Thursday, 2018-02-22

*** yamamoto has joined #openstack-meeting-500:08
*** yamamoto has quit IRC00:14
*** mjturek has quit IRC00:58
*** yamamoto has joined #openstack-meeting-501:10
*** yamamoto has quit IRC01:16
*** slaweq has quit IRC01:34
*** slaweq has joined #openstack-meeting-501:35
*** slaweq has quit IRC01:39
*** yamamoto has joined #openstack-meeting-502:12
*** yamamoto has quit IRC02:18
*** yamahata has quit IRC02:34
*** ricolin has joined #openstack-meeting-502:40
*** yamamoto has joined #openstack-meeting-503:14
*** yamamoto has quit IRC03:20
*** yamamoto has joined #openstack-meeting-503:37
*** slaweq has joined #openstack-meeting-503:59
*** slaweq has quit IRC04:04
*** slaweq has joined #openstack-meeting-505:25
*** slaweq has quit IRC05:30
*** tdasilva has quit IRC06:10
*** tdasilva has joined #openstack-meeting-506:25
*** slaweq has joined #openstack-meeting-506:27
*** slaweq has quit IRC06:31
*** slaweq has joined #openstack-meeting-507:45
*** slaweq_ has joined #openstack-meeting-507:47
*** slaweq has quit IRC07:48
*** slaweq has joined #openstack-meeting-507:49
*** slaweq_ has quit IRC07:52
*** slaweq_ has joined #openstack-meeting-508:00
*** slaweq_ has quit IRC08:04
*** MarkBaker_ has quit IRC08:11
*** slaweq_ has joined #openstack-meeting-508:28
*** slaweq_ has quit IRC08:45
*** MarkBaker_ has joined #openstack-meeting-509:31
*** mab__ has joined #openstack-meeting-509:36
*** MarkBaker_ has quit IRC09:38
*** yamamoto has quit IRC10:06
*** mab__ has quit IRC10:12
*** yamahata has joined #openstack-meeting-510:32
*** ricolin_ has joined #openstack-meeting-510:49
*** ricolin has quit IRC10:52
*** kei-ichi_ has joined #openstack-meeting-510:57
*** kei-ichi has quit IRC10:59
*** yamahata has quit IRC11:00
*** kei-ichi_ has quit IRC11:01
*** kei-ichi has joined #openstack-meeting-511:04
*** yamamoto has joined #openstack-meeting-511:06
*** yamamoto has quit IRC11:13
*** mab__ has joined #openstack-meeting-511:19
*** diman has joined #openstack-meeting-512:06
*** yamamoto has joined #openstack-meeting-512:09
*** yamamoto has quit IRC12:14
*** yamamoto has joined #openstack-meeting-512:27
*** wxy has quit IRC12:52
*** wxy has joined #openstack-meeting-512:52
*** zhipeng has joined #openstack-meeting-513:28
*** yamamoto has quit IRC13:34
*** yamamoto has joined #openstack-meeting-513:44
*** yamamoto has quit IRC13:44
*** lujinluo has joined #openstack-meeting-513:50
*** annp has joined #openstack-meeting-513:50
*** TuanVu has joined #openstack-meeting-513:50
*** TuanVu_ has joined #openstack-meeting-513:53
*** TuanVu has quit IRC13:56
*** TuanVu_ has left #openstack-meeting-514:00
*** TuanVu_ has joined #openstack-meeting-514:00
*** ihrachys_ has joined #openstack-meeting-514:05
*** ihrachys has joined #openstack-meeting-514:09
ihrachyseh folks, are you still up for a meeting?14:10
lujinluoi am14:10
ihrachysmy irc client makes me a fool. I just started a meeting in a wrong channel14:10
ihrachysembarrassing14:10
lujinluolol, glad you found out14:10
ihrachys#startmeeting neutron_upgrades14:10
openstackMeeting started Thu Feb 22 14:10:53 2018 UTC and is due to finish in 60 minutes.  The chair is ihrachys. Information about MeetBot at http://wiki.debian.org/MeetBot.14:10
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:10
*** openstack changes topic to " (Meeting topic: neutron_upgrades)"14:10
openstackThe meeting name has been set to 'neutron_upgrades'14:10
ihrachyssorry for that14:11
ihrachys#topic PTG14:11
*** openstack changes topic to "PTG (Meeting topic: neutron_upgrades)"14:11
lujinluonot a problem14:11
ihrachyslujinluo, I thought you were going to Dublin but then I saw you don't, as per ptg etherpad. right?14:12
TuanVu_no problem at all, Ihar :)14:12
lujinluowell, i am going. which etherpad do you mean?14:12
*** zhipeng has quit IRC14:12
lujinluoi remember putting my name in neutron etherpad14:12
ihrachyshm. maybe it was a bad dream lol. let me find it.14:12
lujinluohttps://etherpad.openstack.org/p/neutron-ptg-rocky i put my name here14:13
ihrachysthe pad here: https://etherpad.openstack.org/p/neutron-ptg-rocky14:13
lujinluoLine 2214:13
*** zhipeng has joined #openstack-meeting-514:14
ihrachyslujinluo, oh right! I mixed you and liuyulong in line 35. sorry for that.14:14
ihrachysgreat14:14
lujinluonice14:14
*** derekh has joined #openstack-meeting-514:15
lujinluospeaking of that, since we have resolved the new engine facade conflict, i do not think we will have any OVO specific topics14:15
lujinluoexcepting for neutron-lib ones, proposed by boden?14:15
ihrachyswell, there are some neutron-lib14:15
ihrachysalso, enginefacade issue is not solved until all objects (and their plugin consuming code) switches to the new facade14:16
lujinluoyeah, but boden says he won't be there. so i am wondering how the topics would proceed14:16
lujinluoalso, for the new engine facade, i want to bring this bug to your attention. it is purely about tests refactoring, but i think it worth your attention14:17
ihrachysI walked through his specs that are related to OVO and both seem quite solid to just move forward with them (at least pieces that are OVO specific)14:17
lujinluohttps://bugs.launchpad.net/neutron/+bug/175073514:17
openstackLaunchpad bug 1750735 in neutron "[OVO] UT fails when setting new_facade to True" [Undecided,New] - Assigned to Lujin Luo (luo-lujin)14:17
ihrachysyeah we were in a hurry to deliver that so some tests are missing14:18
lujinluoi went through his two specs as well. but i feel like i need more investigation to understand what he wants to implement, lol14:19
*** hungpv has joined #openstack-meeting-514:19
ihrachyslujinluo, I see you assigned yourself to the bug. so the idea there would be to add a fake test object in test_base, and add a unit test class for that by inheriting from base 'iface' class. we don't necessarily need to switch an object to new_facade to trigger the tests, fake object will do for the start.14:20
lujinluoyeah, i assigned myself but i am afraid i won't have time to code before PTG. but once i get back, i will follow your suggestion14:21
ihrachysit's not critical (yet), as long as it's just unit tests14:22
ihrachysas for boden specs, we can walk through them to see what he's up to14:22
ihrachysyou know the general goal of neutron-lib and what he does, don't you?14:23
ihrachyshe = boden14:23
lujinluoactually no.. that's why i am struggling to understand the big picture14:23
*** hungpv has quit IRC14:23
ihrachysok. so in the past, we had a single neutron tree with all plugins14:24
ihrachyscisco, midonet; lbaas, fwaas, ...14:24
ihrachyseverything imported from neutron.* and it was fine because it's same tree14:24
ihrachysthen the decision was made to split all but essential pieces into separate repos14:24
ihrachyswhich was done around liberty14:24
ihrachys(liberty as in release name)14:25
lujinluohaha14:25
ihrachysbut then the problem became clear that whenever we change smth in neutron repo, we often break some stadium project because now we don't gate on their tests14:25
ihrachysso neutron-lib is an attempt to answer to that challenge14:25
lujinluoi see. this is what happened to bgp this time, right?14:26
ihrachysinstead of subprojects importing from neutron.*, they would import from neutron_lib.* that 1) keeps just the right amount of shared code; and 2) keeps API stability promise to consumers.14:26
lujinluowe reverted some ovo changes in neutron repo, then bgp broke14:26
ihrachyslujinluo, not exactly because what they were hit is something implicit - the context of callback execution changed facades14:27
ihrachysbut failure vector is same14:27
*** mjturek has joined #openstack-meeting-514:27
lujinluoi see. then it makes sense to put ovo into neutron-lib14:27
lujinluoat least part of ovo14:27
ihrachysso now boden works for a bunch of cycles already moving pieces of neutron code into the library14:28
ihrachysof course sometimes we don't just move but modify api a bit so that it makes more sense as public api14:28
lujinluothanks, i got the general idea now14:29
ihrachyslujinluo, yes, one thing is, bagpipe imports neutron.objects.base to define their objects. while we try to keep this interface more or less stable, it may still break at some point.14:29
ihrachysso now to the specs14:29
ihrachysone spec is to decouple 'db api and utils'14:30
ihrachysthere are simple cases there, miscelaneous util helpers. they just go as is into neutron-lib.14:31
ihrachysbut there are cases where plugins just import 'mixin' classes with db implementation of some API endpoints verbatim14:31
ihrachyslike PortBindingMixin14:32
ihrachysthose bits are really not meant to be shared (as-is)14:33
ihrachysso the spec discusses what to move and how to approach it.14:33
ihrachysand the other spec is about OVO / models14:34
ihrachysone pattern that a lot of subprojects will need is to e.g. add a port, or fetch a network in their plugin code14:34
ihrachyscurrently it means f.e. either sqlalchemy query (that refers/imports models from neutron) or OVO get_object[s] (which also requires import of OVO definition)14:35
ihrachysboden suggests that we introduce a wrapper that would allow projects to do e.g.: util.get_ovo_class('Port') and get the right class.14:36
ihrachysthen they just know that all OVO objects have get_objects14:36
ihrachysso it just calls it14:36
ihrachysand at no point there is a direct import to neutron OVO class definition14:37
lujinluoyeah, this sounds reasonable14:37
ihrachysand then neutron (and subprojects) will have a util helper to register OVO classes for that14:37
ihrachysplugins can then even do things like: sfc_cls = util.get_ovo_class('SFCResource'); if sfc_cls: do some additional work specific to sfc14:39
ihrachysok that's the general background behind the goal there14:40
lujinluook, i see. thanks for the explanation!14:40
lujinluohis design sounds straightforward14:40
ihrachysok great. if you will have time, please try to join the sessions since they will probably touch on OVO and facade one way or another14:42
lujinluosure, i will be there14:43
ihrachys#topic OVO patches14:43
ihrachyshttps://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/adopt-oslo-versioned-objects-for-db14:43
*** openstack changes topic to "OVO patches (Meeting topic: neutron_upgrades)"14:43
ihrachyshttps://review.openstack.org/#/c/507772/ "Use Network OVO in db_base_plugin"14:43
ihrachysTuanVu_, I saw your email but didn't have time to dive in / reply14:44
TuanVu_no problem, Ihar14:44
ihrachysapart from what's in there, do you have more problems with the patch?14:44
TuanVu_I really appreciate your great help so far14:44
TuanVu_thanks, that's the only blocking point at this moment14:44
ihrachysok14:45
*** yamamoto has joined #openstack-meeting-514:45
ihrachyshttps://review.openstack.org/#/c/537320/ "Use Port OVO in neutron/db/external_net_db.py"14:45
ihrachyslujinluo, are results already with the new_facade engine fix in?14:46
lujinluothis one hit by the bug i mentioned before14:46
lujinluocause i need to switch new_facade to true14:46
ihrachysah I see14:46
ihrachysis it the only place where we use Port object?14:47
lujinluowe had other places using Port object before this one, but they do not conflict with new engine (as far as i checked)14:48
ihrachysyeah but if you switch the object to new facade, it's for those other cases too no?14:48
lujinluooh, right. i need to switch those places to new facade14:49
ihrachysright. that may be a problem.14:49
lujinluoi will take a note about that somewhere..14:49
ihrachysone crazy thing I was thinking about though is, what if we could detect which engine is used and use the right facade?14:50
ihrachysso that when ovo call happens under writer.using, it does one thing, but another if under begin(...)14:50
lujinluothis would be great!14:50
ihrachysthen you could have your Port object behave correctly in either context and not care about calling context14:50
ihrachysright. now the question is, how to make it happen lol14:51
*** yamamoto has quit IRC14:51
lujinluolol14:51
ihrachysit may be the case that writer.using / begin leave some marker in session object that we could (ab)use for this14:51
*** lamt_ has joined #openstack-meeting-514:52
lujinluowe could check whether they do14:52
ihrachysand if not, we could introduce our own facade entry points that do, and switch all code to using them14:52
lujinluoi will take a note about this somewhere too..14:52
ihrachysok. that may be part of solution for facade problem14:54
ihrachysat least for ovo14:54
lujinluolol14:54
ihrachyswe would still need to switch to new one in all places14:54
ihrachysbut not as synchronized and intertwined as it would be right now14:54
ihrachysand if we have it, we could also just drop new_facade attribute since it would be of no use14:55
lujinluoyes!14:55
ihrachysdreams dreams dreams14:55
ihrachysbut i don't think it's impossible. we can do it!14:55
ihrachys:)14:55
lujinluoendless work!14:55
ihrachyshttps://review.openstack.org/#/c/544206/ "Integration of (Distributed) Port Binding OVO"14:56
ihrachysis it the patch we reverted a while ago? or was it a different one?14:56
lujinluoit is the one we reverted14:56
lujinluonow it is hit by the same bug14:57
lujinluocause i need to switch portbinding to new engine facade..14:57
ihrachysok, I see. we need a generic solution here, switching it across the board is hard and unsafe14:57
lujinluoyes14:57
ihrachysalso, InvalidRequestError handling in _detach_db_obj is probably not needed with the new_engine patch since it added a wrapped subtransaction there14:58
lujinluoi am not fully sure, but in test_base iface class is not working correctly (the one we want to refactor). i think i added invalid* to mitigate that. after the refactor, invalid* would not be necessary14:59
ihrachysok we need to have a look at specific cases. but just adding that handler is probably unsafe.15:00
lujinluohttps://github.com/openstack/neutron/blob/master/neutron/tests/unit/objects/test_base.py#L685 this class i mean15:00
lujinluotrue. will remove in next PS15:00
ihrachyslujinluo, the class already mocks out expunge / refresh so I am not sure how one would get an error from there15:01
ihrachyssee lines 695-69615:01
ihrachysok let's revert this part. and think about generic solution to auto-detect engine facade version15:02
lujinluoin my local test, f.e. https://github.com/openstack/neutron/blob/master/neutron/tests/unit/objects/test_base.py#L1007 is still calling expunge()15:02
ihrachyswe are out of time15:02
lujinluosure15:02
ihrachysthanks for joining15:02
ihrachys#endmeeting15:02
*** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings/"15:02
openstackMeeting ended Thu Feb 22 15:02:47 2018 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:02
openstackMinutes:        http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-22-14.10.html15:02
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-22-14.10.txt15:02
openstackLog:            http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-22-14.10.log.html15:02
TuanVu_thank you very much, Ihar15:03
lujinluothanks!15:03
TuanVu_see you later, Luo15:03
ihrachyso/15:03
lujinluosee you15:03
TuanVu_:)15:03
*** annp has quit IRC15:05
*** lujinluo has quit IRC15:06
*** felipemonteiro has joined #openstack-meeting-515:25
*** slaweq has quit IRC15:30
*** slaweq has joined #openstack-meeting-515:30
*** slaweq has quit IRC15:35
*** krtaylor has quit IRC15:38
*** yamahata has joined #openstack-meeting-515:40
*** zhipeng has quit IRC15:42
*** yamamoto has joined #openstack-meeting-515:47
*** yamamoto has quit IRC15:51
*** TuanVu_ has quit IRC15:54
*** slaweq has joined #openstack-meeting-516:40
*** slaweq has quit IRC16:45
*** yamamoto has joined #openstack-meeting-516:48
*** mab__ has quit IRC16:50
*** yamamoto has quit IRC16:52
*** ihrachys has quit IRC16:56
*** ricolin_ has quit IRC17:17
*** slaweq has joined #openstack-meeting-517:29
*** diman has quit IRC17:30
*** diman has joined #openstack-meeting-517:31
*** diman has quit IRC17:35
*** slaweq has quit IRC17:48
*** yamamoto has joined #openstack-meeting-517:48
*** yamamoto has quit IRC17:54
*** derekh has quit IRC18:00
*** lamt__ has joined #openstack-meeting-518:03
*** lamt_ has quit IRC18:03
*** lamt__ is now known as Guest8721018:03
*** slaweq has joined #openstack-meeting-518:15
*** sshank has joined #openstack-meeting-518:21
*** yamahata has quit IRC18:22
*** diman has joined #openstack-meeting-518:24
*** slaweq has quit IRC18:25
*** diman has quit IRC18:29
*** lamt has quit IRC18:30
*** lamt has joined #openstack-meeting-518:33
*** lamt is now known as Guest495218:33
*** Guest4952 has quit IRC18:35
*** Guest4952 has joined #openstack-meeting-518:35
*** Guest4952 has quit IRC18:37
*** lamt_ has joined #openstack-meeting-518:37
*** slaweq has joined #openstack-meeting-518:39
*** Guest87210 has quit IRC18:39
*** lamt_ has quit IRC18:40
*** lamt_ has joined #openstack-meeting-518:41
*** slaweq has quit IRC18:43
*** yamamoto has joined #openstack-meeting-518:50
*** yamamoto has quit IRC18:56
*** yamamoto has joined #openstack-meeting-518:59
*** slaweq has joined #openstack-meeting-519:00
*** yamamoto has quit IRC19:04
*** slaweq has quit IRC19:05
*** yamamoto has joined #openstack-meeting-519:14
*** yamamoto has quit IRC19:19
*** diman has joined #openstack-meeting-519:28
*** yamamoto has joined #openstack-meeting-519:29
*** sshank has quit IRC19:33
*** diman has quit IRC19:33
*** yamamoto has quit IRC19:34
*** yamamoto has joined #openstack-meeting-519:44
*** yamamoto has quit IRC19:52
*** slaweq has joined #openstack-meeting-520:01
*** yamahata has joined #openstack-meeting-520:03
*** slaweq_ has joined #openstack-meeting-520:03
*** slaweq has quit IRC20:06
*** slaweq_ has quit IRC20:07
*** lamt_ has quit IRC20:20
*** slaweq has joined #openstack-meeting-520:22
*** slaweq_ has joined #openstack-meeting-520:24
*** slaweq_ has quit IRC20:29
*** diman has joined #openstack-meeting-520:30
*** diman has quit IRC20:34
*** yamamoto has joined #openstack-meeting-520:48
*** yamamoto has quit IRC20:55
*** slaweq_ has joined #openstack-meeting-521:09
*** slaweq_ has quit IRC21:14
*** iyamahat has joined #openstack-meeting-521:22
*** diman has joined #openstack-meeting-521:26
*** diman has quit IRC21:30
*** iyamahat_ has joined #openstack-meeting-521:40
*** sshank has joined #openstack-meeting-521:41
*** iyamahat has quit IRC21:41
*** slaweq_ has joined #openstack-meeting-521:46
*** slaweq_ has quit IRC21:50
*** yamamoto has joined #openstack-meeting-521:53
*** yamamoto has quit IRC21:57
*** yamamoto has joined #openstack-meeting-522:01
*** slaweq_ has joined #openstack-meeting-522:18
*** slaweq has quit IRC22:22
*** slaweq has joined #openstack-meeting-522:22
*** slaweq_ has quit IRC22:23
*** diman has joined #openstack-meeting-522:26
*** slaweq has quit IRC22:27
*** mjturek has quit IRC22:28
*** diman has quit IRC22:30
*** felipemonteiro has quit IRC22:31
*** diman has joined #openstack-meeting-522:39
*** diman has quit IRC22:44
*** yamamoto has quit IRC23:02
*** slaweq has joined #openstack-meeting-523:33
*** slaweq has quit IRC23:39
*** slaweq has joined #openstack-meeting-523:43
*** slaweq has quit IRC23:48

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!