18:01:41 #startmeeting networking_policy 18:01:42 Meeting started Thu Mar 26 18:01:41 2015 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:43 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:01:46 The meeting name has been set to 'networking_policy' 18:02:04 #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#March_26th.2C_2015 18:02:24 anyone want to announce anything? 18:02:41 #topic Bugs 18:02:43 hello 18:02:55 Shall we list the accepted vancouver talks GBP related? 18:03:19 ivar-lazzaro: sure, lets do that in the open discussion 18:03:39 ack 18:03:39 ivar-lazzaro: lets also update the wiki page (i have the proposed talks listed at the bottom) 18:03:56 ivar-lazzaro: since we are at it, please go ahead :-) 18:04:24 ivar-lazzaro: i mean announce the talk that got accepted 18:04:31 SumitNaiksatam: oh ok 18:04:53 GBP Lab session got accepted (let me retrieve the link) 18:04:56 #undo 18:04:57 Removing item from minutes: 18:05:18 ivar-lazzaro: yeah, thats great 18:05:32 #link https://openstacksummitmay2015vancouver.sched.org/event/fe712eb5f27186ffd64ed787244af4f3?iframe=yes&w=i:0;&sidebar=no&bg=no#?iframe=yes&w=i:0;&sidebar=no&bg=no 18:05:36 so the hands-on lab session that ivar-lazzaro and rkukura had proposed has been accepted ^^^ 18:06:13 we are all on the hook for this one, i would request everyone to participate the preparation for this one 18:06:16 Everyone's input/help is accepted of course! That's a great opportunity to show GBP up and running 18:06:28 SumitNaiksatam: ++ 18:06:42 and also please make yourself available at the time of the session, since we will need voluteers to help with conducting the session 18:06:43 alright! 18:06:57 and of course the team here is the expert on GBP ;-) 18:07:00 Is there anything else? I don't remember all the proposals 18:07:09 great! 18:07:19 there are at least a couple of more “user-story” kind of talks 18:07:30 i dont have the information handy, but i will update the wiki page with those 18:07:30 ivar-lazzaro: oh well, it conflicts with my other talk... 18:07:48 ivar-lazzaro: nice picture, though :-) 18:07:48 s3wong: darn 18:07:57 if you have a GBP related talk that got accepted, please add here: 18:08:12 #link https://wiki.openstack.org/wiki/GroupBasedPolicy#Team_Activities 18:08:19 s3wong: thanks :) 18:08:27 we will change the “proposed” talks to “accepted” talks 18:08:41 ivar-lazzaro: lets discuss more on the hands on lab in the open discussion 18:08:42 SumitNaiksatam: my accepted talk is about Tacker 18:08:54 s3wong: ah great, very relevant 18:09:20 ivar-lazzaro: perhaps we can have a standing item on the agenda in the lead up to the conference to discuss the hands-on-lab session 18:09:37 so that we are all well prepared 18:10:13 anyone else with any other announcements? 18:10:26 ok moving on 18:10:32 #topic Bugs 18:10:47 magesh: how is this coming along: #link https://bugs.launchpad.net/group-based-policy/+bug/1433530 18:10:48 Launchpad bug 1433530 in Group Based Policy "GBP Kilo release should be in sync with Neutron Kilo" [Critical,In progress] - Assigned to Magesh GV (magesh-gv) 18:11:42 SumitNaiksatam: Ran into a lot of issues on this, but most are fixed now, just a few more UTs failing still 18:11:53 magesh: ah okay 18:12:07 magesh: thanks for following up individually with the vendor driver owners on this 18:12:17 magesh: are any of those blocking you at this point? 18:12:24 SumitNaiksatam: this needs a thorough devstack testing as well 18:12:40 magesh: yes 18:12:58 SumitNaiksatam: We may also parallely need a devstack branch updated to master + gbp 18:13:30 magesh: yes lets discuss that offline, since i am trying to get the gate job to work 18:13:46 magesh: and it will be nice to have that in place to test 18:14:02 SumitNaiksatam: ok, I think the pending issues should be sorted out in a day or two 18:14:22 magesh: ok, anything you want to discuss with the rest of the team here on that? 18:15:08 SumitNaiksatam: sure, so basically delete External segment doesnt seem to happen when run with just the implicit policy driver 18:15:09 SumitNaiksatam: You have a commenty on https://review.openstack.org/#/c/165378/2/test-requirements.txt suggesting pinning to a neutronclient commit, right? 18:15:44 The delete api request is not throwing any error, but the resource is not getting cleaned up in UTs 18:15:45 rkukura: yes, i was more asking 18:16:19 ivar-lazzaro: can you take a quick look at the issue magesh is running into? 18:16:33 rkukura: that is the client counter-part of what magesh is doing 18:16:50 SumitNaiksatam: sure 18:16:54 rkukura: the server side patch is: #link https://review.openstack.org/165377 18:17:03 sorry should have posted that link earlier 18:17:09 ivar-lazzaro: great, thanks! 18:17:34 rkukura, SumitNaiksatam: For both the client and server, should we pin on some commit now or wait till may be kilo rc1 is cut on Neutron 18:18:00 ivar-lazzaro: thanks 18:18:03 magesh: i dont think we need to pin on commit for neutron server 18:18:03 I think we should track master for the server, but maybe pin the client 18:18:14 magesh: at least until kilo release 18:18:23 yeah basically what rkukura said 18:18:31 rkukura, SumitNaiksatam: okay 18:18:43 magesh: my understanding is that we are supporting 2.3.10 neutron client, right? 18:18:53 magesh: are is there an even later version? 18:19:01 magesh: Is making the IPD usable without the RMD being handled as part of this same patch? 18:19:13 magesh: the Jenkins logs are weird 18:19:14 rkukura: i hope not 18:19:21 magesh: 'Zero tests were run. At least one test should have been run.' 18:19:27 SumitNaiksatam: I did not check the corresponding released versions, will check 18:19:36 magesh: ok 18:20:22 SumitNaiksatam, magesh: The bit about “… with just the implicit policy driver” above, plus some code I noticed in the patch, seem to indicate this might be the case 18:20:36 rkukura: yeah 18:20:46 magesh: rkukura’s question ^^^ 18:20:58 ivar-lazzaro: I think it might be some temporary issue, with the latest patch, when I ran UTs all tests were successfull except 5 18:21:28 rkukura: No, that is not handled as part of this patch 18:21:44 SumitNaiksatam: rkukura: There have always been unit tests for the IPD alone 18:21:58 ivar-lazzaro: ah okay got it, magesh was referring to those 18:22:04 the issue we have is that RMD doesn't work without IPD, not the way around 18:22:11 ivar-lazzaro: true 18:22:15 right, the UTs for IPD doesnt run with RMD configured 18:22:28 magesh: Are the changes in https://review.openstack.org/#/c/165377/10/gbpservice/neutron/extensions/group_policy_mapping.py needed due to something that changed in neutron during kilo? 18:22:31 magesh: I sure hope not ;) 18:24:06 rkukura: I couldnt get what caused this, but I think the L4-L7 service splitting from Neutron may have something to do with this 18:24:28 magesh: that sounds plausible - thanks 18:24:36 rkukura: Basically here the extensions got initialized twice and it was causing an issue 18:24:42 But that operation is already done here #link https://github.com/stackforge/group-based-policy/blob/master/gbpservice/neutron/extensions/group_policy.py#L34 18:24:43 magesh: thats interesting, we already have that delta in the group_policy.py 18:24:52 ivar-lazzaro: yeah 18:25:13 and that is imported by group_policy_mapping too, so it should run 18:25:22 magesh: i dont think the split should have an impact 18:25:49 rkukura: good point, please comment on the patch 18:25:55 SumitNaiksatam: OK 18:25:59 so we are 25 mins into the meeting, lets move on 18:26:10 magesh: thanks for the update, lets work to get the loose ends tied on this 18:26:21 SumitNaiksatam: ok 18:26:33 #link https://bugs.launchpad.net/group-based-policy/+bug/1432779 18:26:34 Launchpad bug 1432779 in Group Based Policy "redirect actions don't work with external policies" [Critical,In progress] - Assigned to Ivar Lazzaro (mmaleckk) 18:26:41 this in review 18:26:51 #link https://review.openstack.org/164920 18:27:00 needs at least one more core’s attention 18:27:11 As per igordcard comment on the ML, I've also enabled EP providing chains 18:27:21 ivar-lazzaro: nice! 18:27:35 and i think igordcard has obliged with a +1 ;-) 18:28:03 ivar-lazzaro: anything more to discuss on that critical bug? 18:28:04 I have one thing to say though, if one of the chain experts here could quickly test it in devstack it would be great :) 18:28:28 ivar-lazzaro: hopefully we can exercise the test suite on this 18:28:49 ivar-lazzaro: however we dont have tests for EP providing chains 18:28:50 ivar-lazzaro: I wanted to do that a few days back, but this got lost from the TODO list, will try this out 18:28:57 SumitNaiksatam: it would be nice to do it before merging though 18:29:04 magesh: thanks! 18:29:36 ivar-lazzaro: i had confidence in this patch because i had seen it work (or at least part of it) in the PoC we were doing here 18:29:42 magesh: I still have some concerns about having the EP providing the chains... but I've done it so that we have both versions available 18:30:09 SumitNaiksatam: this is different, especially if we enable chain providing 18:30:09 ivar-lazzaro: unless there are obvious issues with the patch, i would like to move ahead (in the interest of iterative progress) 18:30:20 move ahead with the patch i mean 18:30:33 okay, any other bugs we need to discuss today? 18:30:43 SumitNaiksatam: I actually wanted to discuss the providing a little bit more... For instance, what happens with the LBs? 18:30:51 ivar-lazzaro: okay, I will go through it, I will follow up on it with you may be offline 18:30:57 SumitNaiksatam: there are no policy targets in that EP 18:30:59 ivar-lazzaro: okay sure 18:31:43 ivar-lazzaro: what happens which LBs? 18:31:48 *with which 18:31:50 SumitNaiksatam: what do we expect? to create a LB without members (should be the behavior today) or to have a specific set of services that are "providable" by EPs? 18:32:25 ivar-lazzaro: correct me if i am wrong, but the EP does not have a policy targets, right? 18:32:42 SumitNaiksatam: exactly 18:32:57 SumitNaiksatam: therefore a LB on that side is completely useless 18:33:15 ivar-lazzaro: so my simplistic assumption would be that there is external/manual configuration 18:33:16 SumitNaiksatam: also, EPs don't have NSP associated with them 18:33:23 ivar-lazzaro: true 18:33:32 perhaps LB is not a good example in this case? 18:33:47 ivar-lazzaro, sumitNaiksatam: EP or EPG ? 18:33:52 it could be a logging service like hemanthravi was mentioning last week 18:34:01 SumitNaiksatam: it is a good example to understand how to deal with unusable services 18:34:04 magesh: EP as in External Policy 18:34:06 magesh: External Policy 18:34:29 got confused with the acronyms ! thanks 18:34:36 magesh: :-) 18:34:47 SumitNaiksatam: yeah I agree with that point, what we miss though is a way to avoid bad things to happen when a PRS with a LB is provided 18:34:56 SumitNaiksatam: LB or any other unusable service 18:35:05 ivar-lazzaro: i think we always make the assumption that not all service-types (or their flavors) can be used everywhere 18:35:37 ivar-lazzaro: the “integrity check” is left to the service-chain provider (at least the way it is today) 18:36:23 and perhaps we are not implementing any robust integrity check either ;-) 18:36:28 SumitNaiksatam: ok, so this part is missing from my implementation 18:36:44 SumitNaiksatam: I don't think so :) but we can address that in a separate issue 18:36:47 this is mostly because we are only tying into services that are provided to us by Neutron 18:37:12 SumitNaiksatam: as long as we all agree that we need to properly screen the service types that an EP can provide 18:37:12 so we made many simplifying assumptions to get off the ground :-) 18:37:25 ivar-lazzaro: oh absolutely agree on that 18:38:08 ivar-lazzaro: thanks for bringing this up 18:38:09 SumitNaiksatam: that makes sense, then I suggest we file an issue regarding this particular problem and we can discuss that 18:38:15 SumitNaiksatam: I'll take the action 18:38:25 ivar-lazzaro: yes, great, i would prefer not to block the current patch 18:38:48 ivar-lazzaro: anything else to discuss on this? 18:39:00 SumitNaiksatam: Nope, let's wait for magesh's approval and we are set :) 18:39:11 ivar-lazzaro: i put a process comment on #link https://review.openstack.org/164907 18:39:22 ivar-lazzaro: other than that patch is good 18:39:56 we are running behind today 18:40:04 sharing is caring! 18:40:05 #topic Functional/Integration Tests 18:40:11 ivar-lazzaro: ;-) 18:40:48 i am still working on the fix this: #link https://review.openstack.org/#/c/161532/ 18:41:09 #topic Packaging update 18:41:21 rkukura: there was a question around CentOS 6.5 support 18:41:33 I didn’t see that. 18:41:42 rkukura: is this tested, and which Fedora packages can be used for this? 18:42:20 My understanding was that RDO does not support EL6 18:42:52 at least for juno and beyond 18:43:09 rkukura: okay, but if someone wants to run CentOS 6.5? 18:43:28 what openstack distro are they using? 18:43:48 rkukura: good question, i dont know :-) 18:44:14 rkukura: okay i will follow up offline with you 18:44:50 moving on 18:44:52 #topic GBP Project Proposal 18:44:58 #link https://review.openstack.org/#/c/161902/ 18:45:10 this has been added to next week’s TC agenda 18:45:27 i believe the meeting is on IRC at 22.00 UTC, March 31st 18:45:40 sorry, 20.00 UTC 18:45:56 #link https://wiki.openstack.org/wiki/Meetings/TechnicalCommittee 18:46:17 our favorite topic 18:46:22 #topic Re-factor Group Based Policy with Neutron RESTful APIs 18:46:40 so last week #link https://review.openstack.org/#/c/159725 was merged 18:46:49 thanks yapeng and Yi 18:47:10 Yi: #link https://review.openstack.org/#/c/156776 is ready to go? 18:47:38 SumitNaiksatam: yes 18:48:15 Yi: okay nice 18:48:18 https://review.openstack.org/#/c/156856/ needs some extra work, but is very close to be done 18:48:21 requesting the cores to take a look 18:48:29 Yi: good, about to ask you that 18:48:45 Yi: yapeng: any blockers? 18:48:54 everything seems fine 18:49:13 Yi: I have a question about the mock_neutron.meta_mock decorator 18:49:20 Ivar helped us a lot, and the patched UT seems fine 18:49:30 yapeng: Yi: as with the discussion with magesh on testing, this needs to be tested thoroughly with devstack 18:49:37 Yi: If I understand correctly, that is needed in all the UTs, is that correct? 18:49:38 ivar-lazzaro: go ahead 18:50:01 ivar-lazzaro: pretty much yes 18:50:20 Yi: isn't there any way to do it in the setUp method then? 18:50:57 Yi: I say that because I'm pretty sure people are going to forget/not know to use it 18:51:16 ivar-lazzaro: how to do it with setUp to patch those methods? 18:52:14 Yi: I have to look at it, but I think you can just do the same thing you to on meta_mock in the setUp 18:52:28 well, my using the patch method instead of the decorator ofc 18:52:33 s/my/by 18:53:09 let me find I link for an example 18:53:23 ivar-lazzaro: that would be great 18:54:08 https://github.com/stackforge/group-based-policy/blob/master/gbpservice/neutron/tests/unit/services/grouppolicy/test_resource_mapping.py#L70-L71 18:54:08 i would prefer if this can be done in the setup as well 18:54:18 Yi: would something like this work? 18:54:40 ivar-lazzaro: let me take a look 18:55:02 Yi: yapeng: thanks for the update 18:55:14 #topic Floating IP support 18:55:14 I will ping you offline if I have any issues.:-) 18:55:18 #link https://review.openstack.org/157298 18:55:30 Yi: ok :) I'll be flying for a couple of days though 18:55:35 magesh: i know you have been busy with the pre-requisite step for this 18:56:39 SumitNaiksatam: yes, but comments are welcome on this patch, this needs some more work 18:56:40 magesh: any update that you want to provide? 18:56:56 Right now, the implementation is like this 18:57:14 magesh: i believe you will get more comments once you post a response to the current comments, and perhpas post a new patch set 18:57:23 #link https://review.openstack.org/157298 18:58:00 ah, my bad the spec isnt updated 18:58:07 magesh: np 18:58:27 #link https://review.openstack.org/167174 18:58:58 I have a WIP patch , may be all of you can take a look at this and see which way we want to head 18:59:11 magesh: ah okay, did not show up on my radar 18:59:18 magesh: have added reviewers now 18:59:25 magesh: thanks for the update there 18:59:33 #topic Open Discussion 18:59:37 we have a minute 18:59:43 anything we missed? 19:00:20 SumitNaiksatam: stable/juno plans? 19:00:37 rkukura: yes, not tagged it, i will try to do that EoD 19:00:43 thanks for bringing that up 19:00:45 SumitNaiksatam: thanks 19:01:01 alrighty thanks everyone for your time today! 19:01:02 bye 19:01:11 #endmeeting