| *** zzzeek has left #openstack-oslo | 00:02 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow injected atom args to be persisted https://review.openstack.org/163159 | 00:03 |
|---|---|---|
| *** e0ne has quit IRC | 00:06 | |
| openstackgerrit | Ian Cordasco proposed openstack/oslo.policy: Publicize oslo_policy.opts.register https://review.openstack.org/163163 | 00:06 |
| *** tsufiev_ has quit IRC | 00:08 | |
| *** tsufiev_ has joined #openstack-oslo | 00:14 | |
| *** tsufiev_ has quit IRC | 00:17 | |
| *** tsekiyama has quit IRC | 00:18 | |
| *** achanda has quit IRC | 00:18 | |
| *** achanda has joined #openstack-oslo | 00:18 | |
| *** dims_ has joined #openstack-oslo | 00:20 | |
| *** mtanino has quit IRC | 00:21 | |
| *** dims has quit IRC | 00:22 | |
| *** dims_ has quit IRC | 00:22 | |
| *** dims has joined #openstack-oslo | 00:23 | |
| *** achanda has quit IRC | 00:23 | |
| openstackgerrit | Merged openstack/tooz: Add a driver feature compatibility matrix https://review.openstack.org/162742 | 00:24 |
| *** tsufiev_ has joined #openstack-oslo | 00:25 | |
| *** david-lyle_afk has quit IRC | 00:34 | |
| *** zzzeek has joined #openstack-oslo | 00:58 | |
| *** zzzeek has quit IRC | 00:59 | |
| *** zzzeek has joined #openstack-oslo | 01:00 | |
| *** pblaho has quit IRC | 01:05 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add failure json schema validation (WIP) https://review.openstack.org/163248 | 01:06 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add failure json schema validation (WIP) https://review.openstack.org/163248 | 01:08 |
| *** ChuckC has quit IRC | 01:10 | |
| *** liusheng has quit IRC | 01:12 | |
| *** liusheng has joined #openstack-oslo | 01:12 | |
| *** crc32 has joined #openstack-oslo | 01:20 | |
| *** crc32 has quit IRC | 01:20 | |
| *** david-lyle_afk has joined #openstack-oslo | 01:21 | |
| *** crc32 has joined #openstack-oslo | 01:22 | |
| *** dims has quit IRC | 01:28 | |
| *** dims has joined #openstack-oslo | 01:29 | |
| *** salv-orlando has quit IRC | 01:30 | |
| *** devlaps has joined #openstack-oslo | 01:31 | |
| *** dims has quit IRC | 01:34 | |
| *** dims has joined #openstack-oslo | 01:40 | |
| *** ChuckC has joined #openstack-oslo | 01:42 | |
| openstackgerrit | Dan Krause proposed openstack/taskflow: Added EventTimeListner to record when events occur https://review.openstack.org/163125 | 01:44 |
| *** ChuckC has quit IRC | 01:46 | |
| *** ChuckC has joined #openstack-oslo | 01:47 | |
| *** devlaps has quit IRC | 01:51 | |
| openstackgerrit | Dan Krause proposed openstack/taskflow: update uses of TimingListener to DurationListener https://review.openstack.org/163258 | 01:53 |
| *** dims_ has joined #openstack-oslo | 01:55 | |
| *** dims has quit IRC | 01:58 | |
| *** takedakn has joined #openstack-oslo | 02:04 | |
| *** harlowja_ is now known as harlowja_away | 02:09 | |
| *** yamahata has quit IRC | 02:13 | |
| *** zzzeek has quit IRC | 02:30 | |
| *** salv-orlando has joined #openstack-oslo | 02:31 | |
| *** david-lyle_afk has quit IRC | 02:36 | |
| *** salv-orlando has quit IRC | 02:38 | |
| *** takedakn1 has joined #openstack-oslo | 02:40 | |
| *** takedakn has quit IRC | 02:40 | |
| openstackgerrit | Dan Krause proposed openstack/taskflow: (WIP) lazy loading for logbooks and flowdetails https://review.openstack.org/163269 | 02:43 |
| *** takedakn1 has quit IRC | 02:46 | |
| *** salv-orlando has joined #openstack-oslo | 02:57 | |
| *** achanda has joined #openstack-oslo | 02:58 | |
| *** dims_ has quit IRC | 03:02 | |
| *** salv-orlando has quit IRC | 03:10 | |
| *** achanda has quit IRC | 03:10 | |
| *** david-lyle_afk has joined #openstack-oslo | 03:17 | |
| *** stevemar has quit IRC | 03:34 | |
| *** jgrimm is now known as zz_jgrimm | 03:40 | |
| *** zz_jgrimm is now known as jgrimm | 03:47 | |
| *** YorikSar has joined #openstack-oslo | 03:52 | |
| *** dims has joined #openstack-oslo | 04:03 | |
| *** achanda has joined #openstack-oslo | 04:07 | |
| *** dims has quit IRC | 04:10 | |
| *** miqui has quit IRC | 04:14 | |
| openstackgerrit | Merged openstack/taskflow: Denote issue 17911 has been merged/accepted https://review.openstack.org/162796 | 04:28 |
| *** harlowja_at_home has joined #openstack-oslo | 04:35 | |
| *** amotoki has joined #openstack-oslo | 04:36 | |
| *** YorikSar has quit IRC | 04:38 | |
| *** crc32 has quit IRC | 04:39 | |
| *** salv-orlando has joined #openstack-oslo | 04:57 | |
| *** rushiagr_away is now known as rushiagr | 04:58 | |
| *** salv-orlando has quit IRC | 05:10 | |
| *** salv-orlando has joined #openstack-oslo | 05:12 | |
| *** salv-orlando has quit IRC | 05:16 | |
| *** salv-orlando has joined #openstack-oslo | 05:29 | |
| *** achanda has quit IRC | 05:31 | |
| *** salv-orlando has quit IRC | 05:34 | |
| *** achanda has joined #openstack-oslo | 05:34 | |
| *** salv-orlando has joined #openstack-oslo | 05:35 | |
| *** salv-orlando has joined #openstack-oslo | 05:41 | |
| *** salv-orlando has quit IRC | 05:45 | |
| *** achanda has quit IRC | 05:48 | |
| *** bnemec has quit IRC | 05:49 | |
| *** bnemec has joined #openstack-oslo | 05:52 | |
| *** achanda has joined #openstack-oslo | 05:52 | |
| *** YorikSar has joined #openstack-oslo | 05:56 | |
| *** achanda has quit IRC | 05:58 | |
| *** achanda has joined #openstack-oslo | 06:01 | |
| *** ajo has joined #openstack-oslo | 06:03 | |
| *** salv-orlando has joined #openstack-oslo | 06:04 | |
| *** salv-orlando has quit IRC | 06:08 | |
| *** inc0 has joined #openstack-oslo | 06:11 | |
| *** harlowja_at_home has quit IRC | 06:11 | |
| *** achanda has quit IRC | 06:14 | |
| *** salv-orlando has joined #openstack-oslo | 06:15 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/oslo.utils: Imported Translations from Transifex https://review.openstack.org/160649 | 06:18 |
| *** salv-orlando has quit IRC | 06:20 | |
| *** salv-orlando has joined #openstack-oslo | 06:21 | |
| *** salv-orlando has quit IRC | 06:26 | |
| *** salv-orlando has joined #openstack-oslo | 06:27 | |
| *** andreykurilin_ has joined #openstack-oslo | 06:28 | |
| *** kiseok7_ has quit IRC | 06:31 | |
| *** kiseok7_ has joined #openstack-oslo | 06:32 | |
| *** sreshetn1 has joined #openstack-oslo | 06:32 | |
| *** salv-orlando has quit IRC | 06:32 | |
| *** salv-orlando has joined #openstack-oslo | 06:38 | |
| *** salv-orlando has quit IRC | 06:43 | |
| *** andreykurilin_ has quit IRC | 06:57 | |
| *** takedakn has joined #openstack-oslo | 07:00 | |
| *** ajo has quit IRC | 07:19 | |
| *** ajo has joined #openstack-oslo | 07:23 | |
| *** dulek has joined #openstack-oslo | 07:30 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow injected atom args to be persisted https://review.openstack.org/163159 | 07:36 |
| *** sreshetn1 has quit IRC | 07:45 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow injected atom args to be persisted https://review.openstack.org/163159 | 07:45 |
| openstackgerrit | Lan Qi Song proposed openstack/oslo.config: Support deprecate both name and group https://review.openstack.org/163316 | 07:45 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow injected atom args to be persisted https://review.openstack.org/163159 | 07:47 |
| *** shardy has joined #openstack-oslo | 07:48 | |
| *** takedakn has quit IRC | 07:50 | |
| *** lan has joined #openstack-oslo | 07:53 | |
| *** takedakn has joined #openstack-oslo | 07:57 | |
| *** salv-orlando has joined #openstack-oslo | 08:05 | |
| *** e0ne has joined #openstack-oslo | 08:11 | |
| *** salv-orlando has quit IRC | 08:15 | |
| *** salv-orlando has joined #openstack-oslo | 08:25 | |
| *** salv-orlando has quit IRC | 08:25 | |
| *** jungleboyj has quit IRC | 08:28 | |
| *** jungleboyj has joined #openstack-oslo | 08:32 | |
| *** i159 has joined #openstack-oslo | 08:41 | |
| *** ajo has quit IRC | 08:46 | |
| *** ajo has joined #openstack-oslo | 08:47 | |
| *** sreshetn1 has joined #openstack-oslo | 08:55 | |
| openstackgerrit | Mehdi Abaakouk proposed openstack/oslo.messaging: Update to oslo.context https://review.openstack.org/159778 | 09:10 |
| openstackgerrit | Mehdi Abaakouk proposed openstack/oslo.messaging: Deprecates the localcontext API https://review.openstack.org/140616 | 09:10 |
| *** e0ne has quit IRC | 09:27 | |
| *** ihrachyshka has joined #openstack-oslo | 09:28 | |
| *** ihrachyshka has quit IRC | 09:37 | |
| *** ihrachyshka has joined #openstack-oslo | 09:38 | |
| *** salv-orlando has joined #openstack-oslo | 09:44 | |
| *** salv-orlando has quit IRC | 09:46 | |
| jd__ | dhellmann: btw you owe me a response in https://review.openstack.org/#/c/148500/ I think | 09:55 |
| *** salv-orlando has joined #openstack-oslo | 10:10 | |
| *** sreshetn1 has quit IRC | 10:13 | |
| *** salv-orlando has quit IRC | 10:14 | |
| *** salv-orl_ has joined #openstack-oslo | 10:15 | |
| *** salv-orl_ has quit IRC | 10:15 | |
| *** e0ne has joined #openstack-oslo | 10:16 | |
| *** dims has joined #openstack-oslo | 10:18 | |
| *** vigneshvar has joined #openstack-oslo | 10:40 | |
| *** takedakn has quit IRC | 10:42 | |
| *** salv-orlando has joined #openstack-oslo | 10:48 | |
| *** _amrith_ is now known as amrith | 11:08 | |
| *** red_ has joined #openstack-oslo | 11:14 | |
| *** e0ne is now known as e0ne_ | 11:21 | |
| *** e0ne_ is now known as e0ne | 11:26 | |
| *** haypo has quit IRC | 11:46 | |
| *** haypo has joined #openstack-oslo | 11:47 | |
| *** amrith is now known as _amrith_ | 11:47 | |
| *** e0ne is now known as e0ne_ | 11:51 | |
| *** inc0 has quit IRC | 11:54 | |
| *** e0ne_ has quit IRC | 11:56 | |
| *** inc0 has joined #openstack-oslo | 12:06 | |
| *** vigneshvar has quit IRC | 12:07 | |
| viktors | dhellmann: ping | 12:12 |
| *** e0ne has joined #openstack-oslo | 12:18 | |
| *** sreshetn1 has joined #openstack-oslo | 12:22 | |
| *** exploreshaifali has joined #openstack-oslo | 12:28 | |
| *** _amrith_ is now known as amrith | 12:29 | |
| *** dims has quit IRC | 12:33 | |
| *** pblaho has joined #openstack-oslo | 12:34 | |
| *** dims has joined #openstack-oslo | 12:34 | |
| *** dulek has quit IRC | 12:38 | |
| *** rushiagr is now known as rushiagr_away | 12:46 | |
| *** gordc has joined #openstack-oslo | 12:53 | |
| *** rushiagr_away is now known as rushiagr | 13:05 | |
| sdague | dhellmann: fyi, towards the end of my nova philly ops meeting email, there are some interesting bits around rabbit / oslo.messaging, especially a patch everyone wants landed. | 13:06 |
| *** amotoki has quit IRC | 13:08 | |
| *** jecarey has joined #openstack-oslo | 13:12 | |
| *** amotoki has joined #openstack-oslo | 13:15 | |
| *** rushiagr is now known as rushiagr_away | 13:20 | |
| *** cdent has joined #openstack-oslo | 13:22 | |
| *** red_ has quit IRC | 13:23 | |
| *** amrith is now known as _amrith_ | 13:24 | |
| *** mriedem has joined #openstack-oslo | 13:27 | |
| dims | sdague: ack the rabbitmq heartbeat patch | 13:27 |
| dims | sdague: first trying to verify that the patch does work as we lack functional testing | 13:27 |
| dims | sileht: ^^ :) | 13:27 |
| sileht | dims, sdague, yes it's hard to test | 13:30 |
| *** dulek has joined #openstack-oslo | 13:30 | |
| dims | sileht: so how did you test it when you wrote the patch? | 13:31 |
| * dims inviting ozamiatin to listen in | 13:31 | |
| *** inc0 has quit IRC | 13:32 | |
| dims | ozamiatin: and i can try to replicate that same testing sileht hence asking | 13:32 |
| *** jungleboyj has quit IRC | 13:33 | |
| sileht | I have a small script to create rpc client and server, I have make some call() then sleep | 13:33 |
| sileht | I restart rabbitmq | 13:34 |
| sileht | the server immedialty detect the connection lost since oslo.msg 1.6 (when we start poll the broker at least every 1 sec) | 13:34 |
| sileht | and then see on the client part that all polled connection used for call() have been restart when next heartbeat occurs | 13:35 |
| *** gordc_ has joined #openstack-oslo | 13:36 | |
| sileht | dims, also the large ops jobs is a good tests to ensure that rabbit doesn't unexpectly closes the connection because it doesn't receive a heartbeat (all previous implementations tentative was failing randomly on that job) | 13:37 |
| dims | sileht: cool, can you share the script please? | 13:39 |
| sileht | dims, https://gist.github.com/sileht/571a0a9204c3a96896ab, this not the exact version I have used for my testing | 13:40 |
| dims | sileht: thanks, | 13:41 |
| sileht | dims, you have to reintroduce the sleep() | 13:41 |
| dims | sileht: approx which line? | 13:41 |
| *** gordc has quit IRC | 13:42 | |
| sileht | dims, just after the main() to not close the connections when all the calls have finished | 13:43 |
| dims | cool thanks | 13:43 |
| *** stpierre has joined #openstack-oslo | 13:47 | |
| sileht | dims, thanks for testing it ! | 13:49 |
| *** e0ne has quit IRC | 13:57 | |
| *** e0ne has joined #openstack-oslo | 13:57 | |
| dhellmann | Nakato: let's not let the fact that we don't have a maintainer block us from starting the graduation work | 13:59 |
| e0ne | dhellmann, Nakato: +1 from me on it | 13:59 |
| dhellmann | jd__: I don't think we can remove that function since it has so many users | 14:00 |
| *** jgrimm is now known as zz_jgrimm | 14:00 | |
| *** gordc_ is now known as gordc | 14:00 | |
| dhellmann | sdague: I'll check it out. | 14:00 |
| jd__ | dhellmann: uh? since when is it impossible to deprecate for removal? :( | 14:01 |
| dhellmann | jd__: the fact that the function is small and apparently useless to you isn't, imo, sufficient justification to require all of the users of it to have to go touch code that is working for them | 14:02 |
| *** exploreshaifali has quit IRC | 14:02 | |
| dhellmann | sileht: is that test script part of your patch? | 14:02 |
| dims | dhellmann: no | 14:02 |
| jd__ | dhellmann: it's more than useless, it's dangerous, bad idea to use it, we should deprecate it to stop having it spread everywhere | 14:03 |
| dhellmann | dims, sileht : it would be useful to add it, at least for documentation purposes | 14:03 |
| *** zzzeek has joined #openstack-oslo | 14:03 | |
| *** sreshetn1 has quit IRC | 14:04 | |
| dhellmann | jd__: then you need to publicize the plan more and get some buy-in from some of the other projects. Maybe a ML thread? | 14:04 |
| jd__ | really? a ml thread to bike shed on that? :( | 14:04 |
| jd__ | it's a deprecation warning | 14:04 |
| dhellmann | jd__: people do not pay close enough attention to plans we make in reviews or in here, so what I predict happening is someone filing a bug against the library because it is logging deprecation warnings too much and causing a nova job to fail | 14:05 |
| *** rushiagr_away is now known as rushiagr | 14:06 | |
| jd__ | dhellmann: which is orthogonal to the fact that this patch is good or bad | 14:06 |
| dhellmann | there will be a lot of anger, and more calls to just stop oslo from even existing, and I'm not in the mood to deal with that this cycle | 14:06 |
| dhellmann | the patch is bad because it is likely to cause issues downstream from us, and we need to start thinking about that more | 14:06 |
| sileht | dhellmann, this is not really a test script, this is a tools to help me to play around oslo.messaging | 14:07 |
| dhellmann | sileht: ah, ok. still, it might be useful to have it in the repo to share it | 14:07 |
| sileht | dhellmann, it needs some cleanup to do that, it have a lot of hardcoded thing | 14:08 |
| jd__ | I thought we were supposed to resolve the technical debt, not maintaining it | 14:08 |
| *** mdbooth has quit IRC | 14:09 | |
| dhellmann | jd__: we can remove technical debt. You're asking to discard a function that is working as designed in some large number of places. To do that, you need more of a plan than "this thing is going away". We need to provide a replacement or a fix. | 14:10 |
| jd__ | dhellmann: there already is a replacement and a fix (IIRC from my commit message) | 14:10 |
| jd__ | dhellmann: there's just code to change in $project | 14:10 |
| *** dulek has quit IRC | 14:11 | |
| dhellmann | is replacing strtime() with isotime() going to work if we're mixing them? can the time parser handle both formats of values? | 14:11 |
| *** mdbooth has joined #openstack-oslo | 14:14 | |
| *** dulek has joined #openstack-oslo | 14:15 | |
| *** kgiusti has joined #openstack-oslo | 14:20 | |
| *** salv-orlando has quit IRC | 14:23 | |
| *** bknudson has joined #openstack-oslo | 14:24 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:27 | |
| *** jungleboyj has joined #openstack-oslo | 14:27 | |
| *** salv-orlando has joined #openstack-oslo | 14:31 | |
| *** e0ne is now known as e0ne_ | 14:32 | |
| *** e0ne_ is now known as e0ne | 14:33 | |
| *** mtanino has joined #openstack-oslo | 14:36 | |
| *** dulek has quit IRC | 14:37 | |
| *** jungleboyj has quit IRC | 14:38 | |
| *** inc0 has joined #openstack-oslo | 14:41 | |
| *** ChuckC_ has joined #openstack-oslo | 14:43 | |
| *** ChuckC has quit IRC | 14:44 | |
| *** ChuckC_ has quit IRC | 14:49 | |
| openstackgerrit | Mehdi Abaakouk proposed openstack/oslo.messaging: tools: add simulator script https://review.openstack.org/163464 | 14:55 |
| sileht | dims, dhellmann I have proposed by tools: https://review.openstack.org/#/c/163464/ | 14:55 |
| *** zz_jgrimm is now known as jgrimm | 14:55 | |
| sileht | dims, the example in the script is exactly what I have done to test the heartbeat patch | 14:55 |
| otherwiseguy | dhellmann: With neutron needing rootwrap daemon mode, and it just having a security hole fixed and release made, I assume bumping the version in global-requirements so neutron can use 1.6.0 is ok? (https://review.openstack.org/#/c/163089/) | 14:56 |
| ihrachyshka | dhellmann, hey. some weird bug in oslo.config with --config-dir, not sure whether it's me or the library. | 14:57 |
| sileht | dims, when the client part print 'calls finished, wait 15000', you restart rabbitmq | 14:57 |
| ihrachyshka | dhellmann, I try to make neutron l3 agent to use --config-dir | 14:57 |
| ihrachyshka | so I created a dir (/etc/neutron/l3_agent) | 14:57 |
| sileht | dims, without heartbeat patch client doesn't detect the closed connection and with it detects it | 14:57 |
| dhellmann | otherwiseguy: see https://review.openstack.org/162656 | 14:57 |
| ihrachyshka | moved fwaas_driver.ini file into the dir, and pass --config-dir=/etc/neutron/l3_agent to l3_agent | 14:57 |
| ihrachyshka | but it complains that fwaas is not enabled (so it does not see enabled=True from the fwaas_driver.ini file in the dir) | 14:58 |
| dims | sileht: going to fast approve it! :) | 14:58 |
| dhellmann | ihrachyshka: does the log output show you where it is loading config settings? | 14:58 |
| otherwiseguy | dhellmann: great, thanks. | 14:59 |
| ihrachyshka | dhellmann, not really, I don't see any relevant output | 14:59 |
| ihrachyshka | (debug is True in neutron.conf) | 14:59 |
| dhellmann | ihrachyshka: I'm in the middle of something else now, but I would start by making oslo.config show what it's doing. If it doesn't already log that at at least debug level, I'd be happy to have a patch that makes it do it. | 15:00 |
| openstackgerrit | Timofey Durakov proposed openstack/oslo.db: InsertFromSelect class refactored https://review.openstack.org/163469 | 15:01 |
| ihrachyshka | dhellmann, oh, once I got into the code, I see the problem. it seems --config-dir parses *.conf files only, and mine is *.ini | 15:01 |
| ihrachyshka | I wonder whether it should limit filenames like that | 15:01 |
| ihrachyshka | oh, it's actually mentioned in option description | 15:02 |
| *** stevemar has joined #openstack-oslo | 15:02 | |
| ihrachyshka | so - the bug is in my head :) | 15:02 |
| otherwiseguy | dhellmann: so the global-requirements.txt patch won't merge until after feature freeze? | 15:04 |
| otherwiseguy | or did I read that wrong? | 15:04 |
| dhellmann | otherwiseguy: right, but in the mean time pip will cause the latest version to be used | 15:05 |
| otherwiseguy | dhellmann: my patch fails tests due to the requirements test failing, so wouldn't be mergeable until after the requirements patch passed, though, right? | 15:06 |
| otherwiseguy | which also needs to merge before feature freeze. | 15:06 |
| *** prometheanfire has joined #openstack-oslo | 15:06 | |
| *** daniel3_ has joined #openstack-oslo | 15:07 | |
| prometheanfire | is there a reason oslo.db had test deps added to the main requirements file? https://github.com/openstack/oslo.db/blob/master/requirements.txt#L16-L17 | 15:07 |
| otherwiseguy | dhellmann: does that mean that I should not be putting the the requirement in our requirements.txt yet? That seems odd. | 15:08 |
| dhellmann | otherwiseguy: which patch is failing? | 15:08 |
| otherwiseguy | dhellmann: https://review.openstack.org/#/c/158805/ | 15:08 |
| dhellmann | prometheanfire: see https://review.openstack.org/#/c/157135/ for the rationale | 15:08 |
| otherwiseguy | gate-neutron-requirements http://logs.openstack.org/05/158805/8/check/gate-neutron-requirements/aa766f0/ : Incompatible requirement found; see https://wiki.openstack.org/wiki/Requirements in 17s | 15:09 |
| dhellmann | otherwiseguy: remove the change to requirements.txt in your patch and it will work | 15:09 |
| otherwiseguy | dhellmann: ok. just checking that that is what is supposed to happen. felt weird. | 15:09 |
| dhellmann | pip will do the right thing, and when the requirements pins are put into place they will ensure the right version is used | 15:09 |
| otherwiseguy | dhellmann: works for me. thanks. | 15:10 |
| dhellmann | otherwiseguy: np | 15:10 |
| *** inc0 has quit IRC | 15:11 | |
| prometheanfire | dhellmann: I guess I'm more questioning the inclusion of a test library in the main code | 15:11 |
| dhellmann | prometheanfire: the library provides a test fixture which is part of the public API, so the test library is a dependency | 15:12 |
| dhellmann | prometheanfire: if you disagree with that policy, please respond to the spec review | 15:12 |
| dhellmann | prometheanfire: the paragraph starting on line 44 is a relevant section | 15:13 |
| prometheanfire | ya, read that | 15:13 |
| openstackgerrit | Dan Krause proposed openstack/taskflow: add _listeners_from_job method to Conductor base https://review.openstack.org/162729 | 15:14 |
| openstackgerrit | Mehdi Abaakouk proposed openstack/oslo.messaging: Deprecates the localcontext API https://review.openstack.org/140616 | 15:15 |
| *** ChuckC_ has joined #openstack-oslo | 15:16 | |
| prometheanfire | I can see why testresources was added, but not testscenarios, as that seems to only be used in tests, not part of the public api | 15:16 |
| prometheanfire | https://github.com/openstack/oslo.db/commit/6ccea346d9609d47b188b33c686f49ce3f4f9b14#diff-1bd3274a6859c8124b815fdb5481f38fR13 | 15:16 |
| prometheanfire | the closest it could get to the public api is oslo_db/sqlalchemy/test_base.py | 15:17 |
| dhellmann | prometheanfire: please respond on the review, I'm in the middle of some other stuff I have to concentrate on | 15:18 |
| prometheanfire | dhellmann: ok | 15:19 |
| dhellmann | prometheanfire: thanks, we're collecting input on this so it will be good to have it all in one place | 15:19 |
| prometheanfire | dhellmann: I don't have a problem with the policy, it's more of an additional dep slipping from test to normal dep | 15:19 |
| prometheanfire | as a packager less deps is good, all these oslo packages getting added are hell | 15:20 |
| prometheanfire | here's the discussion https://review.openstack.org/#/c/120870/23/requirements.txt | 15:24 |
| dhellmann | prometheanfire: which platform do you package? | 15:24 |
| prometheanfire | dhellmann: gentoo :P | 15:24 |
| dhellmann | ah | 15:24 |
| dhellmann | I didn't realize anyone was working on gentoo support, that's great | 15:24 |
| prometheanfire | set up a nice release pipeline | 15:24 |
| *** inc0 has joined #openstack-oslo | 15:24 | |
| prometheanfire | dhellmann: I think we package quicker then anyone else, also have test support | 15:25 |
| prometheanfire | etcetc | 15:25 |
| prometheanfire | anyway, I can see why this was added now, just feels odd | 15:26 |
| dhellmann | prometheanfire: you aren't the first person to react that way, so we're reconsidering our approach | 15:26 |
| *** YorikSar has quit IRC | 15:27 | |
| prometheanfire | ya, I heard that there might be an oslo-requests, which kinda seems insane :P | 15:27 |
| *** hogepodge has quit IRC | 15:28 | |
| *** david-lyle_afk is now known as david-lyle | 15:28 | |
| ihrachyshka | :w | 15:28 |
| *** mtanino is now known as mtanino_away | 15:30 | |
| *** sreshetn1 has joined #openstack-oslo | 15:32 | |
| *** exploreshaifali has joined #openstack-oslo | 15:34 | |
| prometheanfire | dhellmann: as an example... I've had to package nearly 100 packages for openstack | 15:34 |
| dhellmann | prometheanfire: no, I think oslo.requests was more of a joke | 15:35 |
| prometheanfire | 17 oslo things | 15:35 |
| prometheanfire | dhellmann: that's good at least :D | 15:35 |
| dhellmann | prometheanfire: that list is going to grow again next cycle | 15:35 |
| prometheanfire | it continuously grows | 15:35 |
| prometheanfire | that's not a problem, it's the rate of churn that sucks | 15:36 |
| prometheanfire | I totally stopped packaging horizon because of the node stuff too :( | 15:36 |
| *** dulek has joined #openstack-oslo | 15:36 | |
| prometheanfire | saw that requirements update and nopenopenope | 15:36 |
| *** inc0 has quit IRC | 15:38 | |
| *** jungleboyj has joined #openstack-oslo | 15:38 | |
| *** dulek has quit IRC | 15:41 | |
| *** exploreshaifali has quit IRC | 15:42 | |
| *** yamahata has joined #openstack-oslo | 16:00 | |
| openstackgerrit | Victor Stinner proposed openstack/oslo.utils: Add exception_to_unicode() function https://review.openstack.org/163507 | 16:06 |
| haypo | ^^ i just proposed a new function to format an exception because it looks like there is no reliable way to log an exception in python 2 and python 3 when the exception contains a non-ASCII character | 16:08 |
| haypo | i'm working on the dummy bug https://bugs.launchpad.net/nova/+bug/1419905 considered as fixed | 16:08 |
| openstack | Launchpad bug 1419905 in OpenStack Compute (nova) "Nova may not start instances when OS is installed with locale not en_US" [Low,Fix committed] - Assigned to Davanum Srinivas (DIMS) (dims-v) | 16:08 |
| haypo | but i'm not sure that it works | 16:08 |
| haypo | the final fix is something like LOG.debug(u'... %(e)s', {'e': exc, ...}) | 16:09 |
| dims | haypo: https://review.openstack.org/#/c/162282/4/nova/virt/libvirt/firewall.py,cm | 16:09 |
| openstackgerrit | Merged openstack/oslo.messaging: Properly distinguish between server index zero and no server https://review.openstack.org/158444 | 16:10 |
| dims | haypo: the u'' is the critical part i guess | 16:10 |
| *** enikanorov__ has joined #openstack-oslo | 16:10 | |
| *** enikanorov has quit IRC | 16:10 | |
| *** jecarey has quit IRC | 16:10 | |
| *** jecarey has joined #openstack-oslo | 16:11 | |
| haypo | dims: see http://paste.openstack.org/show/191635/ - the new code doesn't look to work | 16:12 |
| haypo | dims: i don't understand the fix | 16:12 |
| *** rossella_s has quit IRC | 16:12 | |
| *** wendar_ has joined #openstack-oslo | 16:14 | |
| *** freyes has quit IRC | 16:14 | |
| *** ttx has quit IRC | 16:14 | |
| *** lbragstad has quit IRC | 16:14 | |
| *** jmeridth has quit IRC | 16:14 | |
| *** jroll has quit IRC | 16:14 | |
| *** Nakato has quit IRC | 16:14 | |
| *** flwang1 has quit IRC | 16:14 | |
| *** wendar has quit IRC | 16:14 | |
| *** gus has quit IRC | 16:14 | |
| *** russellb has quit IRC | 16:14 | |
| *** zigo_ has quit IRC | 16:14 | |
| *** zigo_ has joined #openstack-oslo | 16:14 | |
| otherwiseguy | For some reason I seem to be completely incapable of installing oslo.context==0.2.0. It's like pip can't see the links for the 0.2.0 version? Example: http://www.fpaste.org/196722/60899141/ | 16:15 |
| *** russellb has joined #openstack-oslo | 16:15 | |
| *** jroll has joined #openstack-oslo | 16:15 | |
| *** lbragstad has joined #openstack-oslo | 16:15 | |
| *** jmeridth has joined #openstack-oslo | 16:15 | |
| otherwiseguy | The links are there at pypip.python.org/simple/oslo.context/ when I browse there, they just don't appear to be noticed by pip? | 16:16 |
| haypo | dims: i tried to document the best practice a few weeks ago in https://wiki.openstack.org/wiki/Python3#logging_module_and_format_exceptions | 16:16 |
| *** Nakato has joined #openstack-oslo | 16:16 | |
| *** gus has joined #openstack-oslo | 16:16 | |
| haypo | dims: but there is no solution which works on all cases | 16:16 |
| *** ttx has joined #openstack-oslo | 16:16 | |
| *** jroll has quit IRC | 16:16 | |
| *** jroll has joined #openstack-oslo | 16:16 | |
| *** rossella_s has joined #openstack-oslo | 16:16 | |
| *** flwang1 has joined #openstack-oslo | 16:17 | |
| *** freyes has joined #openstack-oslo | 16:17 | |
| openstackgerrit | Timofey Durakov proposed openstack/oslo.db: InsertFromSelect class refactored https://review.openstack.org/163469 | 16:20 |
| dims | haypo: does this new guidance work? - http://docs-draft.openstack.org/12/147312/5/check/gate-oslo.log-docs/5b2a401//doc/build/html/usage.html#migrating-to-oslo-log | 16:27 |
| dims | haypo: https://review.openstack.org/#/c/147312 << from this review | 16:27 |
| *** inc0 has joined #openstack-oslo | 16:29 | |
| *** mtanino_away has quit IRC | 16:32 | |
| *** dulek has joined #openstack-oslo | 16:33 | |
| haypo | dims: no, i don't think that examples at http://docs-draft.openstack.org/12/147312/5/check/gate-oslo.log-docs/5b2a401//doc/build/html/usage.html#migrating-to-oslo-log work with non-ASCII characters | 16:34 |
| haypo | >>> logging.exception(Exception('\xe9'.encode('utf8'))) | 16:34 |
| haypo | UnicodeDecodeError: 'ascii' codec can't decode byte 0xe9 in position 0: ordinal not in range(128) | 16:34 |
| haypo | oops, it's: | 16:35 |
| haypo | >>> logging.exception(Exception('\xe9'.decode('utf8'))) | 16:35 |
| haypo | UnicodeDecodeError: 'utf8' codec can't decode byte 0xe9 in position 0: unexpected end of data | 16:35 |
| haypo | way, no, what i am doing? :-p | 16:35 |
| haypo | >>> logging.exception(Exception(u'\xe9'.encode('utf8'))) | 16:35 |
| haypo | ERROR:root:é | 16:35 |
| dims | haypo: there's one more review to consider https://review.openstack.org/#/c/163027/ | 16:35 |
| haypo | ok, this one works ;) | 16:35 |
| haypo | >>> logging.critical(u"error: %s", Exception(u'\xe9'.encode('utf8'))) | 16:36 |
| haypo | UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 0: ordinal not in range(128) | 16:36 |
| haypo | this one doesn't work | 16:36 |
| dims | will that work with 163027? | 16:36 |
| openstackgerrit | Mehdi Abaakouk proposed openstack/oslo.messaging: tools: add simulator script https://review.openstack.org/163464 | 16:36 |
| haypo | dims: hum, we don't use directly the logging module? | 16:37 |
| dims | haypo: "from oslo_log import log as logging" is a common way for us | 16:38 |
| haypo | dims: i was not aware of the magic unicode conversion | 16:40 |
| haypo | dims: but no, converting the message to unicode is not enough :-( | 16:40 |
| haypo | dims: the problem occurs at fmt % args, even if fmt type is unicode | 16:41 |
| dims | haypo: fmt is unicode and args is unicode it would work right? | 16:42 |
| haypo | well, i should dig all these issues | 16:42 |
| haypo | dims: yes | 16:42 |
| haypo | dims: the problem is very specific to exceptions | 16:42 |
| haypo | dims: in short, it's not possible to convert an exception to unicode in python 2 | 16:42 |
| haypo | dims: so u'...%s...' % exc will fail | 16:42 |
| haypo | we must write u'...%s...' % exception_to_unicode(exc) to avoid issues | 16:43 |
| dims | "Now, the error should be converted to unicode either by calling six.text_type() or by using a unicode formatting string to provide context." | 16:43 |
| dims | right? | 16:44 |
| *** exploreshaifali has joined #openstack-oslo | 16:44 | |
| dims | that takes care of the args. the automatic conversion in 163027 takes care of the fmt | 16:44 |
| dims | no? | 16:45 |
| *** i159 has quit IRC | 16:46 | |
| dims | haypo? | 16:46 |
| haypo | dims: unicode(Exception) doesn't work if the exception message contains a non-ASCII character | 16:48 |
| haypo | dims: let me check 163027 | 16:49 |
| *** inc0 has quit IRC | 16:51 | |
| haypo | dims: i have to go. i will read all these patches | 16:51 |
| *** andreykurilin_ has joined #openstack-oslo | 16:51 | |
| haypo | dims: thanks for the pointer to 163027 | 16:51 |
| *** openstack has joined #openstack-oslo | 16:55 | |
| david-lyle | prometheanfire: question, do you prefer all the javascript libraries vendored in the horizon code base as they used to be? | 16:56 |
| david-lyle | we're just being honest about our dependencies now | 16:56 |
| *** andreykurilin_ has quit IRC | 16:56 | |
| prometheanfire | yes | 16:57 |
| *** takedakn has joined #openstack-oslo | 16:57 | |
| prometheanfire | if only to avoid that insanity | 16:57 |
| prometheanfire | not the right answer, but meh | 16:57 |
| david-lyle | prometheanfire: I understand, but don't you have to gut those out in the end anyway? | 16:58 |
| *** zul has joined #openstack-oslo | 16:58 | |
| prometheanfire | what do you mean? | 16:58 |
| david-lyle | prometheanfire: actually I don't know your use case, if it's only for internal consumption that's probably not a concern | 17:01 |
| prometheanfire | it's not | 17:02 |
| prometheanfire | it was packaged wrong in the first place | 17:02 |
| prometheanfire | webapps are not fun to package | 17:03 |
| david-lyle | indeed | 17:03 |
| david-lyle | web tools are built for site deployments not general tool reusable site deployment | 17:03 |
| *** salv-orlando has quit IRC | 17:03 | |
| david-lyle | which where most of our difficulties come in | 17:04 |
| david-lyle | *is where | 17:04 |
| prometheanfire | we kinda have something for that in gentoo (webapp-config) | 17:04 |
| prometheanfire | I could use a site deploy type thing maybe | 17:04 |
| *** dulek has quit IRC | 17:05 | |
| david-lyle | but most deployments don't want to be pulling down random packages from github, nor can they in a lot of cases | 17:05 |
| *** takedakn has quit IRC | 17:05 | |
| prometheanfire | that's what happens? | 17:06 |
| david-lyle | no, that's what we're avoiding by making packages | 17:06 |
| prometheanfire | what is causing you to need to avoid that | 17:07 |
| prometheanfire | why is it not a static file included in horizon? | 17:07 |
| david-lyle | versioning becomes a problem, additionally, feedback from other distros has been that they have to remove that code anyway and just link to the version of the javascript package they already ship | 17:08 |
| david-lyle | so that the security, maintenance burden is less | 17:08 |
| *** takedakn has joined #openstack-oslo | 17:09 | |
| david-lyle | but when they did that, they had no idea what version of the library they were replacing, and that caused problems with horizon expecting features in versions (newer or older) that weren't there | 17:09 |
| prometheanfire | ok, my complaint is more on the end of packaging all those missing deps | 17:10 |
| david-lyle | I get it | 17:11 |
| prometheanfire | ya | 17:11 |
| david-lyle | imagine trying to push all those through the openstack requirements process | 17:11 |
| david-lyle | :P | 17:11 |
| david-lyle | wonder if one package to rule them all would be better? | 17:12 |
| david-lyle | one big blob of javascript dependencies rolled up together in a toxic little ball | 17:12 |
| prometheanfire | lol,ya | 17:12 |
| prometheanfire | so toxic | 17:12 |
| david-lyle | I'm open to suggestions | 17:13 |
| david-lyle | because both the old and new way are a mess | 17:13 |
| prometheanfire | ya | 17:13 |
| prometheanfire | the toxic ball worksforme | 17:13 |
| *** harlowja_away is now known as harlowja_ | 17:14 | |
| david-lyle | will look into | 17:14 |
| david-lyle | thanks for the feedback, I really want a workable solution for as many as I can | 17:15 |
| prometheanfire | asking in #openstack-stable and the operators mailing list might get you more feedback | 17:16 |
| david-lyle | prometheanfire: will continue there, thanks again | 17:17 |
| *** ihrachyshka has quit IRC | 17:18 | |
| *** e0ne is now known as e0ne_ | 17:18 | |
| *** YorikSar has joined #openstack-oslo | 17:20 | |
| *** sputnik13 has joined #openstack-oslo | 17:28 | |
| *** wendar_ is now known as wendar | 17:28 | |
| *** e0ne_ has quit IRC | 17:29 | |
| *** exploreshaifali has quit IRC | 17:29 | |
| *** e0ne has joined #openstack-oslo | 17:30 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow injected atom args to be persisted https://review.openstack.org/163159 | 17:32 |
| *** viktors is now known as viktors|afk | 17:33 | |
| *** takedakn has quit IRC | 17:37 | |
| *** cdent has quit IRC | 17:39 | |
| *** e0ne has quit IRC | 17:41 | |
| *** devlaps has joined #openstack-oslo | 17:42 | |
| *** sreshetn1 has quit IRC | 17:43 | |
| *** inc0 has joined #openstack-oslo | 17:47 | |
| *** mtanino has joined #openstack-oslo | 17:52 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow using ordered(set/dict) to retain ordering https://review.openstack.org/160558 | 17:53 |
| *** achanda has joined #openstack-oslo | 17:54 | |
| *** inc0 has quit IRC | 17:57 | |
| *** exploreshaifali has joined #openstack-oslo | 18:01 | |
| *** salv-orlando has joined #openstack-oslo | 18:04 | |
| *** stevemar has quit IRC | 18:04 | |
| *** stevemar has joined #openstack-oslo | 18:05 | |
| *** _amrith_ is now known as amrith | 18:07 | |
| *** inc0 has joined #openstack-oslo | 18:08 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Use ordered[set/dict] to retain ordering https://review.openstack.org/160558 | 18:12 |
| *** takedakn has joined #openstack-oslo | 18:12 | |
| *** kbyrne has quit IRC | 18:13 | |
| dhellmann | dims: I have a tempest patch up to convert to use all of the current oslo libs: https://review.openstack.org/#/c/163549/1 | 18:14 |
| dhellmann | bnemec: ^^ | 18:14 |
| *** kbyrne has joined #openstack-oslo | 18:14 | |
| *** salv-orlando has quit IRC | 18:14 | |
| *** vigneshvar has joined #openstack-oslo | 18:16 | |
| *** ndipanov has joined #openstack-oslo | 18:17 | |
| mriedem | there is still not a thing in oslo.config to mark an option as deprecated and not being renamed, but being gone in a release, right? | 18:19 |
| dhellmann | mriedem: we *just* added that in the last release | 18:19 |
| bnemec | mriedem: There is now. :-) | 18:19 |
| mriedem | omfg | 18:19 |
| * mriedem is all over it | 18:19 | |
| bnemec | dhellmann: That looks like a really fun patch. ;-) | 18:19 |
| dhellmann | bnemec: I got surprisingly far by using "sed -i" :-) | 18:20 |
| bnemec | dhellmann: That's good. I'd hate to think you had to do the whole thing by hand. | 18:20 |
| bnemec | Ick, those lock_path references though. :-/ | 18:21 |
| dhellmann | I did have to go back and sort the imports by hand in most places, but emacs + flake8 in a compilation buffer made that fairly quick | 18:21 |
| dhellmann | bnemec: yeah, I talked with mtreinish about that. We need to decide if we want to provide a way to ask lockutils what path it is using, or if we want tempest to somehow own that value and just send it down into lockutils | 18:22 |
| dhellmann | I figured this was ok for now, though, since the patch was originally just going to be about adopting oslo.log so barbican's tests would be unblocked | 18:22 |
| bnemec | dhellmann: Yeah, since we had to drop the config filter anyway I guess there's no need to worry about it immediately. | 18:25 |
| dhellmann | bnemec: right | 18:25 |
| bnemec | It kind of looks like those things maybe don't belong in the lock path and should be in a separate temp directory. | 18:25 |
| mtreinish | bnemec: the issue is that dir has to be the same between all the test runner processes so using tempfile wouldn't work | 18:25 |
| dhellmann | bnemec: yeah, it wasn't clear exactly what they were so I left them alone for now | 18:26 |
| dhellmann | so it sounds like it doesn't have to be the lock dir, it just has to be the same for all processes and right now the lock dir fits that requirement | 18:26 |
| mtreinish | dhellmann: exactly | 18:26 |
| dhellmann | jungleboyj: have a sec to chat about https://bugs.launchpad.net/oslo.db/+bug/1417018 ? | 18:26 |
| openstack | Launchpad bug 1417018 in Cinder "Cinder encounters dbapi error: NoSuchColumnError: "Could not locate column in row for column '%(140070887032400 anon)s.volumes_id'"" [High,In progress] - Assigned to Eric Harney (eharney) | 18:26 |
| bnemec | mtreinish: Yeah, that's a pain. | 18:26 |
| dhellmann | maybe we can put something in oslotest to help solve this | 18:27 |
| mtreinish | dhellmann: I was thinking we could just add a lock_path tempest option back to the default group, and just use set_defaults() to set it for lockutils in tempest/config.py | 18:27 |
| dhellmann | sort of like the lockutils wrapper in oslo.concurrency, but more general | 18:27 |
| dhellmann | mtreinish: you'd need to use a different name, I think, but that could work too | 18:27 |
| *** amrith is now known as _amrith_ | 18:27 | |
| mtreinish | oh right because of the deprecated on the lockutils option | 18:28 |
| dhellmann | yeah | 18:28 |
| dhellmann | but yeah, that's a pretty straightforward solution | 18:28 |
| *** bknudson has quit IRC | 18:30 | |
| mtreinish | dhellmann: if I added that opt would using deprecated opts on the one in lockutils one work? So if people who have default.lock_path or oslo_concurrency.lock_path in existing config files it would be seemless? | 18:31 |
| dhellmann | right now if they have default.lock_path that will be picked up and used as the value for the existing option in oslo.concurrency | 18:32 |
| dhellmann | I don't know what would happen if you registered another option with the same name in default, but I suspect it would not work as you want. | 18:32 |
| mtreinish | no I meant if I added one like default.lock_dir and set deprecated opt on oslo_concurrency.lock_path would it pick up both that and default.lock_path? | 18:33 |
| dhellmann | why would you set the deprecated flag on the option in oslo.concurrency? | 18:33 |
| bnemec | mtreinish: dhellmann: I'm confused. How is Tempest getting a common lock_path without using the lockutils-wrapper script? | 18:33 |
| bnemec | I don't see it used anywhere in the repo. | 18:33 |
| *** jaypipes is now known as jaypipes-afk | 18:34 | |
| dhellmann | bnemec: could the test job in project-config be setting the value? | 18:34 |
| dhellmann | or maybe one of the devstack-gate scripts? | 18:34 |
| dhellmann | it works by setting an environment variable, right? | 18:34 |
| *** vigneshvar has quit IRC | 18:35 | |
| bnemec | It's possible, but that would be bad since it wouldn't work in local runs. | 18:35 |
| mtreinish | tempest has it's own config file, so it just loads the lock_path from config before it runs any tests | 18:35 |
| mtreinish | the config file path is loaded with env vars | 18:35 |
| mtreinish | bnemec: or were you talking about the unit tests? | 18:35 |
| bnemec | mtreinish: No, but that makes sense. Since you have your own config that's your global source of truth. | 18:36 |
| bnemec | We didn't have that for unit tests in the other projects. | 18:36 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add failure json schema validation https://review.openstack.org/163248 | 18:37 |
| bnemec | In that case, +1 to just letting Tempest set its own lock path. | 18:37 |
| mtreinish | dhellmann: for setting the deprecated flag, after your patch lands the sample config will say use the oslo.concurrency option. I want to handle the window between that and the future patch w/ the tempest option gracefully so if someone follows the sample config at any point it'll just work | 18:40 |
| dhellmann | mtreinish: if you want the tempest option to take precedence over the default or user-provided value for the oslo.concurrency option you can call set_overrides() to force-set the oslo.concurrency value (although we don't provide a public way to do that either, yet) | 18:42 |
| dhellmann | bnemec: I wonder if we should decouple the lock path from the config a bit more. Use a module-level value to store it, provide an API to set/get it, and use the config setting if the value isn't set when something asks for it. | 18:43 |
| dhellmann | alternately we could provide a function to fetch the current lock path value, that would be a simple change to add | 18:43 |
| mtreinish | dhellmann: the inverse case. Lets say I pull tempest after your patch lands, I set lock_path in the oslo_concurrency group because that's what the sample tells me. Then next week we land a patch adding lock_dir to default in tempest and I pull that. Things will break if I try to reuse the config file | 18:44 |
| mtreinish | because the code in accounts.py is looking for lock_dir | 18:44 |
| *** salv-orlando has joined #openstack-oslo | 18:44 | |
| dhellmann | mtreinish: you can set the default for your option to %(oslo_concurrency.lock_path)s (I think that's the syntax) | 18:45 |
| *** inc0 has quit IRC | 18:45 | |
| dhellmann | although really that gives us the same issue, which is that you're now relying on us to not change that name | 18:45 |
| mtreinish | ok, that'll work (although we're still dependent on the lib config option then) | 18:45 |
| bnemec | dhellmann: We could do that. The module-level variable would just be another addition to the lock_path precedence we already have. | 18:45 |
| dhellmann | right, so that's not great either | 18:45 |
| *** prometheanfire has left #openstack-oslo | 18:45 | |
| dhellmann | bnemec: does that meet mtreinish's needs, though? | 18:46 |
| dhellmann | I wonder if we really just need a public way to call set_overrides() for that option | 18:46 |
| dhellmann | sort of like what we have for set_defaults() | 18:46 |
| dhellmann | then we aren't syncing the values between different global storage locations | 18:46 |
| bnemec | Seems reasonable. | 18:46 |
| bnemec | And something to query the value, but that's trivial. | 18:47 |
| dhellmann | yeah, I'll do both in the same patch | 18:47 |
| mtreinish | if there is a lock_path query function in the api, that solves all of tempest's needs, because I can just use that in accounts.py | 18:48 |
| *** vigneshvar has joined #openstack-oslo | 18:48 | |
| ndipanov | hey guise - I have a oslo.messaging question | 18:48 |
| dhellmann | mtreinish: ok, we'll start small and just add that then | 18:49 |
| ndipanov | or more a oslo + kombu question | 18:49 |
| ndipanov | so correct me if I am wrong here - but for sending threads -we create a connection (or use one from the pool) and then each connection has only one channel | 18:50 |
| ndipanov | does that sound right? | 18:50 |
| dhellmann | ndipanov: I'm afraid I'm not sure. The best person to ask is sileht, but it's late in Europe so he might not be online any more. | 18:52 |
| ndipanov | dhellmann, thanks - I am europe too so I will try to catch him tomorrow morning | 18:53 |
| dhellmann | ndipanov: sounds good | 18:53 |
| *** mtanino has quit IRC | 18:57 | |
| *** mtanino has joined #openstack-oslo | 18:57 | |
| *** e0ne has joined #openstack-oslo | 18:58 | |
| *** exploreshaifali has quit IRC | 19:03 | |
| openstackgerrit | Doug Hellmann proposed openstack/oslo.concurrency: Add lockutils.get_lock_path() function https://review.openstack.org/163565 | 19:07 |
| openstackgerrit | Doug Hellmann proposed openstack/oslo.concurrency: Remove py33 env from default tox list https://review.openstack.org/163566 | 19:07 |
| dhellmann | bnemec, mtreinish : ^^ | 19:07 |
| *** rushiagr is now known as rushiagr_away | 19:10 | |
| *** bknudson has joined #openstack-oslo | 19:13 | |
| harlowja_ | dims ya, http://logs.openstack.org/76/162976/2/check/check-tempest-dsvm-full/9c47a5d/logs/screen-n-api.txt.gz makes me sad, lol | 19:15 |
| harlowja_ | seems like not so easy to just close stuff | 19:15 |
| *** pblaho has quit IRC | 19:19 | |
| harlowja_ | to many hidden sockets and stuff being shared | 19:20 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add + use failure json schema validation https://review.openstack.org/163248 | 19:21 |
| *** sreshetn1 has joined #openstack-oslo | 19:22 | |
| jungleboyj | dhellmann: I am around now. | 19:24 |
| dims | harlowja_: yea! | 19:26 |
| *** achanda has quit IRC | 19:27 | |
| harlowja_ | although maybe nova should be recreating that socket when it gets a 'Bad file descriptor' on its expected socket | 19:28 |
| harlowja_ | dims lets see what happens with https://review.openstack.org/#/c/163582/ | 19:37 |
| *** achanda has joined #openstack-oslo | 19:38 | |
| harlowja_ | probably still won't work though, lol | 19:42 |
| harlowja_ | but lets see, ha | 19:42 |
| *** takedakn has quit IRC | 19:45 | |
| dhellmann | jungleboyj: I wanted to catch up and see how big of an issue that bug is for the cinder folks, and see if we can agree on a way to fix it. | 19:48 |
| *** vigneshvar has quit IRC | 19:51 | |
| openstackgerrit | Doug Hellmann proposed openstack/oslo.concurrency: Remove py33 env from default tox list https://review.openstack.org/163566 | 19:52 |
| openstackgerrit | Doug Hellmann proposed openstack/oslo.concurrency: Add lockutils.get_lock_path() function https://review.openstack.org/163565 | 19:52 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add + use failure json schema validation https://review.openstack.org/163248 | 19:53 |
| jungleboyj | dhellmann: Other than it being mentioned in #openstack-cinder earlier today, that was the first time I really had heard about it. | 19:56 |
| dhellmann | jungleboyj: zzzeek has been hunting down the source of the problem, and I assumed it was something the cinder team was already aware of and involved with | 19:57 |
| *** inc0 has joined #openstack-oslo | 19:57 | |
| zzzeek | dhellmann: looks like people are on it now that I mentioned it on the list | 19:57 |
| jungleboyj | dhellmann: Not that I have been involved in. | 19:57 |
| jungleboyj | Now eharney has a patch proposed. | 19:57 |
| jungleboyj | We have been focused on feature freeze and behind on bugs. | 19:58 |
| dhellmann | jungleboyj: sure, I'm trying to get ahead of the bug fixing thing and this is one we have listed as critical for the oslo team to help you all work through | 19:59 |
| dhellmann | zzzeek: is the fix just for cinder, then? | 19:59 |
| zzzeek | dhellmann: I think there are multiple “fixes” in play | 19:59 |
| zzzeek | dhellmann: one is the “close all the FDs on service start” fix | 19:59 |
| zzzeek | dhellmann: which will “work”, though clumsily, for everyone | 19:59 |
| dhellmann | zzzeek: yeah, that's harlowja_'s and I don't know if that's actually working yet or not | 19:59 |
| *** ajo has quit IRC | 20:00 | |
| zzzeek | the next is the one I’m helping them in cinder directly which is https://review.openstack.org/#/c/163551/ | 20:00 |
| harlowja_ | zzzeek http://logs.openstack.org/76/162976/2/check/check-tempest-dsvm-full/9c47a5d/logs/screen-n-api.txt.gz (looks like people are re-using other sockets elsewhere and aren't handling re-opening them) | 20:00 |
| harlowja_ | mainly seems to be that wsgi parent pid socket | 20:01 |
| harlowja_ | *which child just dups | 20:01 |
| harlowja_ | *after fork (and after closed) | 20:01 |
| harlowja_ | likely more cases that aren't seen yet (but hopefully not to many others) | 20:02 |
| dhellmann | harlowja_: that behavior makes sense in a web server, doesn't it? | 20:03 |
| *** exploreshaifali has joined #openstack-oslo | 20:03 | |
| dhellmann | zzzeek: I wonder if we can generalize your cinder fix? it looks like cinder's Service class expects a Manager instance which is a subclass of Base | 20:04 |
| harlowja_ | dhellmann ya, i think so; but might be better to try to open it if it died in child, not sure | 20:04 |
| harlowja_ | https://review.openstack.org/#/c/163582/ is seeing what happens there | 20:04 |
| harlowja_ | *with reopening it | 20:04 |
| dhellmann | harlowja_: ok | 20:04 |
| zzzeek | dhellmann: um well I dont know that architecutre but if everyone kind of has a “Service” class with a “.db” element, then sure, somethign like that | 20:04 |
| dhellmann | harlowja_: I guess I'm thinking of forking for each worker, but we're probably not doing that | 20:04 |
| zzzeek | dhellmann: I’d sure rather keep it some interaction between oslo libraries but as usual, blah blah blah -1 zeek ! | 20:05 |
| dhellmann | zzzeek: yeah, that's what I'm looking for. I don't think the incubated version of the Service class makes that assumption | 20:05 |
| harlowja_ | dhellmann what do u mean, thought we were forking :-/ | 20:05 |
| harlowja_ | blah blah blah -1 harlowja lol | 20:05 |
| dhellmann | harlowja_: sorry, per *request* not per worker | 20:05 |
| harlowja_ | man, we should just use some other wsgi container, lol | 20:05 |
| harlowja_ | stop recreating apache prefork , lol | 20:06 |
| *** SridharG has joined #openstack-oslo | 20:06 | |
| *** SridharGaddam has quit IRC | 20:06 | |
| harlowja_ | * http://httpd.apache.org/docs/current/mod/prefork.html ... | 20:06 |
| dims | harlowja_: +1 :) | 20:06 |
| dhellmann | zzzeek: hmm, it looks like "Base" is a common concept, but it's not part of any of the incubated code anywhere | 20:06 |
| zzzeek | dhellmann: i really like the idea of: 1. oslo provides patterns 2. the patterns implicitly handle the extremely common case of “service + fork where db already exists” 3. everyone builds off the API and doesn’t have to remember to abide by this case in each individual project | 20:07 |
| harlowja_ | looks like apache prefork tracks more than we are :-P | 20:08 |
| harlowja_ | http://code.metager.de/source/xref/apache/httpd/server/mpm/prefork/prefork.c#495 | 20:08 |
| dhellmann | zzzeek: right, we just don't have some of the common stuff in a library, yet | 20:08 |
| zzzeek | dhellmann: but as I work with all these projects, I’m cosntantly getting hit hard with, “we want to build this ourselves, we want to do this ourselves, we want to do that” | 20:08 |
| dhellmann | zzzeek: yeah, I hear that a lot, too. We can focus on enabling, though. | 20:08 |
| zzzeek | dhellmann: its the definition of Sis·y·phe·an | 20:08 |
| dhellmann | I don't know, maybe the right thing is to just make oslo.db reconnect | 20:08 |
| harlowja_ | syphilis ? | 20:09 |
| dhellmann | I think sileht said oslo.messaging was already doing that | 20:09 |
| dhellmann | harlowja_: sisyphus | 20:09 |
| zzzeek | now keystone wants to stop using foreign keys. sorta cant win | 20:09 |
| zzzeek | :) | 20:09 |
| zzzeek | combine sisyphus with plate spinning | 20:10 |
| zzzeek | so pushing *ten* rocks up hills | 20:10 |
| zzzeek | :) | 20:10 |
| harlowja_ | thats my nightly workout | 20:11 |
| harlowja_ | lol | 20:11 |
| zzzeek | every time I try to task switch back to the VM where I want to clone devstack, I see another “5” lit up in my email icon | 20:11 |
| dhellmann | zzzeek: I keep my email on a separate desktop so it doesn't distract me while I'm working :-) | 20:11 |
| *** vigneshvar has joined #openstack-oslo | 20:12 | |
| *** _amrith_ is now known as amrith | 20:13 | |
| harlowja_ | foreign keys aren't webscale, duh | 20:13 |
| harlowja_ | lol | 20:13 |
| lifeless | https://www.youtube.com/watch?v=b2F-DItXtZs | 20:16 |
| openstackgerrit | Merged openstack/oslo.messaging: tools: add simulator script https://review.openstack.org/163464 | 20:16 |
| zzzeek | lifeless: https://www.youtube.com/watch?v=bzkRVzciAZg more in my wheelhouse :) | 20:19 |
| *** ihrachyshka has joined #openstack-oslo | 20:19 | |
| lifeless | haven't seen that one, will watch after this meeting | 20:20 |
| dhellmann | zzzeek: if you address my comments about the log level and wording on https://review.openstack.org/#/c/156725/2/oslo_db/sqlalchemy/session.py,cm I'll +2 it | 20:23 |
| zzzeek | dhellmann: wow did someone update the CSS for gerrit | 20:23 |
| dhellmann | zzzeek: oh, I use the "new" view and sent you that link (look in your preferences, you can change it) | 20:24 |
| zzzeek | dhellmann: i ilke it, though its kind of falling off the right side of my screen | 20:24 |
| zzzeek | dhellmann: um what log level do you want | 20:24 |
| *** sreshetn1 has quit IRC | 20:24 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow backend connection config (via fetch) to be a string https://review.openstack.org/163597 | 20:25 |
| dhellmann | zzzeek: yeah, it has to be a wide window. You can remove the ,cm bit and I think it will show you the default view | 20:25 |
| dhellmann | zzzeek: debug | 20:25 |
| dhellmann | zzzeek: the rule I'm applying is that the operator can't do anything about it and we're recovering from the condition | 20:25 |
| zzzeek | dhellmann: OK. So the message you have, waht’s inaccurate about it is that if we are in fact invalidating the connecvtion here, then it will *not* cause any race condition. nothing at all will really go wrong | 20:25 |
| dhellmann | zzzeek: actually, let me double-check that: *are* we recovering in the latest version of your patch? | 20:25 |
| zzzeek | dhellmann: the one you’re pointing me at here does recover | 20:25 |
| dhellmann | zzzeek: ok, good. So for the message, maybe say "Parent process forked with an open database connection, which is being discarded and recreated." or something like that? | 20:26 |
| zzzeek | dhellmann: basically in the pool, if you craete a “checkout” event and raise DBDisconnectionError, the pool says, “OK let me reconnect" | 20:26 |
| zzzeek | dhellmann: yes. jstu give me wahtever msg you want :) im all msged out | 20:26 |
| dhellmann | zzzeek: ok, use what I just wrote then and we'll refine it if someone else comes up with better words :-) | 20:27 |
| zzzeek | holy crap, the VM is still sitting there and i havent cloned devstack yet. geezus | 20:27 |
| dhellmann | harlowja_: even if we move on with zzzeek patch in oslo.db, we should still figure out whether there's a way for your close-all-fds patch to be made to work as a more general way to fix the problem | 20:28 |
| harlowja_ | dhellmann agreed, looking at what apache pre-fork does, it tracks the sockets (or thats what it seems to do); we likely need some mechainsm for that :-/ | 20:28 |
| dhellmann | harlowja_: ew. :-/ | 20:29 |
| harlowja_ | http://code.metager.de/source/xref/apache/httpd/server/mpm/prefork/prefork.c#539 i think | 20:29 |
| harlowja_ | orrrrrr like ummm, just start to get people to use apache , lol | 20:29 |
| *** ChuckC_ has quit IRC | 20:29 | |
| harlowja_ | and then remove all wsgi service.py stuff, lol | 20:30 |
| harlowja_ | *serious talk (no lol) | 20:30 |
| harlowja_ | which seems to be quite similar to what i did ( http://code.metager.de/source/xref/apache/httpd/server/listen.c#888 is the socket closing) | 20:31 |
| harlowja_ | but they have more control over what is closed | 20:32 |
| dhellmann | harlowja_: moving off of the home-grown wsgi thing would be my preference, too. I have no idea if that will work for nova | 20:32 |
| harlowja_ | ya, sadly there are expectations that work now, with an unknown amount of code, that make this switch hard :-/ | 20:33 |
| dhellmann | yeah | 20:34 |
| harlowja_ | dhellmann u on TC right? maybe can start getting the TC to see how this could be done? | 20:36 |
| harlowja_ | since i have no idea, haha | 20:36 |
| dhellmann | harlowja_: someone needs to do the work to figure it out and write a spec proposing it | 20:36 |
| *** ndipanov has quit IRC | 20:36 | |
| dhellmann | this is sort of tied up in the "drop eventlet" work, too -- how is that going? | 20:37 |
| harlowja_ | me thinks not just 1 person could do that, lol | 20:37 |
| dhellmann | sure, but the TC isn't going to do it either | 20:37 |
| harlowja_ | ya, not doing to much there, haypo and i need to merge that stuff | 20:37 |
| harlowja_ | haypo when can we do that?? when u around :-P | 20:38 |
| harlowja_ | *and we need to figure out a good reason to even move, lol | 20:38 |
| dhellmann | he's in europe, so I don't suppose you'll overlap online all that much | 20:38 |
| harlowja_ | ya | 20:38 |
| harlowja_ | chillen in france with jd__ | 20:38 |
| dhellmann | yes, well, that's the main thing. without a sufficiently strong argument in favor of moving at all, the options won't even be considered | 20:38 |
| harlowja_ | understood | 20:39 |
| harlowja_ | which leaves us to be the people making an argument (based on some level of facts) that people will refute (i'm sure people will say, well eventlet doesn't seem that bad, or its acceptable to me...) , which is sorta not the most useful place to be, lol | 20:40 |
| harlowja_ | sounds like being in US congress | 20:41 |
| dhellmann | yep, we have to convince enough other people that moving is the right thing to do, we can't just make the decision | 20:41 |
| *** ajo has joined #openstack-oslo | 20:42 | |
| harlowja_ | https://www.youtube.com/watch?v=W9uTPf_I7l0 | 20:42 |
| harlowja_ | lol | 20:42 |
| *** salv-orlando has quit IRC | 20:43 | |
| harlowja_ | eventlet bill, lol | 20:43 |
| *** takedakn has joined #openstack-oslo | 20:43 | |
| dhellmann | heh | 20:43 |
| *** amrith is now known as _amrith_ | 20:45 | |
| *** achanda has quit IRC | 20:55 | |
| *** mtanino has quit IRC | 20:56 | |
| *** hogepodge has joined #openstack-oslo | 20:58 | |
| *** inc0 has quit IRC | 21:02 | |
| *** kgiusti has left #openstack-oslo | 21:09 | |
| openstackgerrit | Dan Krause proposed openstack/taskflow: lazy loading for logbooks and flowdetails https://review.openstack.org/163269 | 21:10 |
| *** andreykurilin has quit IRC | 21:12 | |
| *** ozamiatin has quit IRC | 21:13 | |
| *** viktors|afk has quit IRC | 21:14 | |
| *** ozamiatin has joined #openstack-oslo | 21:14 | |
| *** viktors has joined #openstack-oslo | 21:15 | |
| *** andreykurilin has joined #openstack-oslo | 21:16 | |
| *** sreshetn1 has joined #openstack-oslo | 21:21 | |
| *** sreshetn1 has quit IRC | 21:25 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Avoid duplicated links in memory backend https://review.openstack.org/163619 | 21:26 |
| *** achanda has joined #openstack-oslo | 21:28 | |
| *** salv-orlando has joined #openstack-oslo | 21:28 | |
| jungleboyj | zzzeek: Sorry, I am all over the place today. So, is this proposed patch OK in your opinion? https://review.openstack.org/#/c/163551/2 | 21:31 |
| zzzeek | i have a +1 there | 21:31 |
| jungleboyj | zzzeek: Ah, so you are Michael. :-) | 21:32 |
| *** boris-42 has quit IRC | 21:32 | |
| dhellmann | zzzeek: if you're still around, your patch is under discussion in #openstack-cinder | 21:43 |
| zzzeek | ok | 21:43 |
| *** stpierre has quit IRC | 21:43 | |
| *** amotoki has quit IRC | 21:50 | |
| *** movielovers has joined #openstack-oslo | 21:50 | |
| *** movielovers has left #openstack-oslo | 21:50 | |
| * dhellmann calls it a day | 21:51 | |
| *** mriedem is now known as mriedem_away | 21:55 | |
| *** exploreshaifali has quit IRC | 22:04 | |
| *** jungleboyj has quit IRC | 22:09 | |
| *** dims has quit IRC | 22:11 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Use the ability to chain exceptions correctly https://review.openstack.org/155187 | 22:19 |
| *** sreshetn1 has joined #openstack-oslo | 22:22 | |
| *** hogepodge has quit IRC | 22:24 | |
| *** pblaho has joined #openstack-oslo | 22:25 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Put underscore in-front of alchemist helper https://review.openstack.org/163631 | 22:26 |
| *** sreshetn1 has quit IRC | 22:26 | |
| *** jaypipes-afk is now known as jaypipes | 22:31 | |
| *** exploreshaifali has joined #openstack-oslo | 22:31 | |
| *** jgrimm is now known as zz_jgrimm | 22:36 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Just use the local conf variable https://review.openstack.org/163634 | 22:37 |
| *** gordc has quit IRC | 22:42 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Avoid duplicated links in memory backend https://review.openstack.org/163619 | 22:42 |
| *** dims_ has joined #openstack-oslo | 22:46 | |
| *** amotoki has joined #openstack-oslo | 22:50 | |
| *** jecarey has quit IRC | 22:51 | |
| *** amotoki has quit IRC | 22:56 | |
| *** daniel3_ has quit IRC | 22:58 | |
| *** vigneshvar has quit IRC | 23:03 | |
| *** e0ne is now known as e0ne_ | 23:04 | |
| *** exploreshaifali has quit IRC | 23:10 | |
| *** jecarey has joined #openstack-oslo | 23:28 | |
| *** bknudson has quit IRC | 23:31 | |
| *** ihrachyshka has quit IRC | 23:34 | |
| *** e0ne_ has quit IRC | 23:50 | |
| *** mtanino has joined #openstack-oslo | 23:52 | |
| *** bknudson has joined #openstack-oslo | 23:52 | |
| *** wwriverrat has joined #openstack-oslo | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!