15:01:51 <ihrachys> #startmeeting neutron_upgrades 15:01:52 <openstack> Meeting started Mon Feb 8 15:01:51 2016 UTC and is due to finish in 60 minutes. The chair is ihrachys. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:55 <openstack> The meeting name has been set to 'neutron_upgrades' 15:01:56 <ihrachys> ok let's get it rolling 15:02:11 <ihrachys> hi all. I am back with you. thanks rossella_s for running the previous meeting! 15:02:20 <korzen> hello 15:02:22 <mhickey> rossella_s: ++ 15:02:26 <sc68cal> o/ 15:02:28 <ihrachys> #topic Announcements 15:02:35 <rossella_s> ihrachys, my pleasure :) glad to have you back! 15:02:52 <ihrachys> pc_m was wondering why we don't have a bug to track our OVO work. so here you go: 15:02:58 <ihrachys> #link https://bugs.launchpad.net/neutron/+bug/1541928 15:02:59 <openstack> Launchpad bug 1541928 in neutron "Adopt oslo.versionedobjects for core resources (ports, networks, subnets, ...)" [Medium,Confirmed] - Assigned to Ihar Hrachyshka (ihar-hrachyshka) 15:03:15 <pc_m> ihrachys: Thanks! 15:03:23 <rossella_s> time to update our patches to use that bug id 15:03:26 <ihrachys> please mention the bug in all your patches for objectification. it should answer some concerns of people outside of the team. 15:03:27 * njohnston is listening in from the back row 15:03:38 <ihrachys> next thing is sprint... 15:03:40 <ihrachys> #topic March code sprint 15:03:53 <ihrachys> so as you all know, we have a code sprint in Brno in the middle of March 15:03:59 <ihrachys> #link http://lists.openstack.org/pipermail/openstack-dev/2016-January/085328.html 15:04:06 <ihrachys> (in case you wonder, it's in Europe) 15:04:19 <ihrachys> we have a nearly empty etherpad for now 15:04:20 <ihrachys> #link https://etherpad.openstack.org/p/code-sprint-neutron-objects-brno 15:04:30 <ihrachys> but it already contains some travel related tips 15:04:40 <ihrachys> please add your nickname in the list of attendees if you plan to go 15:04:52 <ihrachys> I hope to see you all there ;) 15:04:59 <mhickey> ihrachsy: good to see connections from other big hub airports in Europe! :) 15:05:16 <mhickey> ihrachys:^ 15:05:35 <ihrachys> mhickey: you are free to update the etherpad :) 15:05:52 <ihrachys> another thing worth mentioning is that we are almost at M3 stage (that will occur end of Feb), and the release is middle of March I guess 15:06:37 <ihrachys> just so everyone is aware that some patches may be blocked for a bit because of core team struggling to push scheduled deliverables. but I hope it won't happened much. 15:06:45 <ihrachys> ok, now to actual business 15:06:46 <ihrachys> #topic Partial Multinode Grenade 15:07:05 <ihrachys> sc68cal: do you have an update? 15:07:20 <sc68cal> We're going through Neutron and trying to get the MTU issue fixed 15:07:21 <ihrachys> I *presume* not much changed since recently, but maybe I miss smth 15:07:58 <sc68cal> I have a suspicion that the last 3 errors we see with floating IPs is because of an MTU mismatch between router interfaces that Sam-I-Am has discovered - possibly. 15:08:13 <ihrachys> sc68cal: yeah. I see Sam-I-Am testing different scenarios for MTU in his local setup. 15:08:22 <sc68cal> although we should be setting the correct MTU to the guests. So, it's still a mystery really. 15:08:36 <ihrachys> I actually checked that assumption in gate, I am not sure it's the case. 15:09:04 <ihrachys> specifically, I pushed a testing patch https://review.openstack.org/274358 to dump interfaces inside network namespaces 15:09:15 <ihrachys> and checked what's there on internal router legs 15:09:21 <sc68cal> ooooh 15:09:24 <ihrachys> it's the same 1450 (assuming you set 1450 for network_device_mtu) 15:10:04 <sc68cal> Gotcha. Yeah I don't like how we have path_mtu, segment_mtu and network_device_mtu 15:10:12 <sc68cal> lots of opportunity for error 15:10:13 <ihrachys> see here: http://logs.openstack.org/59/265759/20/experimental/gate-grenade-dsvm-neutron-multinode/d64a6e6/logs/worlddump-2016-01-30-164508.txt.gz 15:10:42 <ihrachys> I actually planned to set a devstack setup with 1450 mtu to see whether it experiences the failure then 15:11:00 <ihrachys> but was traveling for the whole week and hence did not have a chance. 15:11:55 <ihrachys> if that's a local mtu issue when using 1450 mtu for network_device_mtu and is not related to internode connectivity, then I could reproduce it I guess with all-in-one 15:12:14 <ihrachys> anyhow, need to poke that a bit more 15:12:48 <sc68cal> yup 15:13:06 <ihrachys> nevertheless, it's great we take MTU thing seriously: for most setups, it's not a realistic case to run with underlying infrastructure that does not support usual 1500 size for frames. 15:13:28 <ihrachys> ok, next... the biggest topic to cover... 15:13:34 <ihrachys> #topic Object implementation 15:13:52 <ihrachys> I guess rossella_s is in better position to update than me. rossella_s, your stage :) 15:14:29 <rossella_s> so as you might see from the list of the patches, there's been a lot of progress 15:14:48 <ihrachys> yeah, FYI the list is somewhere here: 15:14:49 <ihrachys> #link https://wiki.openstack.org/wiki/Meetings/Neutron-Upgrades-Subteam 15:14:59 <rossella_s> mhickey, helped a lot in get the tests passing for the port ovo and 2 of the extensions, thanks! 15:15:18 <ihrachys> a lot of patches in flight. not many merged yet, but that will roll quicker, we just set the stage for now :) 15:15:20 <ihrachys> mhickey++ 15:15:27 <mhickey> np 15:15:42 <rossella_s> korzen, also did a great job with the subnet ovo and with handling composite keys 15:16:04 <ihrachys> yeah, composite key patch is magic 15:16:06 <ihrachys> #link https://review.openstack.org/275790 the magic 15:16:21 * ajo reads magic :) 15:16:26 <rossella_s> slunkad, is working on security group extension, it's still WIP #link https://review.openstack.org/#/c/275664/ 15:17:14 <rossella_s> I am working at handling synthetic fields in a base neutron object, will send a patch tomorrow probably 15:18:33 <rossella_s> other people joined our efforts but I don't see them in the meeting..electrocucaracha, dguitarbite, saisriki 15:18:39 <ihrachys> a lot of work is still ongoing in base class. apparently we haven't covered all use cases when we did qos. :) 15:19:03 <saisriki> Hi rossella_s 15:19:10 <ihrachys> that's an impressive list already. seems like we'll be in good position to make progress on the sprint. 15:19:16 <rossella_s> yep 15:19:22 <rossella_s> saisriki, do you have an update? 15:19:43 <rossella_s> saisriki, I don't see any link for a patch in the team page, do you have one? 15:20:48 <saisriki> Not much. I will update the wiki page. I started working on creating DB fields for Neutron similar to that of Nova 15:21:06 <rossella_s> saisriki, great! if you need any help please reach out 15:21:12 <ihrachys> saisriki: thanks. I am not sure what's the scope. Can you describe which fields do you mean? 15:21:14 <saisriki> ok, will do 15:21:34 <ihrachys> like... custom types? 15:21:40 <saisriki> I have identified IPAddress and CIDR fields for now. I am not sure if I need to add more types 15:21:53 <saisriki> ihrachys: yes, custom types 15:22:02 <ihrachys> oh ok. I thought that CIDR is already supported by OVO library. but if not, yes, we need those. 15:22:08 <korzen> ihrachys, it was raised by rossella_s on the privious meeting, we need SQL alchemy decorator 15:22:17 <ihrachys> note that korzen has the first custom field in review: https://review.openstack.org/273517 15:22:27 <saisriki> Oh, if it is already supported, I will re-check. 15:22:34 <rossella_s> ihrachys those fields are supported by OVO but not by sqlalchemy 15:22:39 <ihrachys> ok, I guess I need to educate myself. will wait for the patches. 15:23:06 <rossella_s> ihrachys, we need a translator...look at nova/db/sqlalchemy/types 15:23:12 <ihrachys> rossella_s: oh interesting. ok, let's discuss it in gerrit once patches are up. 15:23:13 <korzen> the decorator is needed to insert IPAddress and IPNetwork OVO fields to DB 15:23:22 <rossella_s> ihrachys, ok 15:23:38 <ihrachys> anything else object-y worth notion? 15:23:43 <mhickey> yes.. 15:23:46 <ihrachys> shoot 15:23:46 <korzen> I have one 15:24:07 <mhickey> I need some love for following patches: https://review.openstack.org/#/c/273072/ and https://review.openstack.org/#/c/268274/ 15:24:34 <ihrachys> mhickey: yeah right. we'll get to them. I was not in position to review till today. 15:24:50 <mhickey> I understand; just putting it out there. thanks. 15:24:57 <ihrachys> korzen: what's up? 15:25:06 <korzen> can I modify the db_api methods to remove the ID param? https://review.openstack.org/#/c/275790/3/neutron/db/api.py 15:26:12 <ihrachys> korzen: well, till now we tried to avoid that. but if that hugely simplifies our lives, we could reconsider it. OR have new functions with proper signature and deprecate old ones. 15:26:57 <rossella_> sorry I've just got disconnected for few minutes 15:27:06 <korzen> ihrachys I guess that it would look cleaner if the ID would be removed 15:27:23 <mhickey> ++ 15:27:34 <ihrachys> korzen: if we can go with deprecation, I would take the path. 15:27:47 <korzen> ok, sounds good to me 15:28:34 <ihrachys> ok I also encourage folks to check the patch that adds a generic RBAC capability for objects: https://review.openstack.org/#/c/250081/ 15:28:54 <ihrachys> it's very magic and metaclass-y, but gives a promise of easy RBAC integration in the future. 15:29:06 <ihrachys> and we'll need it when we get to porting networks to objects. 15:29:27 <korzen> from software developer ro magician, need to update my CV 15:29:40 <ihrachys> Distinguished Alchemist 15:29:42 <jschwarz> aye, Haim is working very hard to get the RBAC patch for M-3 15:29:42 <korzen> s/ro/to 15:30:09 <jschwarz> reviews will be appreciated for that patch 15:30:22 <mhickey> jschwarz: ack 15:30:28 <ihrachys> oh yeah indeed. it's not much time left to get it in, so we should push. :) 15:30:49 <ihrachys> there are also two patches for rpc callbacks rolling upgrades: https://review.openstack.org/265347 and https://review.openstack.org/268040 from ajo 15:30:55 <ihrachys> the 1st one should be ready for review 15:31:11 <ihrachys> also quite tight on time and it blocks some other initiatives like qos dscp. 15:31:25 <ihrachys> mhickey: you also have the test-only patch: https://review.openstack.org/#/c/270309/ 15:31:34 <ihrachys> mhickey: do we still wait for new OVO release? 15:32:22 <mhickey> ihrachys: yes 15:32:38 <ihrachys> mhickey: ok I gotta request a new release today on oslo meeting 15:32:54 <mhickey> ihrachys: I would be grateful! :) 15:32:56 <ihrachys> and that's about it for objects, I guess. 15:32:59 <ihrachys> #topic Open discussion 15:33:05 <ihrachys> anything we have not covered? 15:33:40 <korzen> my open would be about copying test case from OVO 15:34:01 <korzen> rossella_ ihrachys are we on the same page? 15:34:09 <ihrachys> korzen: right. we cannot rely on non-public code from oslo libraries. 15:34:20 <ihrachys> I hate duplication mysefl 15:34:40 <ihrachys> maybe we could talk to oslo folks to get the test declared public 15:34:52 <ihrachys> I think we already have cases like that 15:34:57 <rossella_> ihrachys, good idea 15:35:05 <ihrachys> f.e. oslo.db provides some test class for migrations 15:35:11 <ihrachys> and we inherit from it in neutron 15:35:27 <ihrachys> ok let me talk to oslo folks today on the matter and we'll see what they thing 15:35:29 <ihrachys> *think 15:35:36 <rossella_> ihrachys, thanks! 15:35:42 <korzen> ok, thx ihrachys 15:35:46 <mhickey> rossella_s: Is there a backlog of objects to be done or is it clear at the moment? 15:37:06 <rossella_> mhickey, thanks for asking...well there's a WIP patch for sec group and I haven't see any patch for port security yet 15:37:36 <rossella_> mhickey, if you have spare cycles I guess we can ask people directly if they need help...it that's something that might interest you 15:37:43 <ihrachys> I think we took on ports and subnets and networks and that's already more than we can manage right now. I suggest we collaborate on existing patches to get them merged. 15:37:56 <slunkad> rossella_: I should be able to complete the sec group one this week.. 15:38:21 <rossella_> slunkad, great 15:38:30 <mhickey> rossella_: ok. ihrachys: agree. I will probably have less time this week with the bug deputy work! :) 15:38:38 <slunkad> at least good enough for review ;) 15:38:50 <rossella_> mhickey, right! thanks for that too! we are here if you need help 15:38:51 <ihrachys> slunkad: no one expects it to shine on first iteration 15:38:57 <korzen> mhickey, I'm busy with subnet, so maybe you can look at network OVO, if interested? 15:39:02 <slunkad> ihrachys: right! 15:39:27 <mhickey> rossella_: thanks for offering. I may come asking! :) 15:39:50 <ihrachys> great. I guess we can make the meeting shorter and spend the next 20 mins to review patches. :) 15:39:54 <mhickey> korzen: I will get back to you. Maybe maxed as have patches to drive through myself. 15:40:00 <ihrachys> thanks all and keep up the good work 15:40:06 <ihrachys> #endmeeting