Thursday, 2018-02-08

*** markvoelker has joined #openstack-meeting-500:17
*** markvoelker has quit IRC00:24
*** ricolin has joined #openstack-meeting-501:20
*** mjturek has joined #openstack-meeting-501:31
*** iyamahat has quit IRC02:18
*** yamahata has quit IRC02:20
*** yamamoto has joined #openstack-meeting-503:38
*** slaweq has joined #openstack-meeting-504:34
*** slaweq has quit IRC04:39
*** slaweq has joined #openstack-meeting-505:34
*** slaweq has quit IRC05:39
*** yamahata has joined #openstack-meeting-505:50
*** mjturek has quit IRC06:01
*** yamamoto_ has joined #openstack-meeting-506:26
*** yamamoto has quit IRC06:30
*** slaweq has joined #openstack-meeting-506:34
*** slaweq has quit IRC06:39
*** slaweq has joined #openstack-meeting-506:55
*** slaweq has quit IRC06:59
*** slaweq has joined #openstack-meeting-507:55
*** slaweq has quit IRC08:00
*** slaweq has joined #openstack-meeting-508:13
*** iyamahat has joined #openstack-meeting-508:16
*** jbryce has quit IRC08:17
*** iyamahat_ has joined #openstack-meeting-508:19
*** iyamahat__ has joined #openstack-meeting-508:20
*** iyamahat has quit IRC08:21
*** iyamahat_ has quit IRC08:23
*** yamamoto has joined #openstack-meeting-508:57
*** yamamoto_ has quit IRC09:00
*** derekh has joined #openstack-meeting-509:10
*** MarkBaker has quit IRC09:24
*** iyamahat__ has quit IRC09:25
*** slaweq_ has joined #openstack-meeting-509:25
*** slaweq_ has quit IRC09:30
*** MarkBaker has joined #openstack-meeting-510:00
*** jbryce has joined #openstack-meeting-510:00
*** iyamahat has joined #openstack-meeting-510:23
*** iyamahat has quit IRC10:50
*** mattmceuen has quit IRC11:23
*** powerds0111 has quit IRC11:23
*** yamahata has quit IRC11:27
*** mjturek has joined #openstack-meeting-512:08
*** makowals has joined #openstack-meeting-512:12
*** mattmceuen has joined #openstack-meeting-513:19
*** powerds0111 has joined #openstack-meeting-513:19
*** MarkBaker has quit IRC13:41
*** TuanVu has joined #openstack-meeting-513:54
*** MarkBaker has joined #openstack-meeting-513:55
*** lujinluo has joined #openstack-meeting-513:58
*** ihrachys has joined #openstack-meeting-514:02
ihrachys#startmeeting neutron_upgrades14:02
openstackMeeting started Thu Feb  8 14:02:23 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
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:02
*** openstack changes topic to " (Meeting topic: neutron_upgrades)"14:02
openstackThe meeting name has been set to 'neutron_upgrades'14:02
*** hungpv has joined #openstack-meeting-514:02
lujinluoo/14:02
ihrachyso/14:02
TuanVuHi Ihar14:02
TuanVuHi Luo14:02
hungpvHi guys14:03
slaweqhi14:03
ihrachysPSA: this week is RC1 (some time Friday) which should afaiu create stable/queens for neutron14:03
TuanVuHi Slawek14:03
ihrachysafter which we should be able to cautiously start merging Pike patches14:04
ihrachyssorry, I mean Rocky patches14:04
ihrachysof course, while final release is not shipped, the focus will be diverged to Queens matters14:04
ihrachysas some of you know, we have an OVO related issue in bgpvpn that is release blocker for them14:05
ihrachyshttps://launchpad.net/bugs/174699614:06
openstackLaunchpad bug 1746996 in neutron "bagpipe: IPAllocation DB query failing, engine facade mismatch" [High,Confirmed]14:06
ihrachysit's a rather twisted issue. but basically, it boils down to bgpvpn using OVO objects in callback handlers that are issued from new engine facade subtransactions, so it fails rather spectacularly because it doesn't see models created by the callback caller in its session14:07
ihrachysand there is no clear cut way out of the issue.14:07
ihrachysone idea we are trying right now is switching bgpvpn objects to new facade, while leaving other objects on the old one for Queens14:08
ihrachysthere is patch for bagpipe here: https://review.openstack.org/#/c/541513/14:08
ihrachysbut it would require neutron change: https://review.openstack.org/#/c/541512/14:08
ihrachysand the latter is not ready for prime time, lots of unit tests were failing14:08
ihrachysI see lujinluo picked it up, thanks for that14:09
lujinluoihrachys: it is still not in shape yet14:09
ihrachysyeah I know14:10
ihrachysso some questions about your changes14:10
ihrachysfirst, what's about subnetpool change to new_facade=True? https://review.openstack.org/#/c/541512/4..5/neutron/objects/subnetpool.py14:11
lujinluoone minute, finding the code snippets14:12
ihrachysthere is also https://review.openstack.org/#/c/541512/4..5/neutron/db/db_base_plugin_v2.py that deals with subnetpool14:13
lujinluosorry, the new_facade=True might be garbage code that i forgot to delete14:14
lujinluoas for the latter, it is because of this change https://review.openstack.org/#/c/541512/5/neutron/objects/base.py@31914:15
lujinluowe need to refresh(db_obj) in case synthetic field changed14:15
lujinluobut if the session is not active, we ignored it then we miss the synthetic field change in db14:16
ihrachysshouldn't session be active at this point?14:16
lujinluosomehow no14:17
lujinluoneutron.tests.unit.db.test_db_base_plugin_v2.TestSubnetPoolsV2.test_allocate_any_subnet_with_prefixlen is the failed UT14:18
lujinluohttp://logs.openstack.org/12/541512/4/check/openstack-tox-py27/15c63a0/testr_results.html.gz first one here14:18
lujinluoWe missed to refresh the synthetic field (prefixes) of subnetpool14:20
ihrachysok I guess prefixes are not reloaded in time. maybe we should put refresh in subnetpool object code, I will have a look14:20
lujinluook. that is basically the only failed UT I managed to mitigate, LOL14:21
ihrachysactually, maybe it's the decorator that should open subtransaction so that it always spans throughout func() AND refresh()14:21
ihrachyswell that's still something. there are other valuable changes there where I missed model to obj_cls conversion14:22
ihrachysthanks for looking into it14:22
ihrachysI will take it back over today14:22
ihrachysI thought tmorin will help with that but he probably was busy14:22
lujinluoyes, although i have not fixed any other UTs, but during the debugging it seems that most of the UTs are about timing to commit changes14:23
ihrachysnow to other, non-blocking patches14:23
ihrachyslujinluo, timing to commit?14:23
lujinluoi mean when to open and close transactions14:23
*** MarkBaker has quit IRC14:24
lujinluoalso, i want to address that in the patch we are changing some autonested sessions to context.session, and i am not sure if that is the right thing to do14:25
ihrachyslujinluo, do you have an example of the latter? it's easier to follow with an example.14:25
lujinluohttps://review.openstack.org/#/c/541512/5/neutron/objects/base.py@59914:26
lujinluohere you go14:26
ihrachysright. what's the problem with that?14:26
lujinluoi think when we use autonested we are addressing the transaction inside transaction issue, but with this patch we lost that14:27
ihrachysyou mean like independent rollback?14:27
lujinluoyes14:27
ihrachysI don't think we use it except in some places where it's explicit in the code (I believe ipam layer had rollback per ip allocation)14:28
ihrachysautonested just helps us to not care about whether caller opened a subtransaction already14:28
ihrachysbegin(subtransactions=True) is afaik the same as autonested_transaction14:29
lujinluook, i see. then maybe we do not need to care about it14:29
ihrachysok, moving to other patches14:30
ihrachyshttps://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/adopt-oslo-versioned-objects-for-db14:30
ihrachyshttps://review.openstack.org/#/c/507772/ "Use Network OVO in db_base_plugin"14:30
TuanVuFor this patch, the good news is: there’s just only 1 remaining failed unit test at this moment, which is: “test_get_objects_queries_constant”14:30
ihrachysthere are 3 failures right now, all because of number of queries not scaling well14:30
TuanVuyes14:31
TuanVuI’m trying to find exactly where the number of queries is increased14:31
*** hungpv has quit IRC14:32
*** zhipeng has joined #openstack-meeting-514:32
ihrachysbased on diff of queries in logs, it seems that's because of externalnetworks table fetch14:32
TuanVuthank you, Ihar14:33
ihrachysalso networksecuritybindings is fetched twice14:33
ihrachysI guess once per network object14:33
TuanVuyes, I agree, but I still haven't found where it is increased14:33
*** yamamoto has quit IRC14:34
ihrachysafaiu ExternalNetwork is a type of one of synthetic fields of Network14:34
*** hungpv has joined #openstack-meeting-514:34
TuanVucorrect14:34
ihrachysin theory, it should be eagerly fetched with network model because its relationship is lazy='joined'14:34
TuanVuyes14:35
ihrachyshm, I am looking at Network object from_db_object: https://review.openstack.org/#/c/507772/36/neutron/objects/network.py@32614:36
ihrachysand I don't see where we would extract values for external from db model14:36
ihrachyshm, though maybe it's base class that is driving synthetic fields extraction, I am already not sure, it was so long time ago I wrote it :)14:37
TuanVuthanks for your suggestion, Ihar :)14:38
ihrachysthis may actually be related to the same thing why you needed to catch InvalidRequestError here: https://review.openstack.org/#/c/507772/36/neutron/objects/base.py@32014:38
TuanVuwe'll check it deeper14:38
ihrachysif you don't refresh the relationship, maybe it refetches it somewhere else once per object returned. I think ultimately catching the error is not correct and we should make sure that the session is there for the time when refresh is issued14:39
ihrachysthat boils down to what we discussed for that other patch adding support for new_engine14:39
*** MarkBaker has joined #openstack-meeting-514:40
ihrachysactually, that's a good patch to try my approach with subtransaction opened in decorator. if it works, it will probably fix the failure.14:40
ihrachysI will try it on your patch later.14:40
TuanVuthanks, we'll try to figure out way to handle it more properly14:40
ihrachysok14:41
ihrachyshttps://review.openstack.org/#/c/521797/ "Use Router OVO in external_net_db"14:41
ihrachysthis has pep8 issue, let me fix it quickly14:42
ihrachysdone14:42
ihrachysapart from that, it seems fine14:42
hungpvThanks14:43
ihrachyshttps://review.openstack.org/#/c/537320/ "Use Port OVO in neutron/db/external_net_db.py"14:43
ihrachysthis patch is largely a variation on that other patch adding support for new_engine14:43
ihrachysso we'll need to rebase it on top of the patch at some point to be able to utilize the new engine feature14:44
*** zhipeng has quit IRC14:44
ihrachysit also has this InvalidRequestError exception handler that we'll probably try to avoid14:44
ihrachysseems like this patch will need to wait a bit for when those underlying issues are handled14:45
lujinluoyes, i will rebase this one on top of https://review.openstack.org/#/c/541512/ later when it is ready14:45
ihrachysright. it's not ready yet so no need to rebase right away14:45
ihrachysok next is https://review.openstack.org/#/c/537325/ "Use Meter Label OVO in neutron/db/metering/metering_db.py"14:45
ihrachysI had a question there https://review.openstack.org/#/c/537325/4/neutron/db/l3_dvr_db.py@107714:46
*** zhipeng has joined #openstack-meeting-514:46
ihrachysnot sure you had a chance to think about it14:46
lujinluono... i will check tmr14:46
*** yamamoto has joined #openstack-meeting-514:46
ihrachysI personally hate those functions that try to support polymorphic argument types. seems convoluted. I wonder if it's too hard to switch all callers of is_distributed_router to OVO14:47
*** hungpv has quit IRC14:47
ihrachysok let's move on14:48
ihrachyshttps://review.openstack.org/#/c/530182/ "Use Router OVO in l3_db"14:49
ihrachysthe author is not responsive14:49
ihrachysif people feel like taking that over, I wouldn't mind :)14:49
ihrachysthere are no more patches in the queue that would be worth looking at at this point14:51
*** yamamoto has quit IRC14:51
ihrachysanyone has a topic to discuss?14:53
lujinluoi have one thing14:53
ihrachyslujinluo, shoot14:53
lujinluoi will not be able to join the weekly meeting next week14:53
lujinluoas it is chinese new year eve14:53
*** mjturek has quit IRC14:53
lujinluoi won't have the mood to work, LOL14:53
ihrachysok no problem. happy new year :)14:53
lujinluothanks!14:54
ihrachysyou are lucky, you have two of them!14:54
lujinluoLOL, yeah14:54
lujinluoi think this also applies to TuanVu ? right?14:54
ihrachysis it the same / similar for others? or is it just in your part of Asia?14:54
lujinluoyou guys have public holiday on that day, no?14:55
TuanVuyes, I think so14:55
ihrachysI see. well then we will probably cancel the next meeting, and meet in 2 weeks14:55
lujinluoyeah, Vietnamese also celebrate lunar new year14:55
ihrachysenjoy the holiday!14:55
lujinluothank you!14:55
TuanVucorrect, the same with An-san and Hung-san14:56
lujinluohappy chinese new year of dog to you too ihrachys and TuanVu14:56
TuanVuthank you, Ihar and Luo :)14:56
ihrachysyay14:56
ihrachys*°*”˜˜”*°•.¸☆ ★ ☆¸.•°*”˜˜”*°•.¸☆14:56
ihrachys╔╗╔╦══╦═╦═╦╗╔╗ ★ ★ ★14:56
ihrachys║╚╝║══║═║═║╚╝║ ☆¸.•°*”˜˜”*°•.¸☆14:56
ihrachys║╔╗║╔╗║╔╣╔╩╗╔╝ ★ NEW YEAR ☆14:56
ihrachys╚╝╚╩╝╚╩╝╚╝═╚╝ ♥¥☆★☆★☆¥♥ ★☆❤♫❤♫❤14:56
ihrachys.•*¨`*•..¸☼ ¸.•*¨`*•.♫❤♫❤♫❤14:56
TuanVuhappy chinese new year to you, too14:56
TuanVu:)14:56
ihrachysok, thanks everyone for joining :)14:56
ihrachys#endmeeting14:56
*** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings/"14:56
openstackMeeting ended Thu Feb  8 14:56:47 2018 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:56
openstackMinutes:        http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-08-14.02.html14:56
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-08-14.02.txt14:56
openstackLog:            http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-02-08-14.02.log.html14:56
TuanVu@Ihar: awesome, haha :))14:56
lujinluoihrachys: amazing, haha14:57
slaweqnice :)14:57
TuanVusee you later, Slawek :)14:57
ihrachysslaweq, no holiday for you my European friend. :p14:57
slaweqihrachys: yes, no holiday :(14:58
slaweqbut fortunately I like what I'm doing :)14:58
TuanVuthat's awesome, slawek :D14:58
slaweqthx TuanVu, see You14:59
TuanVuthank you, Ihar and Luo14:59
TuanVusee you guys soon :)15:00
*** iyamahat has joined #openstack-meeting-515:00
*** yamahata has joined #openstack-meeting-515:02
*** lujinluo has quit IRC15:16
*** mjturek has joined #openstack-meeting-515:18
*** mjturek has quit IRC15:23
*** TuanVu has quit IRC15:31
*** yamamoto has joined #openstack-meeting-515:32
*** yamamoto has quit IRC15:37
*** yamamoto has joined #openstack-meeting-515:44
*** yamamoto has quit IRC15:44
*** felipemonteiro has joined #openstack-meeting-515:45
*** felipemonteiro_ has joined #openstack-meeting-515:46
*** MarkBaker has quit IRC15:48
*** felipemonteiro has quit IRC15:50
*** slaweq has quit IRC15:56
*** slaweq has joined #openstack-meeting-515:56
*** slaweq has quit IRC16:01
*** MarkBaker has joined #openstack-meeting-516:02
*** MarkBaker has quit IRC16:09
*** MarkBaker has joined #openstack-meeting-516:09
*** iyamahat has quit IRC16:22
*** zhipeng has quit IRC16:29
*** iyamahat has joined #openstack-meeting-516:31
*** yamamoto has joined #openstack-meeting-516:44
*** yamamoto has quit IRC17:02
*** slaweq has joined #openstack-meeting-517:04
*** felipemonteiro__ has joined #openstack-meeting-517:08
*** slaweq has quit IRC17:09
*** felipemonteiro_ has quit IRC17:12
*** derekh has quit IRC17:18
*** slaweq has joined #openstack-meeting-518:10
*** slaweq_ has joined #openstack-meeting-518:13
*** mjturek has joined #openstack-meeting-518:14
*** slaweq_ has quit IRC18:19
*** felipemonteiro__ has quit IRC18:34
*** felipemonteiro__ has joined #openstack-meeting-518:34
*** slaweq has quit IRC18:35
*** felipemonteiro__ has quit IRC18:36
*** slaweq has joined #openstack-meeting-518:56
*** slaweq has quit IRC19:01
*** slaweq has joined #openstack-meeting-519:14
*** slaweq has quit IRC19:17
*** cargonza has quit IRC19:25
*** srwilkers has quit IRC19:27
*** cargonza has joined #openstack-meeting-519:28
*** srwilkers has joined #openstack-meeting-519:31
*** slaweq has joined #openstack-meeting-519:43
*** ChanServ has quit IRC20:57
*** iyamahat has quit IRC21:16
*** yamahata has quit IRC21:16
*** ChanServ has joined #openstack-meeting-521:30
*** barjavel.freenode.net sets mode: +o ChanServ21:30
*** yamamoto has joined #openstack-meeting-521:32
*** slaweq has quit IRC21:51
*** ricolin has quit IRC21:57
*** mjturek has quit IRC22:10
*** felipemonteiro has joined #openstack-meeting-522:10
*** felipemonteiro_ has joined #openstack-meeting-522:20
*** felipemonteiro has quit IRC22:24
*** felipemonteiro_ has quit IRC22:55
*** felipemonteiro_ has joined #openstack-meeting-522:56
*** felipemonteiro_ has quit IRC23:42
*** yamahata has joined #openstack-meeting-523:45

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