14:02:20 #startmeeting neutron_upgrades 14:02:21 Meeting started Thu Jan 11 14:02:20 2018 UTC and is due to finish in 60 minutes. The chair is ihrachys. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:02:24 The meeting name has been set to 'neutron_upgrades' 14:02:35 o/ 14:02:41 hey, sorry for late start 14:02:52 not a problem at all! 14:02:54 there were no action items 14:03:15 as some of you may have noticed, gate is unstable to the point where we can't land much these days 14:03:57 yeah, portbinding patch has gone through around 10 recheck without passing zuul, LOL 14:04:02 Hi everybody :) 14:04:07 hi 14:04:22 multiple reasons: meltdown CVE patched in infra slowed down some jobs; infra patching services triggered POST_FAILUREs and TIMEOUTs; there are duplicate jobs because while we are moving from project-config jobs to in-tree jobs for zuulv3... 14:04:46 I don't think we will be able to land much until we at least fix gate setup for zuulv3 14:05:02 ok, understood 14:05:06 it doesn't stop us from working on patches and reviewing them, but sets expectations properly 14:05:21 thanks for the info, Ihar 14:05:46 oh good news - duplicate jobs are mostly gone: https://review.openstack.org/#/c/530500/ 14:05:57 so rechecks may be more effective now 14:06:49 +1 14:06:51 anyway... let's dive into patches 14:07:01 https://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/adopt-oslo-versioned-objects-for-db 14:07:27 before we go through those requiring reviews, let's (re)check those with W+1 14:07:53 ok I see both patches are rechecked 14:08:02 not sure if before or after the project-config change landed. 14:08:09 so it may need another round of rechecks :) 14:08:23 ok now top to bottom 14:08:37 first is https://review.openstack.org/521797 "Use Router OVO in external_net_db" 14:10:12 Is there anything needs to be concerned? 14:10:21 I am looking right now 14:10:29 a minor question: https://review.openstack.org/#/c/521797/16..21/neutron/db/external_net_db.py 14:10:41 why do we need to call .all() if you iterate through it right away? 14:10:56 I think when you iterate, it will already call .all() under the hood 14:11:04 (or better, will fetch records lazily) 14:11:29 Oh, i didn't know it 14:11:40 I'll check again 14:11:54 I think that's what will happen. it's a minor thing, not a bother really. 14:12:45 Ok 14:14:35 ok this is more serious: https://review.openstack.org/#/c/521797/21/neutron/objects/router.py 14:15:34 I don't think the code is correct 14:16:28 and I would say, at this point we should first fix test coverage for the code because it's clearly not covered properly if this didn't trigger an error. 14:16:36 so I am with slaweq on adding a UT 14:17:38 why is 'gw_port_id' not updatable? 14:18:56 Hmm, let's me recall 14:19:41 it's a foreign key so maybe that 14:20:03 I'll check again tomorrow 14:20:10 though it should be able to be None I guess? 14:20:17 yes 14:20:26 nullable is true 14:21:06 good question 14:21:35 lujinluo, please leave a comment for that one there 14:22:16 done 14:22:42 ok cool 14:22:48 so clearly there is more work to do there 14:22:59 moving on 14:23:04 https://review.openstack.org/#/c/507772/ "Use Network OVO in db_base_plugin" 14:23:21 I’ve already updated the patch as your recommendation and still continue working on it 14:23:35 At this moment, I’m working on this failed unit test: 14:23:36 “neutron.tests.unit.plugins.ml2.test_plugin.TestNetworksV2.test_update_network_set_not_shared_other_tenant_returns_409” 14:23:36 It’s failing at create_port step 14:24:10 14:24:17 It looks like we have problem with “obj_context.session.refresh” 14:24:17 I’m trying to apply the same solution from xujun’s patch set 14:24:17 https://review.openstack.org/#/c/530182/17/neutron/objects/base.py 14:24:30 14:24:32 do you have any significant comments from previous patchsets not addressed yet? 14:24:34 But I’m not sure if it’ll work, so any suggestion is welcome 14:25:07 hmm, let me see .. 14:26:13 ok, I think all of your comments have been addressed in the latest patch 14:26:21 PS22 14:27:25 my main concern at this moment is above failed unit test: "“neutron.tests.unit.plugins.ml2.test_plugin.TestNetworksV2.test_update_network_set_not_shared_other_tenant_returns_409” 14:28:51 am I looking at the very latest patchset? 14:28:52 22 14:29:07 because I see some replies that something is Done but it's the smae 14:29:09 *same 14:29:22 could you please point out which one? 14:29:37 like https://review.openstack.org/#/c/507772/19..22/neutron/plugins/ml2/plugin.py line 888 14:29:43 or 874 same file 14:30:22 https://review.openstack.org/#/c/507772/19..22/neutron/objects/network.py line 313 seems wrong - you ripped off implementation; I think you should have it there, otherwise it won't detect existing tenants relying on a rbac entry 14:30:45 also https://review.openstack.org/#/c/507772/19..22/neutron/db/db_base_plugin_common.py line 299 14:30:55 it says you went with all Network but the code is same 14:31:58 so I am not sure if it's latest / all comments handled. 14:32:10 the missing implementation for bound tenants is especially concerning 14:32:22 because it should have functional impact on the object 14:32:33 we can't have this merged without rbac behavior fixed for the object 14:32:34 hmm, lot's of concern, so let me answer one by one, is that ok? 14:33:01 https://review.openstack.org/#/c/507772/19..22/neutron/plugins/ml2/plugin.py line 888 14:34:00 yeah go ahead. maybe not every one but this one and bound tenants implementation may be worth a discussion right now 14:34:20 or if you have legit answers to each, we can as well follow up in gerrit 14:34:28 somehow I missed this one, thank you for pointing that out 14:34:39 yeah, better to follow up on gerrit 14:35:01 about bound tenants implementation, I'll update it as your recommendation 14:35:07 thanks a lot, Ihar 14:35:13 ok cool 14:35:35 next in list is https://review.openstack.org/530182 "Use Router OVO in l3_db" 14:35:43 I don't think I ever checked it 14:36:39 this is probably the most controversial thing there: https://review.openstack.org/#/c/530182/17/neutron/objects/base.py 14:36:43 (refresh OVO method) 14:37:53 it looks like xujun is not online 14:38:05 is that ok? 14:38:11 though it kinda makes sense to reflect hacks we already have with sqlalchemy 14:39:11 TuanVu, yeah I think we will move on. I will leave feedback there. there is another issue with 'description' field there (that is supposed to be introduced by standard attributes logic already) 14:40:11 next is https://review.openstack.org/#/c/506037/ "Part II of Integrate Port OVO" 14:40:21 lujinluo, I assume we were waiting for port binding to land 14:40:37 which should hopefully be next days granted the gate works 14:40:52 besides that, it also needs some extra work 14:40:58 so no hurry ;) 14:41:09 oh ok. would you benefit from a review iteration? 14:41:40 no, maybe after i make it pass all the tests 14:42:10 ok good 14:42:38 there are no patches in queue that seem ready for a discussion otherwise 14:43:48 anything that we missed? (like, patches with wrong topic?) 14:44:40 I have this revert in trunk service to unbreak the way we use engine facades in scope of OVO: https://review.openstack.org/#/c/532343/ 14:44:50 that's similar to IPAllocation issue that I was talking a while back 14:46:11 ok seems like nothing 14:46:36 I walked through etherpad with remaining bits: http://etherpad.openstack.org/p/neutron-remaining-ovo and stroke through those patches that merged 14:47:30 oh thanks Ihar 14:47:52 port binding section should be mostly gone this week 14:47:59 thanks for your update, Ihar 14:48:00 there is lots of work still though 14:48:08 yeah 14:48:11 ok, anything else you want to discuss? 14:48:24 none from me 14:48:31 me either 14:49:01 ok 14:49:21 if you have cycles and have no patches on your plate, etherpad would be a good place to look for next thing 14:49:27 thanks everyone for joining 14:49:31 thanks everyone! 14:49:34 we meet next week! 14:49:34 #endmeeting