18:02:55 #startmeeting networking_policy 18:02:56 Meeting started Thu Jun 11 18:02:55 2015 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:58 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:03:01 The meeting name has been set to 'networking_policy' 18:03:20 #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#June_4th.2C_11th_2015 18:04:07 #topic Bugs 18:04:34 the list of “high” is growing 18:04:41 Ajay__ found some more during the week 18:04:55 i am yet to file some sumit will do today 18:05:06 Ajay__: ok :-) 18:05:09 external segment ones 18:05:18 Ajay__: which ones are currently blocking you? 18:05:26 external segment one 18:05:30 #link https://bugs.launchpad.net/group-based-policy/+bug/1463207 18:05:31 Launchpad bug 1463207 in Group Based Policy "Unable to delete external network segment" [High,Confirmed] - Assigned to Ivar Lazzaro (mmaleckk) 18:05:31 hi 18:05:32 where expernal policy update fails 18:05:57 Ajay__: did you log a bug for that one? 18:05:57 sumit_: i need to file that though 18:06:01 ah 18:06:09 sorry, i was not able to get back to your setup on that one 18:06:22 sumit_: external policy update and ping failure parts 18:06:43 Ajay__: please log it and we will treat it as high priority 18:06:56 sumit 18:07:04 will do in next few mins 18:07:14 Ajay__: thanks! 18:07:28 Ajay__: is this one blocking you #link https://bugs.launchpad.net/group-based-policy/+bug/1462024 18:07:29 Launchpad bug 1462024 in Group Based Policy "Concurrent create_policy_target_group call fails" [High,Confirmed] - Assigned to Robert Kukura (rkukura) 18:08:20 sumit_: not blocking really 18:08:41 Ajay__: ok good to know 18:08:51 this one more : https://bugs.launchpad.net/group-based-policy/+bug/1462149 18:08:52 Launchpad bug 1462149 in Group Based Policy "Error while deleting a policy target group" [Undecided,New] 18:08:52 thanfully no critical bugs 18:09:18 SumitNaiksatam: regarding #link https://bugs.launchpad.net/group-based-policy/+bug/1463207 18:09:18 Launchpad bug 1463207 in Group Based Policy "Unable to delete external network segment" [High,Confirmed] - Assigned to Ivar Lazzaro (mmaleckk) 18:09:55 Ajay__: i tried to reproduce but could not, hence was waiting to triage it 18:09:55 SumitNaiksatam: I don't think it's related to the implicit workflow, given that the ES has the same UUID as the deleted one 18:09:57 ivar-lazzaro: yes 18:10:10 SumitNaiksatam: I'll investigate this 18:10:12 ivar-lazzaro: i have traced the code through the debugger 18:10:46 ok 18:11:06 and to the extent i can recollect, the implicit driver goes and reinstates the “default” ES 18:11:25 i believe it happens in the post-commit, but it was a while back when i traced it 18:12:00 Ajay__: for the update external policy issue, will you be able to post your entire neutron log in paste-bin or something? 18:12:13 and link it from launchpad bug? 18:12:43 sumit: will do i have never done that but will ask around and do it 18:13:03 http://paste.openstack.org/ 18:13:27 sumit: thx will do 18:13:32 any other bugs that we need to discuss right now? 18:13:39 not for now 18:13:46 okay, anyone else in the team 18:14:03 if you think that a bug is a blocker for the the kilo release please flag it in launchpad accordingly 18:14:14 and let me know, and I will up the priority to critical 18:14:20 and -> or 18:14:24 sure 18:14:33 hi 18:14:40 that request is to the rest of the team as well 18:14:42 jishnu: ho 18:14:43 hi 18:15:18 #topic Integration and Rally tests 18:15:30 jishnu: joined back at the right time 18:15:41 jishnu: question for you - we are adding new features 18:16:00 jishnu: how do we augment gbpfunc with tests for these new features? 18:16:43 jishnu: there? 18:16:56 if it requires more thought we can take it offline 18:17:35 yes .. we need to augment for sure 18:18:07 jishnu: since the tests are in a different repo, how do we coordinate that? 18:18:21 @Sumit: need a dump on the new feature .. as you said we can discuss details offline 18:18:33 jishnu: ok 18:18:53 jishnu: it might also be good if we have a short documentation on how to add new tests to your test suite 18:19:15 jishnu: that way we can try to scale the process to the rest of the team 18:19:27 SumitNaiksatam: ++ 18:20:03 jishnu: if you are comfortable we can coordinate a one hour session on #openstack-gbp at some time 18:20:06 Yes.. Sumit 18:20:21 and you can perhaps give people a quick intro to the suite 18:20:28 *test suite 18:20:48 jishnu: so let me know accordingly 18:21:25 it would be preferable to update the test suite with the new tests, before we can release kilo 18:21:30 so lets plan accordingly 18:21:33 Will you be in office in tomorow .. we can do it tomorrow .. 18:21:59 jishnu: yes, i meant for the whole team, so we can do IRC or we can do webex 18:22:10 jishnu: okay so lets discuss the logistics offline 18:22:15 jishnu: thanks 18:22:17 yes 18:22:23 Ajay__: over to you for the Rally update 18:22:49 yes rally managed to integrate the cleanup context as we spoke to clean up resources 18:23:05 now trying to add tests for external segment/external policy 18:23:12 and explicit l2/l3 policy 18:23:14 Ajay__: thanks for doing that, and thanks for explaining the various pieces involved 18:23:40 so all i need now is for mandeep to create a fork in noironetworks github 18:23:43 per discussion last week, it might not be as easy to just add the rally scenarios to the gbp repo 18:23:58 for now we can go with a rally fork 18:24:03 Ajay__: will get you the repo 18:24:14 and speak to Boris if he will allow GBP tests upstream 18:24:25 since code changes are required to Rally core itself 18:24:31 Ajay__: yes 18:24:39 yes 18:24:49 so for now we will run Rally from Ajay__’s branch 18:25:06 Ajay__: so on that, we also need to start a new upstream job for this 18:25:23 yes we do 18:25:34 Ajay__: either you can do that, or we can discuss as to what is needed and i can put the infra patch 18:25:55 Ajay__: so first lets find out how we need to set it up (in devstack) 18:25:56 i need help from you i dont have a jenkins server to use 18:26:13 Ajay__: ok lets take this offline (first lets get our code into the repo) 18:26:25 yes correct 18:26:28 any questions for Ajay__ on this? 18:26:41 or for Jishnu on the integration tests? 18:27:11 Ajay__: thanks for the update 18:27:11 i presume no 18:27:23 #topic Packaging update 18:27:41 rkukura: any updates on this front (for fedora/kilo-3)? 18:28:22 No, I haven’t been working on the packaging recently. 18:29:09 rkukura: okay, so we are not planning to do the k3 packages any more? 18:29:36 I’d like to, but am not sure of its priority relative to other work. 18:30:03 rkukura: okay 18:30:16 amit does not seem to be around 18:30:34 #topic Kilo features 18:30:58 the udpates to the service chaining refactor specs have been merged: #link https://review.openstack.org/#/c/174118 18:31:15 thanks ivar-lazzaro for the updates and to others for the review 18:31:44 SumitNaiksatam: :) 18:31:49 the initial impl patches also merged yesterday 18:31:56 thanks again ivar-lazzaro, awesome work 18:32:03 rkukura: saw your comments post the merge 18:32:16 the one on the resrouce counts is very valid 18:32:20 yes - minor possible issues with two of them 18:32:30 i believe it gets used for the quota check as well 18:32:35 yeah I saw that, I'll post an update soon 18:32:42 ivar-lazzaro: thanks 18:32:43 so if its broken we need to fix 18:32:48 service target is not a visible object 18:33:00 so it's not affected by quota 18:33:13 well, it is indirectly since it created a PT 18:33:20 ivar-lazzaro: i believe there was a comment on the service profile count as well? 18:33:25 creates* 18:33:40 that comment was about *removing* the count 18:33:47 unless I missed something recent 18:34:01 it looked like an existing count function got renamed rather than a new one added 18:34:17 rkukura: do you have a pointer? 18:34:50 the remaining patches for this work: #link https://review.openstack.org/#/q/status:open+project:stackforge/group-based-policy+branch:master+topic:bp/node-centric-chain-plugin,n,z 18:35:12 ivar-lazzaro: line 490 in https://review.openstack.org/#/c/179691/20/gbpservice/neutron/db/servicechain_db.py 18:35:13 also there is igordcard’s patch: #link https://review.openstack.org/168733 18:36:18 rkukura: ok found it! Good catch 18:36:24 Mine's not yet for review 18:36:30 the servicechain instances is not visible to tenant, so its okay to not have, but better to keep it 18:36:34 rkukura: not sure how that method was replaced, probably some merge mess I did 18:36:58 the method should definitely be renamed in the plural form, since that is the contract 18:36:59 rkukura: please file a bug and I'll post a patch asap 18:37:02 rkukura: good catch 18:37:34 ivar-lazzaro: i would think its okay to post a follow up just a partial impl of the blueprint too, but bug is just as good 18:37:44 igordcard: okay 18:38:05 SumitNaiksatam: ok 18:38:25 any questions for ivar-lazzaro on this topic? 18:38:52 it would have been good to discuss the heat driver that magesh is writing, but i dont think he is here 18:39:13 rkukura: anything you want to discuss regarding the VM driver that you are planning? 18:39:22 not yet 18:39:27 rkukura: okay 18:40:00 ivar-lazzaro: thanks for the update, and again, amazing work with implementing this feature! 18:40:46 thanks! 18:40:51 the floating IPs for PTs in a PTG was merged earlier #link https://review.openstack.org/167174 18:41:12 the patch was in review for a long time 18:42:17 i think rukhsana had questions on #link https://review.openstack.org/179327 18:42:57 but i dont see her here 18:43:34 that one needs rebasing again 18:43:54 is the neutron issue resolved? 18:44:03 rkukura: we believe yes 18:44:08 the patch was merged 18:44:25 great - I had given this a +2, so should be ablr to quickly re-review 18:44:54 rkukura: okay 18:45:25 there was another issue that Ajay__ brought up 18:45:32 regarding quota support 18:45:51 he was not seeing the Neutron quotas being enforced 18:46:01 and i believe the reason is becasue we are making internal calls 18:46:08 and hence its not hitting the quota check 18:46:17 that would make sense 18:46:28 rkukura: you also have comments in the RMD as general TODOs 18:47:01 so we need to figure out a strategy to get this to work for kilo 18:47:04 Ajay__: there? 18:47:12 SumitNaiksatam: agreed 18:47:19 yes 18:47:30 u summarized it 18:47:44 Ajay__: so this is an absolute requirement for your use case? 18:48:06 Hi 18:48:07 the neutron quota enforcement, that is? 18:48:12 i would say when you have multiple tenants yes we need some form of enforcement 18:48:14 rukansari: hi, just a min 18:48:32 else any one tenant can exhaust resources 18:48:40 Ajay__: we will definitely try to incorporate the GBP quotas (which we dont have today either) 18:49:28 Ajay__: i was thinking if it would be enough to enforce at the GBP resouce quota level 18:49:43 enough -> acceptable (in this iteration) 18:50:04 yes that would be good 18:50:10 Ajay__: perhaps we can think over it a little more 18:50:45 i feel the short term check for quota is not really useful for us since we need the long term GBP quota implementation 18:50:45 rukansari: you had questions regarding #link https://review.openstack.org/179327 18:50:50 short term is ok i feel 18:51:05 Ajay__: yes, we will implement quota checks for GBP 18:51:18 Ajay__: thanks 18:51:21 rukansari: go ahead 18:51:31 we have 9 mins 18:51:40 ok 18:52:25 The question I have is w.r.t sharing PRS pacth:https://review.openstack.org/#/c/179327/ 18:52:34 Will this patch address the requirement: 18:53:00 1. Need 1 external policy with 1 external segment for data connectivity to the Internet with its own L3Policy 2. Need another external policy with 1 external segment for service management network with its own L3policy 18:54:06 Would a WIP patch that has a very specific usecae for the reqs above be acceptable: i.e remove the validations in: 18:54:06 https://github.com/noironetworks/group-based-policy/blob/master/gbpservice/neutron/services/grouppolicy/drivers/resource_mapping.py#L1116 18:54:49 rukansari: you can definitively try 18:55:04 rukansari: I'm not sure that removing the validation is the only thing needed 18:55:27 ivar: did try and it seems to satisfy our specific usecase 18:55:34 rukansari: but in theory using the remote_sg manager should solve the problem 18:56:05 rukansari: no harm in posting a WIP patch, but what would be the final outcome of that patch? 18:56:17 rukansari: nice, for a proper upstream fix I would suggest we use the sg_managers validations 18:56:24 ivar: OK. So the SG mgr should be the final solution. 18:56:37 rkukura: something on top of it 18:56:41 ivar-lazzaro: one request - can we document somewhere on teh wiki page as to what are the current constraints of using the ES and the EP (with teh RMD)? 18:56:42 rukansari: ^^ 18:56:56 ivar: agreed. proper solution is to use SG mgr 18:57:49 rukansari: we don't want to break the existing deployments (especially when we backport) so we need to keep the current behavior when people use the remote subnet SG manager 18:58:30 as long as that requirement is fulfilled, we can experiment with anything else :) 18:58:41 ivar-lazzaro: can you clarify "we need to keep the current behavior when people use the remote subnet SG manager" 18:58:46 SumitNaiksatam: ok 18:59:03 rkukura: the new SG patch has 2 different sg manager 18:59:10 rukansari: ^^ 18:59:36 rukansari: one is a backward compatible sg manager, which uses remote subnets just as we do today 18:59:55 rukansari: when that manager runs, we need to keep the same limitations we have had so far 19:00:05 ivar-lazzaro:got it! 19:00:26 rukansari: hence the suggestion to document the limitations/constraints so that everyone is on the same page 19:00:40 alright we are the hour 19:00:45 #topic Open Discussion 19:00:55 anything we missed? 19:01:14 (name change is still on the table, not forgotten!) 19:01:34 SumitNaiksatam: my favorite! 19:01:38 ivar-lazzaro: :-) 19:01:41 Have a question w.r.t: https://review.openstack.org/#/c/166424/8/gbpservice/neutron/services/servicechain/drivers/simplechain_driver.py 19:02:07 rukansari: can we take it to #openstack-gbp? 19:02:14 we are passed the time 19:02:22 Is this an issue that that this admin user has to be a member of each and every tenant? 19:02:25 ok 19:02:33 rukansari: yes, that will be the case 19:02:43 good observation 19:02:59 alright, thanks everyone for joining! 19:03:02 bye 19:03:06 Bye 19:03:06 bye 19:03:08 #endmeeting