| *** MarkBaker has joined #openstack-meeting-5 | 00:49 | |
| *** MarkBaker has quit IRC | 01:01 | |
| *** mjturek has quit IRC | 01:11 | |
| *** ricolin has joined #openstack-meeting-5 | 01:26 | |
| *** kei-ichi has quit IRC | 02:45 | |
| *** kei-ichi has joined #openstack-meeting-5 | 02:46 | |
| *** kei-ichi has quit IRC | 03:08 | |
| *** kei-ichi has joined #openstack-meeting-5 | 03:09 | |
| *** yamamoto has joined #openstack-meeting-5 | 03:37 | |
| *** kei-ichi has quit IRC | 04:54 | |
| *** kei-ichi has joined #openstack-meeting-5 | 04:55 | |
| *** Mutter_ has joined #openstack-meeting-5 | 06:37 | |
| *** Mutter_ has quit IRC | 06:41 | |
| *** kei-ichi has quit IRC | 06:44 | |
| *** kei-ichi has joined #openstack-meeting-5 | 06:45 | |
| *** kei-ichi has quit IRC | 06:52 | |
| *** markvoelker has quit IRC | 06:52 | |
| *** kei-ichi has joined #openstack-meeting-5 | 06:53 | |
| *** markvoelker has joined #openstack-meeting-5 | 06:53 | |
| *** markvoelker has quit IRC | 06:57 | |
| *** kei-ichi has quit IRC | 06:59 | |
| *** kei-ichi has joined #openstack-meeting-5 | 07:02 | |
| *** markvoelker has joined #openstack-meeting-5 | 08:54 | |
| *** kei-ichi has quit IRC | 09:18 | |
| *** kei-ichi has joined #openstack-meeting-5 | 09:20 | |
| *** markvoelker has quit IRC | 09:27 | |
| *** derekh has joined #openstack-meeting-5 | 09:29 | |
| *** kei-ichi has quit IRC | 09:44 | |
| *** kei-ichi has joined #openstack-meeting-5 | 09:46 | |
| *** yamamoto has quit IRC | 10:04 | |
| *** markvoelker has joined #openstack-meeting-5 | 10:25 | |
| *** markvoelker has quit IRC | 10:58 | |
| *** vkmc has quit IRC | 10:58 | |
| *** yamamoto has joined #openstack-meeting-5 | 11:05 | |
| *** vkmc has joined #openstack-meeting-5 | 11:05 | |
| *** yamamoto has quit IRC | 11:18 | |
| *** yamamoto has joined #openstack-meeting-5 | 11:31 | |
| *** markvoelker has joined #openstack-meeting-5 | 11:55 | |
| *** markvoelker has quit IRC | 12:29 | |
| *** yamamoto has quit IRC | 12:31 | |
| *** yamamoto has joined #openstack-meeting-5 | 12:31 | |
| *** yamamoto has quit IRC | 12:52 | |
| *** makowals has quit IRC | 12:54 | |
| *** makowals has joined #openstack-meeting-5 | 12:56 | |
| *** yamamoto has joined #openstack-meeting-5 | 13:07 | |
| *** makowals has quit IRC | 13:19 | |
| *** annp has joined #openstack-meeting-5 | 13:25 | |
| *** markvoelker has joined #openstack-meeting-5 | 13:26 | |
| *** makowals has joined #openstack-meeting-5 | 13:26 | |
| *** yamamoto has quit IRC | 13:42 | |
| *** cleong has joined #openstack-meeting-5 | 13:42 | |
| *** slaweq has joined #openstack-meeting-5 | 13:51 | |
| *** Mutter has joined #openstack-meeting-5 | 13:55 | |
| *** Mutter is now known as tuanvu | 13:55 | |
| *** tuanvu is now known as tuan_vu | 13:56 | |
| *** ricolin has quit IRC | 13:56 | |
| *** markvoelker has quit IRC | 13:59 | |
| *** lujinluo has joined #openstack-meeting-5 | 13:59 | |
| *** ihrachys_ has joined #openstack-meeting-5 | 14:01 | |
| *** hungpv_ has joined #openstack-meeting-5 | 14:02 | |
| ihrachys_ | #startmeeting neutron_upgrades | 14:02 |
|---|---|---|
| openstack | Meeting started Thu Jan 25 14:02:11 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 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 14:02 |
| *** openstack changes topic to " (Meeting topic: neutron_upgrades)" | 14:02 | |
| openstack | The meeting name has been set to 'neutron_upgrades' | 14:02 |
| ihrachys_ | hi! | 14:02 |
| lujinluo | o/ | 14:02 |
| tuan_vu | Hi everyone | 14:02 |
| *** yamamoto has joined #openstack-meeting-5 | 14:02 | |
| ihrachys_ | before we dive in, some background on the current state of affairs in relation to release | 14:03 |
| ihrachys_ | as you may probably know, a new major release (queens) is getting closer | 14:03 |
| lujinluo | \o/ | 14:03 |
| ihrachys_ | https://releases.openstack.org/queens/schedule.html | 14:04 |
| ihrachys_ | this week is feature freeze | 14:04 |
| ihrachys_ | we'll have a RC1 release (and hence stable/queens) in several weeks from now | 14:05 |
| ihrachys_ | till that moment, we should be cautious landing patches that may affect stability and don't fix clear bugs | 14:05 |
| ihrachys_ | which most OVO patches are | 14:05 |
| ihrachys_ | we can of course continue reshaping patches in review to get them ready for when master is open | 14:06 |
| lujinluo | understood. | 14:06 |
| ihrachys_ | one sad development because of release final coming close is that we were forced to revert port binding patch that took a while to shape | 14:06 |
| ihrachys_ | because of several issues that merging it revealed | 14:07 |
| ihrachys_ | one is that it was not counting with mixed old/new engine facade usage | 14:07 |
| ihrachys_ | another is postgresql installations busted | 14:07 |
| lujinluo | yeah, speaking of that. can i have a (stupid) question? | 14:07 |
| ihrachys_ | we will need to get back to the white board, so to speak, repropose the patch, fix those issues, and land some time in Rocky | 14:08 |
| ihrachys_ | lujinluo, yes! | 14:08 |
| lujinluo | i was trying to reproduce the postgresql issues | 14:08 |
| lujinluo | so i built a devstack with postgresql as backend | 14:08 |
| lujinluo | then i have no idea how i can reproduce the errors.. | 14:08 |
| lujinluo | could you direct me to the conf/settings of that specific tempest tests? | 14:09 |
| ihrachys_ | the periodic job that fails executes full suite of tempest I believe | 14:09 |
| ihrachys_ | but let me check | 14:09 |
| lujinluo | sure, thanks | 14:10 |
| ihrachys_ | so this is what failed there: http://logs.openstack.org/periodic/git.openstack.org/openstack/neutron/master/legacy-periodic-tempest-dsvm-neutron-pg-full/71f9bd8/logs/testr_results.html.gz | 14:11 |
| *** lamt_ has joined #openstack-meeting-5 | 14:12 | |
| lujinluo | got it! will start from there | 14:12 |
| ihrachys_ | I would imagine running those tests in a loop could get you to failure | 14:12 |
| ihrachys_ | specifically check those that raise ServerFault | 14:12 |
| lujinluo | nice suggestion. will follow it ;) | 14:12 |
| ihrachys_ | since that's a sign of neutron-server internal error | 14:13 |
| ihrachys_ | ok | 14:13 |
| ihrachys_ | btw note that revert patch hasn't landed still: https://review.openstack.org/536913 | 14:13 |
| lujinluo | ok | 14:14 |
| ihrachys_ | in other news, gate is quite unstable lately, with multiple issues lingering | 14:14 |
| ihrachys_ | we are trying to fix them one by one but we are not there yet | 14:14 |
| ihrachys_ | things like job timeouts is one, functional is also unstable (ovsdb commands timing out) | 14:14 |
| ihrachys_ | so please be patient if you get those errors over and over, that's sadly expected | 14:15 |
| hungpv_ | Okay :( | 14:15 |
| ihrachys_ | now, back to usual business | 14:16 |
| slaweq | hello, sorry for late | 14:16 |
| ihrachys_ | slaweq, hi! | 14:16 |
| lujinluo | hey slaweq ! | 14:16 |
| ihrachys_ | https://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/adopt-oslo-versioned-objects-for-db | 14:16 |
| ihrachys_ | https://review.openstack.org/#/c/521797/ "Use Router OVO in external_net_db" | 14:16 |
| ihrachys_ | so this patch is fine now but it lacks test coverage for the module / method it touches | 14:17 |
| ihrachys_ | so we can't really safely land it | 14:17 |
| ihrachys_ | hungpv_, do you think you could work on adding some unit tests for the method before merging this OVO patch? | 14:18 |
| hungpv_ | I thought you said it's called by update_rbac_policy and delete_rbac_policy ? | 14:18 |
| ihrachys_ | I just said that I checked there is no mixed old/new engine facade issue there | 14:19 |
| hungpv_ | Oh, if it's the case | 14:20 |
| ihrachys_ | remember in the past, in one of older patchsets, there were some mistakes that absolutely broke the code and gate failed to catch it? that's what I am concerned about, that existing coverage won't catch anything if there are still some bugs there. | 14:20 |
| hungpv_ | I'll be working more this issue | 14:20 |
| ihrachys_ | so adding some unit tests would help us to sleep well :) | 14:21 |
| ihrachys_ | ok thanks hungpv_ ! | 14:21 |
| ihrachys_ | next "https://review.openstack.org/507772 Use Network OVO in db_base_plugin" | 14:21 |
| tuan_vu | I’ve left a comment about current blocking point | 14:21 |
| *** ricolin has joined #openstack-meeting-5 | 14:21 | |
| tuan_vu | We’re still working on it | 14:22 |
| ihrachys_ | the current list of failures is: http://logs.openstack.org/72/507772/29/check/openstack-tox-py27/8b97a84/testr_results.html.gz | 14:22 |
| ihrachys_ | for foreign key failure, I see it's just in object test class itself | 14:23 |
| tuan_vu | Yes, but we haven’t figured out how to fix it | 14:24 |
| slaweq | tuan_vu: I didn't have time to check it yet | 14:24 |
| slaweq | but I will try to help | 14:24 |
| tuan_vu | Hi slaweq, thank you in advance | 14:24 |
| ihrachys_ | I think it's one of those cases that just need self.update_obj_fields in setUp() for whatever field is violated | 14:24 |
| tuan_vu | We really appreciate your help | 14:24 |
| ihrachys_ | slaweq, tuan_vu, ping me if you need me to look at the foreign key issue | 14:25 |
| tuan_vu | Thank you, ihar, we’ll try your suggestion | 14:25 |
| ihrachys_ | another more important failure there is test_*_queries_constant failures | 14:26 |
| tuan_vu | Yes, I’m just wondering if we can change the number of queries? | 14:26 |
| tuan_vu | Because in ovo, the way it works is different a little bit | 14:27 |
| ihrachys_ | which signals that somewhere in the patch, we fetch some models from database in an ineffective way (probably fetching a relationship for networks using a separate query) | 14:27 |
| ihrachys_ | tuan_vu, we can't, we should find a way to avoid it | 14:27 |
| ihrachys_ | tuan_vu, do you know which relationship triggers additional queries? | 14:27 |
| tuan_vu | We need to reload “shared” attribute | 14:28 |
| tuan_vu | When update network | 14:28 |
| ihrachys_ | why is it? can't we calculate new value from what is already passed by user / fetched from db? | 14:29 |
| lujinluo | is it because "shared" is a synthetic field now? | 14:30 |
| tuan_vu | Hi Luo, yes | 14:30 |
| ihrachys_ | besides, test_network_list_queries_constant also fails, which doesn't update network | 14:31 |
| lujinluo | i think i have met this issue while working on Port integration. when we have aynthetic fields, the # of queries does not stay constant | 14:31 |
| ihrachys_ | it lists networks, and probably we trigger a set of queries per network matched | 14:31 |
| lujinluo | yes, because get_objects() will try to fetch all their synthetic fields too | 14:32 |
| tuan_vu | Yes, thank you Luo, your experience does help a lot | 14:33 |
| lujinluo | but i do not recall i found a suitable solution for that :( | 14:33 |
| ihrachys_ | right. so one of relationships is probably of type that doesn't fetch data when parent network model is fetched. that would be a problem, no? | 14:33 |
| ihrachys_ | I mean, type of sqlalchemy model attribute for relationship | 14:34 |
| *** markvoelker has joined #openstack-meeting-5 | 14:34 | |
| ihrachys_ | ok, let's try from another direction. how did it work before the patch? | 14:34 |
| tuan_vu | Hmm | 14:35 |
| tuan_vu | The main problem is “shared” attribute | 14:35 |
| tuan_vu | Before the patch, we use rbac for “shared” | 14:36 |
| ihrachys_ | right. can't we calculate it for OVO the same way? we have .db_obj.rbac_entries already, right? | 14:36 |
| ihrachys_ | if we then make rbac metaclass to use this info directly instead of refetching it (I guess that's what happens?) then we would avoid the problem? | 14:37 |
| tuan_vu | Hmm, let me think more about your suggestion, is that ok? I’ll send you an email if that’s needed | 14:38 |
| tuan_vu | Or contact you via IRC | 14:39 |
| ihrachys_ | yeah sure, we don't need to dig the code right away, but the main point, we can't disable / change this unit test. | 14:39 |
| lujinluo | that should work, by changing how the # of queries are calculated. besides, "shared" is not a synthetic field, please check which attribute is the one that causes this issue | 14:39 |
| tuan_vu | Thank you, Ihar and Luo | 14:39 |
| ihrachys_ | moving on | 14:39 |
| ihrachys_ | https://review.openstack.org/#/c/537320/ "Use Port OVO in neutron/db/external_net_db.py" | 14:40 |
| lujinluo | i did recheck about the timeouts.. but i guess zuul hung somewhere...? | 14:40 |
| ihrachys_ | lujinluo, I was to look at some port patches you have but then we stumbled on port binding revert. I was thinking, are we blocked by that one now? | 14:40 |
| ihrachys_ | lujinluo, yeah, zuul is sick these days. I rechecked. | 14:40 |
| lujinluo | kind of.. it's better to wait for port binding to land first, as those two patches conflicts a lot | 14:41 |
| ihrachys_ | lujinluo, right. speaking of which... do you go to Dublin? I think mlavalle was planning a session on ovo/enginefacade issue. | 14:42 |
| lujinluo | it would be easier to get them in serially, not in parallel | 14:42 |
| *** hoangcx_ has joined #openstack-meeting-5 | 14:42 | |
| lujinluo | my employer has not decided yet, but i guess probably.. | 14:42 |
| lujinluo | i would go | 14:42 |
| ihrachys_ | great. btw I don't. but having someone from this group there would help. I may try to join remotely if there will be such an option | 14:43 |
| lujinluo | oh, it is sad that you are not going :( | 14:43 |
| ihrachys_ | yeah. I kinda lagged on trips lately for family reasons. | 14:44 |
| lujinluo | well, family always goes first | 14:45 |
| ihrachys_ | anyway, moving on to the next which is | 14:45 |
| ihrachys_ | https://review.openstack.org/#/c/537325/ "Use Meter Label OVO in neutron/db/metering/metering_db.py" | 14:45 |
| ihrachys_ | this looks green :) | 14:45 |
| lujinluo | i used label.db_obj (which we should not do normally), but since we cannot use router as a synthetic field, i think this is fine. what do you think? ihrachys_ | 14:46 |
| *** tuan_vu has quit IRC | 14:46 | |
| ihrachys_ | so there, you pass db models in one case and objects in another case | 14:46 |
| lujinluo | hmm right | 14:47 |
| lujinluo | i need to avoid that | 14:47 |
| ihrachys_ | question is, how | 14:48 |
| ihrachys_ | we could of course add routers field to labels but that's a bit silly from API perspective (label is not a container for routers) | 14:48 |
| lujinluo | the first solution came to my mind was _load_object(), but we should avoid that too | 14:49 |
| lujinluo | yeah | 14:49 |
| ihrachys_ | at least for get_sync_data_for_rule, you could avoid fetching labels at all and instead have a method that returns the needed routers. | 14:49 |
| lujinluo | how about using get_objects() | 14:50 |
| ihrachys_ | since you don't use anything but id from the label, and it's already known | 14:50 |
| ihrachys_ | we could have this method embedded into Router.get_objects, so that if metering_label_id filter is passed, it does the right tihng | 14:50 |
| *** tuan_vu has joined #openstack-meeting-5 | 14:51 | |
| lujinluo | that sounds good. will choose that approach | 14:51 |
| ihrachys_ | ah wait not really. you rely on shared there | 14:52 |
| lujinluo | yes, but instead i can pass the router_ids i fetch from label.db_obj.routers, then use get_objects() | 14:53 |
| ihrachys_ | true but wouldn't it trigger another fetch? | 14:54 |
| ihrachys_ | but wait, we already trigger it for shared case... | 14:54 |
| lujinluo | yes | 14:54 |
| ihrachys_ | maybe that's a mistake actually that slipped in | 14:54 |
| ihrachys_ | it was doing get_collection_query before OVO though | 14:55 |
| lujinluo | it got in here https://review.openstack.org/#/c/529551/5/neutron/db/metering/metering_db.py | 14:55 |
| ihrachys_ | so I guess no one bothered about it being optimial | 14:55 |
| lujinluo | anyway, let's dig more in gerrit cause we only have 5 min left now! | 14:57 |
| ihrachys_ | but actually, for that other case with all routers fetched, it's different | 14:57 |
| ihrachys_ | because then label.db_obj.routers won't give same result | 14:57 |
| ihrachys_ | ok let's follow up there | 14:57 |
| ihrachys_ | slaweq, I am sorry; you joined today. did you have smth? | 14:58 |
| slaweq | ihrachys_: no | 14:58 |
| slaweq | I just wanted to join :) | 14:58 |
| ihrachys_ | ok, just checking :) thanks for joining :) | 14:58 |
| lujinluo | slaweq: we are happy that you are here! | 14:59 |
| slaweq | I will try to be here every week | 14:59 |
| slaweq | if I will not forget :) | 14:59 |
| ihrachys_ | we don't really have much time left. if you have questions about some patches that were not discussed, please ping me in irc or email after this meeting, I will try to help. | 14:59 |
| tuan_vu | Awesome, Slawek | 14:59 |
| ihrachys_ | slaweq, great. your help would be of great value. | 15:00 |
| ihrachys_ | ok we are officially out of time | 15:00 |
| ihrachys_ | ciao | 15:00 |
| ihrachys_ | #endmeeting | 15:00 |
| *** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings/" | 15:00 | |
| openstack | Meeting ended Thu Jan 25 15:00:18 2018 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:00 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-01-25-14.02.html | 15:00 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-01-25-14.02.txt | 15:00 |
| tuan_vu | See you guys later | 15:00 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/neutron_upgrades/2018/neutron_upgrades.2018-01-25-14.02.log.html | 15:00 |
| slaweq | bye | 15:00 |
| tuan_vu | :) | 15:00 |
| lujinluo | see you guys | 15:00 |
| *** lujinluo has quit IRC | 15:01 | |
| *** tuan_vu has quit IRC | 15:03 | |
| *** ihrachys_ is now known as ihrachys | 15:08 | |
| *** hoangcx_ has quit IRC | 15:09 | |
| *** annp has quit IRC | 15:22 | |
| *** slaweq has quit IRC | 15:57 | |
| *** slaweq has joined #openstack-meeting-5 | 15:58 | |
| *** slaweq has quit IRC | 16:02 | |
| *** yamahata has quit IRC | 16:11 | |
| *** ricolin has quit IRC | 16:45 | |
| *** slaweq has joined #openstack-meeting-5 | 16:49 | |
| *** slaweq has quit IRC | 16:53 | |
| *** hungpv_ has quit IRC | 17:19 | |
| *** mjturek has joined #openstack-meeting-5 | 17:20 | |
| *** lamt_ has quit IRC | 17:32 | |
| *** yamamoto has quit IRC | 17:40 | |
| *** yamamoto has joined #openstack-meeting-5 | 17:43 | |
| *** yamamoto has quit IRC | 17:44 | |
| *** yamamoto has joined #openstack-meeting-5 | 17:44 | |
| *** derekh has quit IRC | 17:54 | |
| *** yamamoto has quit IRC | 18:42 | |
| *** yamahata has joined #openstack-meeting-5 | 18:56 | |
| *** yamahata has quit IRC | 19:09 | |
| *** yamamoto has joined #openstack-meeting-5 | 19:43 | |
| *** yamamoto has quit IRC | 19:55 | |
| *** cleong has quit IRC | 21:32 | |
| *** yamahata has joined #openstack-meeting-5 | 22:02 | |
| *** yamamoto has joined #openstack-meeting-5 | 22:05 | |
| *** slaweq_ has joined #openstack-meeting-5 | 22:13 | |
| *** slaweq_ has quit IRC | 22:15 | |
| *** slaweq_ has joined #openstack-meeting-5 | 22:16 | |
| *** slaweq_ has quit IRC | 22:18 | |
| *** slaweq_ has joined #openstack-meeting-5 | 22:19 | |
| *** slaweq_ has quit IRC | 22:23 | |
| *** yamahata has quit IRC | 23:08 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!