14:02:00 <ihrachyshka> #startmeeting neutron_qos 14:02:02 <openstack> Meeting started Wed Aug 12 14:02:00 2015 UTC and is due to finish in 60 minutes. The chair is ihrachyshka. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:02:06 <openstack> The meeting name has been set to 'neutron_qos' 14:02:10 <ihrachyshka> #topic Announcements 14:02:36 <ihrachyshka> first thing first: hopefully, feature/qos lives its last days: we'll request merge back today. 14:02:48 <ajo> yipieeee :) 14:02:53 * amuller cries 14:02:57 <ihrachyshka> I'll send an email later once the merge patch is up for review, and ajo said he'll post a video with the feature 14:03:19 * ajo ajo is on it :) 14:03:31 <ihrachyshka> the original plan was to request merge on Mon, but on Mon we had some things to fix, like devref or the fact that we limited ingress instead of egress ;) 14:03:34 * jschwarz a video! \o/ 14:03:45 <ihrachyshka> and yesterday we had gate failures that are now fixed 14:04:00 <ihrachyshka> I expect remaining feature/qos patches to merge in an hour. so stay tuned. 14:04:13 <ihrachyshka> any remaining work will proceed in master 14:04:17 <ajo> ack 14:04:27 <ihrachyshka> (unless there are major flaws that will need fix before merge-back) 14:04:31 <ajo> I found a issue with policy.json that will need fix, 14:05:00 <ajo> it will allow a tenant creating a policy, a rule, but crashes when updating a rule 14:05:20 <ajo> tenants should not create qos policies , or rules for this release 14:05:23 <ihrachyshka> good. we have stuff to fix, that's for sure. 14:05:37 <ajo> yes 14:05:45 <ihrachyshka> #topic ongoing work 14:06:43 <ihrachyshka> so I know that moshele was looking into refactoring qos extension drivers API. he has a patch: https://review.openstack.org/211090 but now it's abandoned 14:06:52 <ihrachyshka> moshele, I guess you'll revive it once in master, right? 14:07:03 <moshele> yes 14:07:34 <ihrachyshka> also, jschwarz was working on fullstack tests. and we have one: https://review.openstack.org/202492 but since it requires neutronclient changes first, it will wait 14:07:47 <jschwarz> yes 14:08:26 <ihrachyshka> finally, we have client changes: https://review.openstack.org/189655 and https://review.openstack.org/198277 thanks to vikram, Ramanjaneya and jschwarz for supporting those 14:08:39 <ihrachyshka> I hope we'll get them in quick after neutron merge 14:08:45 <Ramanjaneya> jschwarz: anything pending on neutron-client part? 14:08:47 <ihrachyshka> then we'll be able to proceed on fullstack 14:08:54 <irenab> ihrachyshka: spec update is waiting for second +2 14:09:01 <ihrachyshka> ajo, was that an issue in client to solve with the name filter? ^ 14:09:12 <ajo> ihrachyshka, correct, 14:09:19 <ihrachyshka> oh right. we have spec to update: 14:09:21 <jschwarz> ihrachyshka, as ajo mentioned, we found a but on the neutronclient where the rule might be searched for using a name 14:09:21 <ihrachyshka> #link https://review.openstack.org/199112 14:09:29 <ihrachyshka> please review ^ and we'll make sure it's in :) 14:09:35 <jschwarz> that's of course an error because rules don't have a name. I'm looking into that now 14:09:41 <ajo> thanks jschwarz 14:09:52 <jschwarz> of course 14:10:07 <ihrachyshka> other than that, we will still have some TODOs here and there to solve, but nothing critical. 14:10:57 <ihrachyshka> I guess once client and server and spec are in, we can claim bp is done? fullstack will go as an optional beyond the blueprint I guess since it may take time to get the depending patches in first. 14:11:04 <ihrachyshka> what do you think? 14:11:27 <ihrachyshka> anything failing beyond it should be handled as a bug I thinik 14:11:30 <jschwarz> sounds good to me 14:12:15 <ihrachyshka> ok, I assume silence means everyone agrees :) 14:12:17 <amuller> nay 14:12:20 <ajo> jschwarz, vikram : looking at the spec: 14:12:21 <ajo> neutron qos-bandwidth-limit-rule-update <policy-name-or-id> <rule-id> 14:12:22 <ihrachyshka> NAY! 14:12:25 <amuller> you cannot claim the blueprint is complete without an integration test imo 14:12:27 <ajo> but the client expects the opposite 14:12:35 <ajo> <rule-id> <policy> which is counter intuitive 14:12:40 <ihrachyshka> amuller, hm. like... any other blueprint? 14:12:51 <jschwarz> ajo, well, that's not really something we can change I think... 14:12:51 <amuller> ihrachyshka: can we try and raise the bar? 14:13:11 <amuller> is there anything blocking the fullstack test after merge to master? 14:13:24 <jschwarz> ajo, the lbaas guys did opposite of what you wrote as well.. changing that goes deep into neutronclient mechanics I think 14:13:26 <ihrachyshka> amuller, we obviously can. I just wonder what it buys us. do you think we will stop working on fullstack right after it's marked done? 14:13:41 <ajo> jschwarz, ok, then we need to fix the spec 14:13:44 <ihrachyshka> amuller, I guess we need to go and review patches that jschwarz rebased.. 14:14:14 <jschwarz> amuller, the neutronclient patches also block the fullstack test, and neutronclient is blocked because the gate is broken for that repo. 14:14:19 <ihrachyshka> amuller, ok, if you are against, I take it as order :) 14:14:22 <jschwarz> (for the last week) 14:14:43 <ihrachyshka> amuller, heh. and here you should put your testing hat on! 14:15:01 <amuller> time to fix a gate then... 14:15:21 <ihrachyshka> amuller, are you into the fix? are you in touch with pc_m? 14:15:28 <amuller> not at all 14:15:33 <amuller> I'll do that though 14:15:39 <ihrachyshka> thank you 14:15:48 <ihrachyshka> I think they may not see the end of the tunnel :) 14:16:08 <ihrachyshka> ok, let's move to open agenda 14:16:12 <ihrachyshka> #topic open agenda 14:16:26 <ihrachyshka> ajo, so can you update us about why tenants should not create policies? 14:16:28 <jschwarz> so we need to revisit the policy.json settings 14:16:37 <ajo> ihrachyshka: yes 14:16:50 <ajo> the initial plan was to only let admins create and maintain policies/rules 14:16:59 <ajo> if a policy is shared, a tenant can attach it to a port or net 14:17:02 <ajo> but not otherwise 14:17:16 <ajo> in the future, RBAC will make that more modular 14:17:39 <ajo> people willing to let their tenants do qos themselves, will have to modify policy.json, and we will document that 14:17:49 <ihrachyshka> ajo, but it's supported, right? 14:17:54 <ihrachyshka> with the policy change of course 14:17:58 <ajo> ihrachyshka, yes 14:18:04 <ihrachyshka> we have the code to handle access, kudos to jschwarz 14:18:11 <ajo> jschwarz++ 14:18:12 <jschwarz> ^_^ 14:18:19 <ajo> yes, we have to review and tune the policies 14:18:26 <ajo> and provide the 2nd alternate (open example) in documentation 14:18:28 <ihrachyshka> if we would have time, we could even get RBAC this cycle 14:18:38 <ajo> I'd leave that for M 14:18:46 <ihrachyshka> I think it's quite pluggable the way kevinbenton implements it 14:18:49 <jschwarz> ihrachyshka, RBAC still has unmerged code so... 14:18:52 <ajo> I believe we should focus in polishing, testing and debugging 14:19:01 <jschwarz> i agree with ajo here 14:19:05 <ihrachyshka> jschwarz, well, I'm almost sure it's a matter of days. 14:19:12 <ihrachyshka> I see a lot of review traction around it 14:19:21 <ihrachyshka> ok, let's polish :) 14:19:24 <ihrachyshka> or czech :) 14:19:27 <ajo> lol 14:19:30 <jschwarz> :) 14:19:38 <ajo> I'm more comfortable with spanish 14:19:44 <jschwarz> hebrew? 14:20:02 <ajo> :) 14:20:03 <jlibosva> nah, no czech :* 14:20:05 <jlibosva> :( 14:20:33 <jschwarz> ihrachyshka, in regards to the open agenda; neutronclient still requires a bit of work - there's that rule-naming thing and a few reviews still left unhandled 14:20:36 <ihrachyshka> anything else ongoing or needs discussion? 14:20:46 <ihrachyshka> jschwarz, I bet you are on it 14:20:49 <jschwarz> I'll get to them by tomorrow I think - it will be ready by time for L for sure. 14:20:51 <jschwarz> XD 14:21:02 <ihrachyshka> see?!! I tell ya 14:21:22 <jschwarz> just letting you know so that there are no surprises ;-) 14:21:42 <ihrachyshka> overall people, good work on the branch, I look forward to see us polishing the feature in another one ;) 14:22:00 <irenab> I have f question regarding nova flavor extra_spec settings for qos, but maybe we can discuss it next week, after pre-merge crisis is over 14:22:14 <ihrachyshka> jschwarz, I don't remember when there were any surprises from your side. it's generally like 'yeah, there is an issue. and btw, the fix is up' 14:22:32 <jschwarz> ihrachyshka, thanks ^_^ 14:22:33 <ihrachyshka> irenab, ok cool. 14:22:56 <ajo> irenab, that's documented on the spec, nova can set flavors for qos, but that's not compatible with ours 14:23:08 <irenab> please check the spec update patch and if something else need correction, lets add it 14:23:21 <ajo> irenab, just added a nit now 14:23:24 <ihrachyshka> so if there is nothing critical, I suggest we end the meeting now and I go check the queue on whether I can proceed with merge request already. 14:23:29 <irenab> ajo: I think we should check ifwe can integrate both 14:23:45 <ajo> irenab, yes, we will certanly need to integrate for scheduling / guarantee purposes 14:23:49 <ihrachyshka> ajo, I think it's two parallel things, but let's move it to the next week :) 14:23:55 <ajo> irenab: I'm already talking with nikola dipanov for that :) 14:23:59 <ihrachyshka> ajo++ 14:24:05 <jschwarz> ajo, do you want to work on the spec change re: neutronclient and I'll do the review afterwards? 14:24:05 <irenab> for users its easier to manage per flavor qos requremens 14:24:17 <irenab> ajo: please loop me in 14:24:28 <ajo> irenab, yes, also assigning neutron qos flavors to nova would be cool 14:24:30 <ajo> if that's what you mean :) 14:24:39 <ajo> irenab: I will :) 14:24:55 <irenab> ajo: I think so :-) 14:26:50 <ihrachyshka> ok, I take the silence as the lack of more critical things to discuss 14:27:00 <ihrachyshka> #endmeeting