15:00:24 #startmeeting neutron_upgrades 15:00:25 Meeting started Mon Apr 3 15:00:24 2017 UTC and is due to finish in 60 minutes. The chair is ihrachys. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:26 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:28 The meeting name has been set to 'neutron_upgrades' 15:01:29 #link https://wiki.openstack.org/wiki/Meetings/Neutron-Upgrades-Subteam Agenda 15:01:38 anyone for the meeting? 15:01:42 hi 15:01:44 o/ 15:01:51 hi 15:01:55 hi 15:01:56 ok good, was checking if I am not alone :) 15:02:25 let's get straight into business 15:02:37 #topic Announcements 15:02:41 o/ 15:02:55 hello 15:03:07 I proposed this to remove korzen from the list of chairs, since he left the team: https://review.openstack.org/452782 15:03:35 no other announcements from my side 15:03:49 #topic Linuxbridge multinode grenade job 15:03:55 manjeets_, your stage 15:04:13 ihrachys, checked l3-logs earlier last week couldn't see anything 15:04:22 i see qrouter namespace being created 15:05:00 and still wasn't lucky enough o get os_timeout experiment get scheduled on rax cloud 15:05:25 enough to test*** 15:05:27 manjeets_, where do you experiment? 15:05:36 manjeets_, since it doesn't land on rax, does it pass? 15:05:42 ihrachys, ovo patches 15:05:51 when it does not land on rax it passes anyway 15:06:05 i was testing timeout just for rax node 15:06:05 oh I thought you would have a separate fake change to play with it 15:06:15 I had a change in devstack-gate 15:06:31 https://review.openstack.org/#/c/448218/ 15:07:03 first I tried forcing concurrency to 2 but it still failed when it got scheduled on rax cloud 15:07:30 so doesn't it prove that it's not concurrency that gets in the way? 15:07:37 yes 15:08:26 concurrency is not the issue I can say 15:09:05 ok so the best course of action then is poking the logs of failed runs, and if there is missing info, consider adding more log messages? 15:09:58 ihrachys, yes but later last week gate wasn't friendly 15:10:46 oh right, there was some log collection breakage 15:10:50 should be fixed by now 15:11:00 it is fixed 15:12:25 ok. I also remember that kevinbenton was going to have a look at the linuxbridge grenade job. Not sure it's still in his focus, but maybe worth checking with him. Maybe he may spot something in the logs that you couldn't. 15:12:55 ok sounds good 15:13:26 ok, let's move on 15:13:29 #topic Mixed server versions 15:14:01 I posted a spec like write-up for /extensions/ cluster convergence at: https://review.openstack.org/#/c/451993/ 15:14:19 I put the patch to WIP because the idea was not discussed by drivers team yet 15:15:14 even with that, reviews may be beneficial. I don't suggest to review it extensively, but comments on the general direction may be helpful at this point. 15:15:42 another bit here is the neutron-db-manage data_migration command: https://review.openstack.org/#/c/432494/ 15:16:07 electrocucaracha, there are some comments there; do you plan to tackle them? 15:16:28 yes, I couldn't have some time this week but hopefully this one 15:16:42 ok, cool 15:17:06 ihrachys: I'll need to touch base with you about the idea of plugins an other things 15:17:13 sure 15:17:25 #topic Object implementation 15:18:20 I see LIKE queries support patch is respinned by tony, nice: https://review.openstack.org/#/c/419152/ 15:20:22 Also I looked at ml2 type allocations adoption: https://review.openstack.org/367810 It seems ready though there is I believe one missing bit in the framework that currently forced dasanind to move a bunch of sqla intensive code into vlanallocation object 15:20:47 we use filter_by(...).update(...) pattern in the current code to 'lock' records 15:21:01 and I think we need an alternative on OVO side 15:21:19 I proposed update_objects here: https://review.openstack.org/452266 (still need to make it pass the gate) 15:21:44 the new api would be modeled similar to e.g. delete_objects 15:21:53 or count 15:21:58 ihrachys: I am working on making the allocation adoption patch dependent on this 15:21:58 e.g. it doesn't return objects 15:22:03 only act on them 15:22:27 dasanind, aye. I think the base patch needs some work, for what I see in gate results 15:23:17 I also had a look at metering patch: https://review.openstack.org/#/c/433125/ it will need some work, I think types picked for object fields are not ideal 15:24:14 ihrachys: thanks for taking a look 15:25:43 from the previous meeting, we also suggested that router extra attrs patch may be ready: https://review.openstack.org/#/c/381209/ 15:26:07 it's currently in merge conflict, so the author will need to respin 15:27:30 sshank: ^^ 15:28:06 #topic Review priorities for the week 15:28:19 I think we already have a bunch. 15:28:39 one is the /extensions/ convergence spec: https://review.openstack.org/#/c/451993/ 15:29:08 another is new CLI (pending respin): https://review.openstack.org/#/c/432494/ 15:29:23 LIKE support: https://review.openstack.org/#/c/419152/ 15:29:46 update_objects: https://review.openstack.org/452266 (pending respin) 15:29:58 and endpoints: https://review.openstack.org/367810 (pending respin) 15:30:32 oh, finally, router extra attrs: https://review.openstack.org/381209 (pending respin) 15:30:51 I think relistically it's already quite a lot to keep us busy 15:31:23 unless I miss smth (in which case please advise) 15:32:09 ihrachys: well sindhu needs some help in how to address few things in this one https://review.openstack.org/#/c/361443/ 15:33:01 ihrachys: specially for the impact that moving/creating a class method https://review.openstack.org/#/c/361443/40/neutron/objects/l3_hamode.py@52 15:33:26 electrocucaracha: yes, thanks for bringing it up :) 15:35:29 ok will try to have a look 15:35:53 #topic Open discussion 15:36:05 I don't have anything; do you? 15:37:41 ok let's wrap up then 15:37:43 thanks folks 15:37:45 #endmeeting