davidsha> Will Zuul also be an issue?
14:14:00 <igordc> davidsha: sounds good to me
14:14:40 <davidsha> ok, will we move on?
14:14:43 <bcafarel> if there are no major objections, having an "initial v0" in tree will be nice indeed
14:14:54 <bcafarel> for zuul, it should be ok ccf-wise
14:14:54 <davidsha> kk
14:15:09 <bcafarel> there may be failures, but "generic" ones
14:15:29 <bcafarel> hopefully in a few days the dust will have settled a bit
14:15:40 <davidsha> Hopefully!
14:15:55 <davidsha> #topic PTG discussion
14:16:27 <davidsha> I was hoping tmorin would be here for this, cause now I need to speak on his behalf
14:16:31 <davidsha> :P
14:17:05 <davidsha> He forwarded me an etherpad with some of the discussions he had outside of the main track: https://etherpad.openstack.org/p/neutron-queens-ptg-ccffwaas-raw-notes-tmorin
14:17:42 <davidsha> Have you all seen the live streamed discussions of the Neutron PTG?
14:18:17 * mlavalle lurks in the shadows
14:18:54 <davidsha> Hey
14:19:13 <bcafarel> I played them while doing other things, guess I missed the ccf discussion :/
14:19:20 <davidsha> people are just reviewing this: https://etherpad.openstack.org/p/neutron-queens-ptg-ccffwaas-raw-notes-tmorin
14:19:29 <igordc> hi mlavalle
14:19:51 <davidsha> bcafarel: I believe the CCF discussion was on the Tuesday morning video
14:19:54 <reedip_> o/
14:19:54 * mlavalle waves to igor
14:20:00 <reedip_> sorry for being late
14:20:05 <igordc> hi reedip_
14:20:08 <davidsha> give me 2 mins and I'll try to grab it
14:20:12 <davidsha> hi
14:21:03 <reedip_> hi igordc
14:21:14 <davidsha> Here is the video: https://www.youtube.com/watch?v=58AyKXHkI-I
14:21:45 <mlavalle> yeap, that's the one
14:22:57 <davidsha> So one of the topics tmorin noted in his etherpad was tracking classification/group usage to know if we can delete them.
14:23:01 <mlavalle> You can also look at the CCF summary here: http://lists.openstack.org/pipermail/openstack-dev/2017-September/122583.html
14:23:50 <davidsha> mlavalle: Thanks! forgot about the mailing list summary.
14:24:52 <davidsha> I'll copy the CCF one into tmorin's etherpad actually
14:25:34 <reedip_> yep that has the gist of the meeting, I remember the fwaas discussion for ccf :)
14:25:48 <igordc> davidsha: cool I'll fix the spec, the one attached seems to be the flow manager
14:26:21 <davidsha> igordc: your right.
14:27:08 <reedip_> guys brb
14:28:15 <davidsha> kk, just on the etherpad, the first 2 topics, knowing that a classification/group has been consumed and classifications being read/write. Any comments?
14:29:07 <davidsha> On the second one, I believe classifications should be read only otherwise we have the update problem.
14:29:27 <igordc> davidsha: I tend towards read-only
14:29:56 <davidsha> bcafarel: your thought?
14:30:33 <bcafarel> davidsha: sorry, was looking for the merged spec
14:30:43 <davidsha> bcafarel: no problem
14:31:03 <davidsha> bcafarel: this one isn't it: https://review.openstack.org/#/c/333993/
14:31:03 <patchbot> patch 333993 - neutron-specs - Neutron Common Classification Framework (MERGED)
14:31:29 <bcafarel> but leaning towards read-only too, updating one and potentially syncing a firewall, a SFC, ... on it does not look like a good idea
14:31:31 <igordc> oh... what is this?
14:31:55 <davidsha> patchbot?
14:32:16 <igordc> davidsha: yeah... oh because bcafarel mentioned it
14:32:22 <igordc> davidsha: you*
14:33:02 <davidsha> Ok, so read only is the decision. What about tracking classification usage?
14:33:57 <davidsha> In earlier patches I could track when a classification was consumed through RPC, however that was against the idea of keeping this at the service plugin layer.
14:35:02 <bcafarel> I guess tracking usage is mandatory if we want to keep the feature consuming services do not need to keep a local copy of the classification
14:35:06 <davidsha> Any suggestions? igordc, bcafarel, mlavalle?
14:35:29 <bcafarel> so we can allow deleting CCF resources only if they are not in use
14:35:38 <igordc> davidsha: I see the tracking useful at least to check if classifcations can be deletd (not being consumed)
14:35:50 <igordc> davidsha: but if there's an alternative to tracking it's also fine I think
14:36:20 <igordc> and now patchbot has quit...
14:36:24 <davidsha> igordc: I think the alternative is not allow deletion :P
14:37:01 <davidsha> bcafarel: Ya
14:37:01 <igordc> davidsha: yes, that is a perfect solution to this problem! KISS, very S
14:37:51 <davidsha> Lol
14:38:17 <mlavalle> KISS works for me
14:39:00 <davidsha> mlavalle is throwing his weight behind not being able to delete the classification resource, a decision has been made ;)
14:39:35 <mlavalle> mhhh, I'm going to be blamed at some point in time ;-)
14:40:30 <davidsha> Lol, we can do some more path finding for this as the PoCs are being developed. so we can kick this can down the road I guess.
14:41:24 <igordc> davidsha: yes but the final impl on this aspect should be before v1
14:41:31 <davidsha> There is mention of FWaas drafting a spec for this and BGP VPn doing the same, so I'll need to find those when they go up
14:41:39 <davidsha> igordc: Of course
14:42:13 <bcafarel> no progress on the SFC side, was too busy on other fronts
14:42:21 <bcafarel> davidsha: so I won't block you! (for now)
14:43:12 <davidsha> bcafarel: same for me, I was working on a qos one, but it's been lost in some obscure branch name now....
14:44:12 <davidsha> The next thing is horizon, what will we need to do to make a component?
14:45:06 <davidsha> Can we extend Horizon from the Neutron Classifier repo like the openstack client or do we need to submit something to horizon?
14:45:52 <igordc> davidsha: I think so, but once ccf becomes neutron-stadium or more than the code would be moved to horizon itself
14:46:08 <igordc> davidsha: which is ccf v2 at least, I expect
14:46:12 <reedip_afk> sorry, my network has issues with me
14:46:30 <davidsha> igordc: ack, something else to revisit in the future.
14:46:36 <davidsha> reedip_afk: np
14:47:01 <bcafarel> yep I don't think keeping it in-house will be a problem short-term
14:47:19 <davidsha> Ok, that sounds like a plan then#
14:47:36 <bcafarel> so akihiro was a volunteer to get it done?
14:47:52 <mlavalle> amotoki: ^^^^
14:48:10 <davidsha> I'm not sure, we'd need tmorin to clarify
14:49:54 <davidsha> We're running out of time, I don't think the points from the summary need to be discussed, will we move onto the next topic?
14:50:45 <davidsha> #topic Call for requirements
14:51:37 <davidsha> There hasn't been any response on the mailing list so I won't bring this up again next meeting
14:52:12 <davidsha> I added it because it was mentioned at the PTG.
14:52:38 <davidsha> So we can move onto open discussion if no one has anything to add.
14:53:02 <mlavalle> do you mean from the FWaaS team and others?
14:53:54 <davidsha> mlavalle: well just in response to the mailing list, I'd need to talk to tmorin and see was it discussed
14:54:39 <mlavalle> ok
14:54:46 <reedip_afk> davidsha : I think as per our last discussion, the protocols mention in the IP_PROTOCOL would be useful, right ?
14:55:02 <davidsha> In neutron lib right?
14:55:15 <reedip_afk> yep
14:55:33 <davidsha> There were a few we were not going to pursue I think but generally yes
14:56:42 <reedip_afk> kk
14:56:47 <davidsha> We're coming up to the end of the hour, I'll switch to open discussion in case people have something they want to bring up
14:56:53 <davidsha> #topic Open discussion
14:57:20 * bcafarel quickly gets the coffee ready
14:57:25 <davidsha> Any topics anyone would like to mention?
14:57:37 <amotoki> I was away from keyboard a while
14:57:40 <davidsha> bcafarel: don't for get to share :P
14:57:41 * igordc forgot post-lunch coffee, will get one now
14:57:58 <igordc> maybe I'll use this time to clarify my positon on the ccf
14:58:16 <mlavalle> I have a few questions
14:58:28 <amotoki> i think horizon support should be done per each *service* project like fwaas, sfc, so I don't think we need ccf-specific support
14:58:53 <igordc> due to other priorities I won't be driving the CCF anymore - however I will continue to follow its developments and provide input if I can
14:59:10 <davidsha> amotoki: If it's a common resource between each service though?
14:59:33 <igordc> amotoki: you do need a specific UI for the definition of classifications
14:59:38 <davidsha> igordc: ack
14:59:49 <amotoki> davidsha: afaik this is the first case, so I think we need more discussion
14:59:55 <davidsha> We're out of time, we'll move this to #openstack-neutron
14:59:57 <bcafarel> igordc: thanks for all the restart/spec effort
15:00:02 <bcafarel> ok
15:00:09 <igordc> bcafarel: ;)
15:00:16 <davidsha> #endmeeting