| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack/oslo.messaging: Fix formatting of code blocks in zmq docs https://review.openstack.org/263482 | 00:03 |
|---|---|---|
| *** gordc has quit IRC | 00:10 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Differentiate failures (internal vs external) https://review.openstack.org/263417 | 00:12 |
| *** browne has quit IRC | 00:24 | |
| *** dims has joined #openstack-oslo | 00:33 | |
| *** dims_ has quit IRC | 00:34 | |
| *** cdent has quit IRC | 00:40 | |
| *** pratikmallya has joined #openstack-oslo | 00:43 | |
| *** pratikmallya has quit IRC | 00:48 | |
| *** zz_dimtruck is now known as dimtruck | 00:49 | |
| openstackgerrit | Vilobh Meshram proposed openstack/tooz: Add Consul Driver https://review.openstack.org/245362 | 00:56 |
| openstackgerrit | Vilobh Meshram proposed openstack/tooz: Add Consul Driver https://review.openstack.org/245362 | 01:03 |
| *** mtanino has quit IRC | 01:08 | |
| *** browne has joined #openstack-oslo | 01:08 | |
| *** zqfan has joined #openstack-oslo | 01:09 | |
| *** jecarey has joined #openstack-oslo | 01:19 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary job priorities https://review.openstack.org/263512 | 01:19 |
| *** sputnik13_ has quit IRC | 01:24 | |
| *** dimtruck is now known as zz_dimtruck | 01:34 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Rename '_emit' -> '_try_emit' since it is best-effort (not ensured) https://review.openstack.org/263520 | 01:41 |
| *** pratikmallya has joined #openstack-oslo | 01:41 | |
| *** amotoki has quit IRC | 01:45 | |
| *** zz_dimtruck is now known as dimtruck | 02:04 | |
| *** pratikmallya has quit IRC | 02:06 | |
| *** vilobhmm11 has quit IRC | 02:18 | |
| *** gcb has joined #openstack-oslo | 02:21 | |
| *** dims has quit IRC | 02:25 | |
| *** dims has joined #openstack-oslo | 02:46 | |
| *** dimtruck is now known as zz_dimtruck | 02:47 | |
| *** zz_dimtruck is now known as dimtruck | 02:48 | |
| *** dimtruck is now known as zz_dimtruck | 02:55 | |
| *** zz_dimtruck is now known as dimtruck | 02:56 | |
| *** dimtruck is now known as zz_dimtruck | 03:00 | |
| *** zz_dimtruck is now known as dimtruck | 03:06 | |
| *** dims has quit IRC | 03:14 | |
| *** jecarey has quit IRC | 03:21 | |
| *** yamahata has quit IRC | 03:33 | |
| *** EinstCrazy has joined #openstack-oslo | 03:44 | |
| *** links has joined #openstack-oslo | 03:52 | |
| *** amotoki has joined #openstack-oslo | 03:54 | |
| *** amotoki has quit IRC | 03:58 | |
| *** amotoki has joined #openstack-oslo | 03:59 | |
| *** yamahata has joined #openstack-oslo | 04:01 | |
| *** browne1 has joined #openstack-oslo | 04:08 | |
| *** browne has quit IRC | 04:10 | |
| *** jecarey has joined #openstack-oslo | 04:21 | |
| *** jecarey has quit IRC | 04:26 | |
| *** harlowja_at_home has joined #openstack-oslo | 04:27 | |
| *** jecarey has joined #openstack-oslo | 04:55 | |
| *** jecarey has quit IRC | 04:59 | |
| openstackgerrit | Oleksii Zamiatin proposed openstack/oslo.messaging: (WIP) [zmq] Heartbeat implementation https://review.openstack.org/257346 | 05:08 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary job priorities https://review.openstack.org/263512 | 05:31 |
| *** harlowja_at_home has quit IRC | 05:39 | |
| *** vilobhmm11 has joined #openstack-oslo | 05:48 | |
| *** sputnik13 has joined #openstack-oslo | 05:59 | |
| *** sputnik13 has quit IRC | 06:01 | |
| *** sputnik13 has joined #openstack-oslo | 06:06 | |
| *** sputnik13 has quit IRC | 06:09 | |
| *** Guest63053 has joined #openstack-oslo | 06:18 | |
| *** Guest63053 has quit IRC | 06:18 | |
| *** sputnik13 has joined #openstack-oslo | 06:23 | |
| *** sputnik13 has quit IRC | 06:24 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary and/or non-optimized job priorities https://review.openstack.org/263512 | 06:30 |
| openstackgerrit | Einst Crazy proposed openstack/oslo.config: Replace assertEqual(*, None) with assertIsNone in tests https://review.openstack.org/263583 | 06:47 |
| *** jecarey has joined #openstack-oslo | 06:55 | |
| *** EinstCrazy has quit IRC | 06:56 | |
| *** EinstCrazy has joined #openstack-oslo | 06:56 | |
| *** jecarey has quit IRC | 07:00 | |
| *** vilobhmm11 has quit IRC | 07:03 | |
| openstackgerrit | Alexandre Viau proposed openstack-dev/pbr: exclude from files in _find_git_files https://review.openstack.org/263297 | 07:13 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/oslo.cache: Replace deprecated LOG.warn with warning https://review.openstack.org/263601 | 07:21 |
| *** ozamiatin_ has quit IRC | 07:23 | |
| *** ozamiatin_ has joined #openstack-oslo | 07:24 | |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/oslo.privsep: Replace deprecated LOG.warn with warning https://review.openstack.org/263605 | 07:28 |
| *** e0ne has joined #openstack-oslo | 07:43 | |
| *** salv-orlando has joined #openstack-oslo | 07:44 | |
| *** salv-orl_ has joined #openstack-oslo | 08:01 | |
| *** salv-orlando has quit IRC | 08:05 | |
| *** salv-orl_ has quit IRC | 08:05 | |
| *** markus_z has joined #openstack-oslo | 08:08 | |
| *** browne1 has quit IRC | 08:18 | |
| *** zakora has joined #openstack-oslo | 08:21 | |
| *** shardy has joined #openstack-oslo | 08:33 | |
| *** tobe has joined #openstack-oslo | 08:41 | |
| *** tobe has quit IRC | 08:41 | |
| mhorban | lxsli: nova-compute shows trace after reloading with SIGHUP. I proposed fixes https://review.openstack.org/#/c/259066/, https://review.openstack.org/#/c/258499/, https://review.openstack.org/#/c/258441/ to resolve that issue | 08:42 |
| openstackgerrit | ChangBo Guo(gcb) proposed openstack-dev/oslo-cookiecutter: Remove unused file openstack-common.conf https://review.openstack.org/263635 | 08:49 |
| *** jecarey has joined #openstack-oslo | 08:57 | |
| *** jecarey has quit IRC | 09:01 | |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/oslo.cache: Replace deprecated LOG.warn with LOG.warning https://review.openstack.org/263601 | 09:14 |
| *** jecarey has joined #openstack-oslo | 09:17 | |
| *** yassine__ has joined #openstack-oslo | 09:19 | |
| *** nihilifer has quit IRC | 09:20 | |
| *** nihilifer has joined #openstack-oslo | 09:21 | |
| *** jecarey has quit IRC | 09:22 | |
| *** zakora has quit IRC | 09:23 | |
| *** ndipanov has joined #openstack-oslo | 09:26 | |
| *** zakora has joined #openstack-oslo | 09:27 | |
| mhorban | Hey guys, could you please look at https://review.openstack.org/#/c/260384/ | 09:36 |
| *** ndipanov has quit IRC | 09:42 | |
| *** ndipanov has joined #openstack-oslo | 09:43 | |
| *** mhickey has joined #openstack-oslo | 09:58 | |
| *** mdbooth has quit IRC | 10:00 | |
| *** mdbooth has joined #openstack-oslo | 10:01 | |
| *** boris-42 has quit IRC | 10:03 | |
| *** mdbooth_ has joined #openstack-oslo | 10:09 | |
| *** yamahata has quit IRC | 10:10 | |
| *** mdbooth has quit IRC | 10:10 | |
| *** mdbooth_ is now known as mdbooth | 10:10 | |
| *** mdbooth has quit IRC | 10:19 | |
| *** mdbooth has joined #openstack-oslo | 10:27 | |
| *** mdbooth has quit IRC | 10:35 | |
| *** salv-orlando has joined #openstack-oslo | 10:44 | |
| *** dims has joined #openstack-oslo | 10:47 | |
| *** yamamoto has joined #openstack-oslo | 10:56 | |
| *** cdent has joined #openstack-oslo | 11:02 | |
| *** lucas-dinner is now known as lucasagomes | 11:03 | |
| *** yamamoto has quit IRC | 11:04 | |
| openstackgerrit | Zhihai Song proposed openstack/oslo.config: Make oslo-config-generator fail gracefully when no arguments https://review.openstack.org/262758 | 11:06 |
| *** salv-orlando has quit IRC | 11:13 | |
| openstackgerrit | Marian Horban proposed openstack/oslo.service: Removed double stopping of service on SIGHUP https://review.openstack.org/258499 | 11:14 |
| *** salv-orlando has joined #openstack-oslo | 11:14 | |
| *** salv-orlando has quit IRC | 11:16 | |
| *** salv-orlando has joined #openstack-oslo | 11:17 | |
| lxsli | mhorban: thanks for adding some info to 258499 but I was hoping for some line numbers | 11:17 |
| lxsli | I understand what double-stop means, I just don't see the conditions which cause it to happen | 11:17 |
| *** jecarey has joined #openstack-oslo | 11:18 | |
| lxsli | something like "when X and Y are true, we stop on L111 and L222" | 11:18 |
| *** salv-orlando has quit IRC | 11:18 | |
| *** salv-orlando has joined #openstack-oslo | 11:20 | |
| openstackgerrit | Marian Horban proposed openstack/oslo.service: Refactoring of tests/eventlet_service.py https://review.openstack.org/260384 | 11:21 |
| *** salv-orl_ has joined #openstack-oslo | 11:21 | |
| *** salv-orlando has quit IRC | 11:21 | |
| openstackgerrit | Julien Danjou proposed openstack/tooz: tests: do not hardcode /tmp https://review.openstack.org/263683 | 11:21 |
| *** jecarey has quit IRC | 11:23 | |
| *** salv-orl_ has quit IRC | 11:23 | |
| *** salv-orlando has joined #openstack-oslo | 11:23 | |
| mhorban | lxsli: We have double stop each time when we run service with Servicelauncher in daemon mode and send SIGHUP to it | 11:23 |
| *** salv-orlando has quit IRC | 11:24 | |
| mhorban | lxsli: Because we have one stop in finally block and other inside self.restart() | 11:24 |
| mhorban | lxsli: self.restart() consists of service.stop(); service.start() | 11:25 |
| *** dimtruck is now known as zz_dimtruck | 11:25 | |
| lxsli | but after we restart we loop | 11:25 |
| lxsli | it's in a while True | 11:25 |
| *** salv-orlando has joined #openstack-oslo | 11:26 | |
| mhorban | lxsli: It is loop . And natural reload is : stop-start-stop-start.... But we have stop-stop-start-stop-stop-start...Or what do you mean? | 11:28 |
| lxsli | the thread sits on line 280 most of the time | 11:28 |
| mhorban | agree | 11:28 |
| lxsli | if we get SIGUSR1 (any non-SIGHUP) we go to line 284 and restart | 11:29 |
| *** salv-orlando has quit IRC | 11:29 | |
| mhorban | actually inside super(ServiceLauncher, self).wait() (line 260) | 11:29 |
| lxsli | then we loop to 278 and sit on 280 again | 11:29 |
| *** salv-orlando has joined #openstack-oslo | 11:29 | |
| lxsli | oooh those methods are the other way around... give me a minute... | 11:30 |
| mhorban | if we got any NON-SIGHUP we do not loop but exit | 11:30 |
| *** salv-orlando has quit IRC | 11:30 | |
| lxsli | yes I get it now, somehow I thought _wait_for_exit_or_signal called wait, not the other way around | 11:31 |
| lxsli | +1 | 11:31 |
| mhorban | self.restart() (line 284) contains stop-start of services | 11:31 |
| *** salv-orlando has joined #openstack-oslo | 11:32 | |
| *** salv-orlando has quit IRC | 11:35 | |
| *** salv-orlando has joined #openstack-oslo | 11:36 | |
| openstackgerrit | Alexis Lee proposed openstack/oslo.log: Only format isotime if it's needed https://review.openstack.org/263692 | 11:42 |
| *** salv-orlando has quit IRC | 11:43 | |
| *** salv-orl_ has joined #openstack-oslo | 11:43 | |
| *** ericksonsantos has joined #openstack-oslo | 11:43 | |
| *** salv-orl_ has quit IRC | 11:46 | |
| *** salv-orlando has joined #openstack-oslo | 11:47 | |
| *** salv-orlando has quit IRC | 11:47 | |
| *** salv-orl_ has joined #openstack-oslo | 11:51 | |
| openstackgerrit | javeme proposed openstack/oslo.messaging: rabbit: Missing to pass parameter timeout to next https://review.openstack.org/263694 | 11:52 |
| *** salv-orl_ has quit IRC | 11:53 | |
| *** ihrachys has joined #openstack-oslo | 11:55 | |
| *** salv-orl_ has joined #openstack-oslo | 11:55 | |
| *** salv-orl_ has quit IRC | 11:56 | |
| *** salv-orlando has joined #openstack-oslo | 11:57 | |
| *** salv-orlando has quit IRC | 11:58 | |
| *** salv-orlando has joined #openstack-oslo | 12:04 | |
| *** zz_dimtruck is now known as dimtruck | 12:04 | |
| *** salv-orlando has quit IRC | 12:05 | |
| *** salv-orlando has joined #openstack-oslo | 12:10 | |
| *** salv-orlando has quit IRC | 12:12 | |
| *** salv-orl_ has joined #openstack-oslo | 12:12 | |
| *** mhickey has quit IRC | 12:12 | |
| *** salv-orl_ has quit IRC | 12:13 | |
| *** mhickey has joined #openstack-oslo | 12:13 | |
| *** mhickey has quit IRC | 12:15 | |
| *** mhickey has joined #openstack-oslo | 12:15 | |
| *** dimtruck is now known as zz_dimtruck | 12:15 | |
| *** mhickey is now known as mhickey_ | 12:20 | |
| *** mhickey_ has quit IRC | 12:23 | |
| *** mhickey has joined #openstack-oslo | 12:24 | |
| openstackgerrit | javeme proposed openstack/oslo.messaging: rabbit: fix unit conversion error of expiration https://review.openstack.org/263712 | 12:39 |
| *** salv-orlando has joined #openstack-oslo | 12:47 | |
| *** salv-orlando has quit IRC | 12:48 | |
| *** gcb has quit IRC | 12:48 | |
| openstackgerrit | Merged openstack/oslo.privsep: Updated from global requirements https://review.openstack.org/262915 | 12:49 |
| openstackgerrit | Merged openstack/tooz: utils: replace exception_message by exception_to_unicode https://review.openstack.org/263365 | 12:54 |
| *** zakora has quit IRC | 13:02 | |
| openstackgerrit | Merged openstack/oslo.messaging: Fix formatting of code blocks in zmq docs https://review.openstack.org/263482 | 13:06 |
| openstackgerrit | Merged openstack/taskflow: Use shared util helper for driver name + config extraction https://review.openstack.org/260761 | 13:07 |
| *** zz_dimtruck is now known as dimtruck | 13:07 | |
| *** links has quit IRC | 13:10 | |
| *** gordc has joined #openstack-oslo | 13:12 | |
| *** jecarey has joined #openstack-oslo | 13:19 | |
| *** dimtruck is now known as zz_dimtruck | 13:20 | |
| *** jecarey has quit IRC | 13:23 | |
| *** _amrith_ is now known as amrith | 13:24 | |
| *** edmondsw has joined #openstack-oslo | 13:27 | |
| *** pradk has joined #openstack-oslo | 13:27 | |
| *** pradk has quit IRC | 13:27 | |
| *** kgiusti has joined #openstack-oslo | 13:30 | |
| *** shardy has quit IRC | 13:47 | |
| lxsli | dhellmann: need to discuss 254821 when you can | 13:48 |
| *** shardy has joined #openstack-oslo | 13:49 | |
| *** lucasagomes is now known as lucas-hungry | 14:00 | |
| *** amotoki has quit IRC | 14:00 | |
| *** zz_dimtruck is now known as dimtruck | 14:04 | |
| *** rlrossit has joined #openstack-oslo | 14:11 | |
| *** ericksonsantos has quit IRC | 14:22 | |
| openstackgerrit | amrith proposed openstack/oslo.utils: Add a mechanism to mask passwords in dictionaries https://review.openstack.org/257561 | 14:22 |
| *** links has joined #openstack-oslo | 14:24 | |
| dims | dukhlov : are we ready to merge pika feature branch to master? | 14:24 |
| dims | ozamiatin : what can we do to fix the ceilometer failure in your zmq review? | 14:25 |
| ozamiatin_ | dims: recheck is probably enough | 14:27 |
| openstackgerrit | Monty Taylor proposed openstack-dev/pbr: Split changelog on nulls instead of ( https://review.openstack.org/263748 | 14:27 |
| dims | ozamiatin_ : ok, let's try that | 14:28 |
| dukhlov | dims, I think yes, but but before this we could merge current pending patch with unit tests and probably add ceilometer job for pika-driver | 14:29 |
| *** dimtruck is now known as zz_dimtruck | 14:29 | |
| dukhlov | dims: as far as I understood merge to master means just add new pika driver as optional - default driver remains kombu-driver, correct? | 14:30 |
| dims | dukhlov : it would be better to setup ceilometer job on master for pika driver rather than the feature branch (harder to do) | 14:31 |
| dims | dukhlov : yes | 14:31 |
| dukhlov | dims, ah ok | 14:31 |
| dims | dukhlov : we need to document how to switch from one to another if someone wants to | 14:32 |
| dukhlov | dims: yes, sure. should it be pika_driver.rst in solomessaging/doc? | 14:33 |
| dims | dukhlov : sure | 14:33 |
| openstackgerrit | Monty Taylor proposed openstack-dev/pbr: Split changelog on nulls instead of ( https://review.openstack.org/263748 | 14:33 |
| *** kbyrne has quit IRC | 14:33 | |
| *** ozamiatin_ has quit IRC | 14:34 | |
| dukhlov | dims, should I do it after merging to master or create patch in pika branch? | 14:34 |
| *** regXboi has joined #openstack-oslo | 14:34 | |
| *** ozamiatin_ has joined #openstack-oslo | 14:34 | |
| dims | dukhlov : which ever is easier | 14:34 |
| dukhlov | it would be easier to do it after merge to master if merge process will be fast. | 14:35 |
| dukhlov | dims: but if we need pass through review process and it may take more then two days, I will create patch to pika branch | 14:37 |
| dims | dukhlov : ok | 14:37 |
| *** kbyrne has joined #openstack-oslo | 14:40 | |
| *** jecarey has joined #openstack-oslo | 14:41 | |
| *** jecarey has quit IRC | 14:42 | |
| *** jecarey has joined #openstack-oslo | 14:43 | |
| *** jecarey_ has joined #openstack-oslo | 14:44 | |
| *** jecarey has quit IRC | 14:47 | |
| *** mriedem_away is now known as mriedem | 14:48 | |
| *** links has quit IRC | 14:49 | |
| openstackgerrit | Graham Hayes proposed openstack/debtcollector: Add updated_kwarg_default_value decorator https://review.openstack.org/255941 | 14:51 |
| ozamiatin_ | dims: one more patch from zmq side: https://review.openstack.org/#/c/261546/ | 14:51 |
| dims | ack will watch that too | 14:52 |
| ozamiatin_ | dims: Alexey is trying to add HA support for redis with sentinel (more options in the driver so on), need your advice what to do with 'inspect' package he used | 14:52 |
| *** kbyrne has quit IRC | 14:52 | |
| ozamiatin_ | dims: it is not in global requirements, should we add it there first? | 14:53 |
| dims | ozamiatin_ : which review? | 14:53 |
| *** zz_dimtruck is now known as dimtruck | 14:53 | |
| *** kbyrne has joined #openstack-oslo | 14:53 | |
| ozamiatin_ | dims: https://review.openstack.org/#/c/261546/ | 14:54 |
| dims | ozamiatin that "import inspect" is builtin to python itself | 14:54 |
| dims | so we are good | 14:54 |
| ozamiatin_ | dims: ah, ok | 14:55 |
| *** kbyrne has quit IRC | 14:55 | |
| *** zakora has joined #openstack-oslo | 14:59 | |
| *** kbyrne has joined #openstack-oslo | 15:00 | |
| *** kbyrne has quit IRC | 15:01 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 15:02 | |
| *** kbyrne has joined #openstack-oslo | 15:02 | |
| *** lucas-hungry is now known as lucasagomes | 15:05 | |
| *** shardy has quit IRC | 15:35 | |
| *** cdent has quit IRC | 15:42 | |
| dansmith | dims: what is the purpose of the stable/$foo branches on the oslo repos? | 15:42 |
| dansmith | dims: legacy the-script-creates-them perhaps? | 15:43 |
| dims | dansmith : the stable/liberty one is not used at all right now. we created them just like we used to...before we decided the policy change about new oslo libraries should work with all stable branches | 15:44 |
| *** shardy has joined #openstack-oslo | 15:44 | |
| dims | y | 15:44 |
| dansmith | dims: okay, the confusing thing is that we have those branches, but what we actually run with a given stable branch of, say, nova is actually what is in upper-constraints | 15:45 |
| dims | dansmith : right | 15:45 |
| dansmith | for o.vo, we were at 0.10 at liberty, but we're now running liberty in CI with 1.1.0 | 15:45 |
| dims | yep | 15:45 |
| dansmith | so our packaging guys were confused, and still using 0.10 | 15:45 |
| dims | dansmith : when we do mitaka, we will not create stable/ branches | 15:45 |
| dansmith | okay cool | 15:46 |
| dims | lifeless : where can we add some information to packagers? ^^^ | 15:46 |
| dansmith | yeah, my next question was going to be.. if this was documented somewhere I can point people to | 15:46 |
| dims | dansmith : https://review.openstack.org/#/c/226157/ is the spec | 15:48 |
| dansmith | dims: would it be crazy to delete the stable/liberty branch, and/or add a README_EOL.md file to it or something? | 15:48 |
| dims | dansmith : y good point, we should do something | 15:49 |
| *** pblaho has quit IRC | 15:53 | |
| *** cdent has joined #openstack-oslo | 15:53 | |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Remove deprecated use-syslog-rfc-format option https://review.openstack.org/263785 | 16:00 |
| *** mtanino has joined #openstack-oslo | 16:00 | |
| *** vilobhmm11 has joined #openstack-oslo | 16:03 | |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Added public method to getting default log levels https://review.openstack.org/263468 | 16:06 |
| openstackgerrit | Alexandre Viau proposed openstack-dev/pbr: exclude from files in _find_git_files https://review.openstack.org/263297 | 16:06 |
| *** pblaho has joined #openstack-oslo | 16:07 | |
| *** cdent has quit IRC | 16:09 | |
| openstackgerrit | Alexandre Viau proposed openstack-dev/pbr: exclude from files in _find_git_files https://review.openstack.org/263297 | 16:10 |
| openstackgerrit | Merged openstack/oslo.config: Replace assertEqual(*, None) with assertIsNone in tests https://review.openstack.org/263583 | 16:10 |
| *** pblaho has quit IRC | 16:10 | |
| *** pblaho has joined #openstack-oslo | 16:11 | |
| *** shakamunyi has joined #openstack-oslo | 16:14 | |
| *** barra204 has quit IRC | 16:15 | |
| *** pblaho has quit IRC | 16:17 | |
| *** pblaho has joined #openstack-oslo | 16:17 | |
| *** pblaho has quit IRC | 16:20 | |
| *** pblaho has joined #openstack-oslo | 16:20 | |
| *** nkrinner has quit IRC | 16:21 | |
| *** pblaho has quit IRC | 16:22 | |
| *** pblaho has joined #openstack-oslo | 16:22 | |
| *** pradk has joined #openstack-oslo | 16:26 | |
| mhickey | rlrossit: Hey. Thanks for the code for the temp pattern. My attempt was not working as expected! :) | 16:29 |
| *** cdent has joined #openstack-oslo | 16:32 | |
| *** harlowja_at_home has joined #openstack-oslo | 16:37 | |
| *** pblaho has quit IRC | 16:44 | |
| openstackgerrit | Martin Hickey proposed openstack/oslo.versionedobjects: There is a temp registry pattern [1] where you can backup the object registry, register a class locally, and then restore the original registry. This could be used for test objects that do not need to be registered permanently but will have calls which l https://review.openstack.org/263800 | 16:44 |
| *** pblaho has joined #openstack-oslo | 16:44 | |
| rlrossit | mhickey: glad I could help! | 16:45 |
| openstackgerrit | Martin Hickey proposed openstack/oslo.versionedobjects: Add temporary registry pattern to VersionedObjectRegistry https://review.openstack.org/263800 | 16:47 |
| mhickey | rlrossit: btw, I have just added a patch to wrap the temp registry pattern ^ | 16:49 |
| rlrossit | mhickey: neat! I'll take a look at it if I get some free time later | 16:50 |
| mhickey | rlrossit: not a bother. it might not be best way but feel free to take it apart! :) | 16:51 |
| *** zakora has quit IRC | 16:53 | |
| *** pblaho has quit IRC | 16:54 | |
| *** pblaho has joined #openstack-oslo | 16:54 | |
| *** markus_z has quit IRC | 16:55 | |
| *** ericksonsantos has joined #openstack-oslo | 16:55 | |
| *** pblaho has quit IRC | 16:56 | |
| *** pblaho has joined #openstack-oslo | 16:57 | |
| *** browne has joined #openstack-oslo | 16:59 | |
| *** ttx has quit IRC | 17:00 | |
| *** vilobhmm11 has quit IRC | 17:01 | |
| *** ttx has joined #openstack-oslo | 17:01 | |
| *** pblaho has quit IRC | 17:07 | |
| *** pblaho has joined #openstack-oslo | 17:07 | |
| lxsli | dhellmann: re: mutable opts, to make oslo.log mutable I need to offer a notify_oslo_config method which Nova must remember to call. If oslo.xxx which Nova uses adds config mutability, we'll need to patch Nova to support that. Is this OK, desirable or a developer-scaling problem please? | 17:09 |
| lxsli | tcammann suggested using a decorator on these methods but that's really just a syntax for a callback framework which I know you didn't want | 17:10 |
| *** mhickey has quit IRC | 17:10 | |
| *** e0ne has quit IRC | 17:11 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/oslo.utils: Add "configdrive" to the list of keys used by mask_password() https://review.openstack.org/263811 | 17:11 |
| dhellmann | lxsli : at the summit we talked about having the application code that triggers the config reload (probably a signal handler, but maybe an amqp message) do all of the work of resetting libraries when the config is reloaded by calling public functions in those libraries. Is that what you mean? | 17:19 |
| lxsli | yes that area | 17:20 |
| dhellmann | ok. I was hesitant to build any sort of callback system into oslo.config, because that makes initialization order of the libraries important. We did talk about doing that in oslo.service, I think, but I'm not sure how far we got. | 17:21 |
| dhellmann | so how about if we continue with the manual implementation and get something working, and then refine it? | 17:21 |
| lxsli | yeah sounds good | 17:21 |
| lxsli | I'd like to be able to say the reload order is not to be relied upon | 17:22 |
| lxsli | the other thing was your comment on https://review.openstack.org/#/c/251471/16/oslo_config/cfg.py , I responded on PS#17 but I'm not sure I understood you | 17:23 |
| dhellmann | lxsli : I'd like to avoid insisting on an order, too | 17:23 |
| dhellmann | lxsli : looking | 17:24 |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Remove deprecated use-syslog-rfc-format option https://review.openstack.org/263785 | 17:25 |
| dhellmann | lxsli : the case I'm worried about it loading a config file, reloading it, then registering a new option that is mutable -- I think, if I interpret your code right -- that would result in the mutable option using the initial value, rather than the reloaded value | 17:25 |
| dhellmann | so what I want to do is reload the entire file from which options can be pulled, and look for any existing options that are *not* mutable but that have changed, and drop those values (log an error and reset to the old value maybe?) | 17:26 |
| dhellmann | that way as new options are registered, they will get the current value. for immutable options, that's fine, since they won't have been used before. for mutable options, that's what we want. | 17:26 |
| lxsli | so values are held even if their options are unregistered? eugh | 17:27 |
| dhellmann | lxsli : yes, we parse the config file to build a database, and then as options are registered their values come from the database | 17:28 |
| harlowja_at_home | dims, the russian xmas is this week right (i am guessing u might remember) | 17:28 |
| *** dims has quit IRC | 17:29 | |
| harlowja_at_home | or was that last week, i can't remember, ha | 17:29 |
| dhellmann | lxsli : I forget what that class is called internally, maybe the namespace? Anyway, yeah, the 2 step process is used to avoid re-reading the file every time an option is registered and to allow an option class to do type conversion | 17:29 |
| dhellmann | harlowja_at_home : epiphany is this weekend, so I think this week? | 17:29 |
| lxsli | I think it's the namespace yeah | 17:29 |
| harlowja_at_home | dhellmann, thx i think so to :) | 17:30 |
| dhellmann | lxsli : does my concern make more sense now? | 17:30 |
| lxsli | yep | 17:30 |
| lxsli | there was some chatter about class structure with harlowja_at_home here btw: https://review.openstack.org/#/c/253125/ | 17:30 |
| harlowja_at_home | :) | 17:31 |
| lxsli | ah right yes it's the Parser that actually holds values which is just weird | 17:31 |
| harlowja_at_home | ya, up to u as to what u want to do there | 17:31 |
| harlowja_at_home | when i started doing some refactoring a long time ago, i started to split that up | 17:31 |
| dhellmann | oslo config is some of the oldest living code in openstack, so it's getting a little long in the tooth | 17:31 |
| harlowja_at_home | many moons ago | 17:32 |
| lxsli | I'd like to leave it alone and circle back after my feature is in | 17:32 |
| lxsli | fair enough :) | 17:32 |
| harlowja_at_home | https://review.openstack.org/#/c/134671/4/oslo/config/cfg.py lxsli | 17:32 |
| harlowja_at_home | ^ was some start of that | 17:32 |
| dhellmann | yeah, you may get farther instantiating a new ConfigOpts object and populating its list of registered options, having it read the file, and then comparing what it contains to what it is allowed to contain | 17:32 |
| dhellmann | rather than trying to refactor the underlying parts at this point, which we should do, but later | 17:33 |
| dhellmann | I have to run, bbiab | 17:33 |
| harlowja_at_home | does anyone know what/if systemd is doing any kind of config reloading thing as well (just curious) | 17:33 |
| harlowja_at_home | u'd start to think it might have some built-in feature for that to | 17:33 |
| lxsli | yeah the ConfigOpts-Namespace-Parser stack was what stopped me just making a new ConfigOpts in the first place :/ | 17:34 |
| lxsli | OK, I'm going home, thanks for the help! | 17:34 |
| elarson | harlowja_at_home: there is some daemon reload thing that sounds destructive, but isn't, that reloads things | 17:34 |
| *** shardy has quit IRC | 17:34 | |
| harlowja_at_home | intersting | 17:34 |
| harlowja_at_home | elarson, any docs on it anywhere? | 17:34 |
| harlowja_at_home | (if u know) | 17:35 |
| elarson | harlowja_at_home: I just used --help | 17:35 |
| *** dims has joined #openstack-oslo | 17:36 | |
| elarson | harlowja_at_home: systemctl daemon-reload I think | 17:36 |
| dims | harlowja : yep | 17:36 |
| *** shardy has joined #openstack-oslo | 17:36 | |
| openstackgerrit | Gauvain Pocentek proposed openstack/oslo.messaging: list_opts: update the notification options group https://review.openstack.org/263818 | 17:37 |
| harlowja_at_home | elarson, thx, will check it out | 17:40 |
| elarson | good luck! | 17:41 |
| harlowja_at_home | :) | 17:42 |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Added public method to getting default log levels https://review.openstack.org/263468 | 17:43 |
| *** yamahata has joined #openstack-oslo | 17:49 | |
| *** sputnik13 has joined #openstack-oslo | 17:50 | |
| *** e0ne has joined #openstack-oslo | 18:05 | |
| lifeless | dims: I thought we were still doing stable branches | 18:11 |
| lifeless | dims: my spec certainly doesn't do away with them | 18:11 |
| dims | lifeless : the confusion is that we are not currently making oslo lib releases from stable/liberty branch. what do packagers who are using latest stable/liberty for say nova use? | 18:13 |
| dims | (set of oslo libraries) | 18:13 |
| lifeless | dims: My understanding is that if they are using latest stable/liberty nova - e.g. unreleased stable, the stable maint team wants them to use latest stable/liberty oslo | 18:14 |
| dims | dansmith : ^^ | 18:15 |
| dims | mriedem : question for you too ^^ | 18:15 |
| lifeless | dims: *unit test* CI is not yet constrained in liberty, but we will be doing so, which will clamp it - we're also going to be adding jobs that test deliberately with latest | 18:15 |
| lifeless | so that we get both signal points | 18:15 |
| dansmith | I'm so confused | 18:15 |
| lifeless | dansmith: first step complete. Now for the lemur | 18:16 |
| dansmith | we're not currently testing stable/liberty nova against stable/liberty oslo libraries | 18:16 |
| lifeless | dansmith: For unit tests | 18:16 |
| lifeless | dansmith: devstack should be | 18:17 |
| dansmith | we are for devstack? | 18:17 |
| dansmith | because upper-constraints.txt shows 1.1.0 for o.vo | 18:17 |
| lifeless | dansmith: unless the stable team have landed changes to upper-constraints | 18:17 |
| dansmith | but the stable/liberty branch is muuuch older | 18:17 |
| lifeless | dansmith: if they have, they've effectively said that thats the known good version | 18:17 |
| dansmith | right, but now there is a conflict between the stale stable oslo branch and the reality of what we're using (and have declared as known-good) | 18:18 |
| lifeless | dansmith: https://review.openstack.org/#/c/226157/ is what I hope we'll be blessing in the TC meeting today | 18:18 |
| dansmith | lifeless: yeah, dims told me a couple hours ago that that review would effectively end the oslo library stable branches | 18:18 |
| dansmith | which made sense to me | 18:18 |
| dansmith | now I'm confused about what and how we're communicating to packagers | 18:19 |
| lifeless | dansmith: well, it actually says that due to packager an distributor feedback | 18:19 |
| lifeless | dansmith: that the spec *does not* end stable branches | 18:19 |
| mriedem | we don't test stable/liberty oslo libs against stable/liberty nova | 18:19 |
| lifeless | dansmith: it specifically sets up testing of both 'stable <-> stable' as well as 'latest <-> stable' | 18:19 |
| dims | dansmith : my data point was what's in stable/liberty upper constraints... | 18:19 |
| mriedem | we test stable/liberty nova against whatever released versions of oslo libs are available and fit the constraints for nova stable/liberty | 18:20 |
| mriedem | the only time we test branches for a lib are in the -src jobs | 18:20 |
| mriedem | as far as i know | 18:20 |
| lifeless | dansmith: obviously its not implemented yet | 18:20 |
| lifeless | dansmith: anyhow, it sounds like right now a) stable/liberty constraints is pointing at much newer code as known good | 18:20 |
| lifeless | dansmith: and things have bitrotten somehow such that that newer code is required ? | 18:21 |
| dansmith | so why are we not just creating branches for things like 0.10.x if we need to backport fixes to create an 0.10.1 if we plan to bump the upper-constraints to 0.10.1 for a given package? | 18:21 |
| dansmith | lifeless: tbh, I'm not sure what events led to bumping that version | 18:22 |
| lifeless | dansmith: which version of what ? | 18:22 |
| lifeless | o.vo ? | 18:22 |
| mriedem | i've also wondered why the libs don't have branches based on versions rather than just stable/x | 18:22 |
| dansmith | yeah, o.vo for stable/liberty | 18:22 |
| mriedem | but in the past the libs were just capped in stable anyway | 18:23 |
| lifeless | dansmith: ok, so branch naming - I have no idea. Its a good question. I can speculate - we're not supporting all released versions of libs, only the ones that matched major cycles from a stable team perspective | 18:23 |
| mriedem | so the next version of a stable branch lib was clear | 18:23 |
| lifeless | so going off of the name lets them get end of lifed without much thought | 18:23 |
| dansmith | lifeless: right, so we just create it when we need to create a .z | 18:23 |
| lifeless | dansmith: well no, we mass created stable/liberty branches of everything I thought | 18:24 |
| dims | yes, we mass created them | 18:24 |
| dansmith | lifeless: I know we do now, | 18:24 |
| lifeless | dansmith: ok, now I'm confused :) | 18:24 |
| dansmith | I mean I think we should be just creating an 0.N branch when we need to release a 0.N.1 fix for a given version we care about | 18:24 |
| lifeless | ok - so I'm saying I don't have a dog in that race :). But I'd guess mriedem and the stable team might, since figuring out which branches to delete is easy right now. | 18:25 |
| mriedem | would we need to delete 0.N.x branches? | 18:25 |
| * dansmith isn't sure why we ever need to delete branches | 18:25 | |
| dansmith | heh | 18:25 |
| lifeless | same reason we delete stable/juno etc | 18:26 |
| mriedem | btw, ^ is how i handle patching packages on older versions internally | 18:26 |
| dansmith | so, I'm completely lost in the requirements repo trying to find out what the history of the o.vo bumps were | 18:26 |
| dansmith | lifeless: I don't really see much point in doing that, FWIW | 18:26 |
| mriedem | i create a branch for the target minor version i need, patch it and produce a .z package | 18:26 |
| dansmith | back to the original problem though, | 18:27 |
| mriedem | lifeless: we delete stable/juno b/c there are jobs tied to those | 18:27 |
| lifeless | dansmith: sure, I'm at best ambivalent on the deletion thing, but its the current norm | 18:27 |
| dansmith | our packagers were just blindly packaging the stable/$release version of everything, which turns out to be completely wrong | 18:27 |
| dansmith | which might be because we're not keeping them up to date with the changes we make to the upper constraints, | 18:27 |
| dansmith | but that seems like a real liability for issues to me | 18:27 |
| mriedem | heh, yeah | 18:27 |
| dansmith | doesn't seem like a given setup.py::version should ever exist in multiple branches | 18:27 |
| dansmith | that's just crazy confusing | 18:28 |
| dansmith | and is never going to be kept right by the humans | 18:28 |
| mriedem | for legal reasons once we GA'ed a release, we had to jump through hoops to ship new minor versions of rpms, so we would just patch the GA'ed versions as needed | 18:28 |
| mriedem | and bump the rev number on the package | 18:28 |
| lifeless | 2217bb7f (OpenStack Proposal Bot 2015-12-13 06:35:52 +0000 203) oslo.versionedobjects===1.1.0 | 18:28 |
| mriedem | ^ meant that you had to explicitly know why you needed to patch something | 18:28 |
| dansmith | lifeless: right, that's not a helpful item of history I think | 18:29 |
| lifeless | well, it took us from 0.12.0 | 18:29 |
| mriedem | yeah, it's just a bot https://review.openstack.org/#/c/246211/ | 18:29 |
| lifeless | its who approved it thats more interesting | 18:29 |
| lifeless | flavio and theirry | 18:30 |
| dansmith | lifeless: but really, lots of changes in there... how are they supposed to vet them all? | 18:30 |
| mriedem | Thierry CarrezDec 14 9:43 AM↩Patch Set 25: Code-Review+2 Workflow+1Living dangerously | 18:30 |
| dansmith | and the proposal bot changes are just auto-+Wd elsewhere | 18:30 |
| mriedem | they saw jenkins was green | 18:30 |
| lifeless | dansmith: its meant to be CI tested; the issue is that stable branch policy is different to master on this I suspect | 18:31 |
| mriedem | the problem is the jobs on the generate-constraints changes are a subset of everything that's tested, a lot of which isn't in tempest runs | 18:31 |
| lifeless | dansmith: and - the backwards compat spec specifically takes a copy of upper-constraints for testing of the at-time-of-release libraries | 18:31 |
| mriedem | there is no stable branch policy for the upper-constraints bot changes | 18:31 |
| mriedem | i've just seen people approving them as they see them passing jenkins | 18:32 |
| dansmith | mriedem: right, per normal behavior I imagine | 18:32 |
| lifeless | thats the policy for master; Idon't think we actually sat down and talked about policy on stable for it | 18:32 |
| lifeless | bbs, ECHILD. | 18:32 |
| mriedem | changes to global-requiremnets on stable are generally much more thought out | 18:32 |
| mriedem | since a human is proposing them | 18:32 |
| mriedem | well, assuming sdague is human | 18:33 |
| dansmith | right, which makes sense, but having to keep the stable/$release branch exactly sync'd with the code in the version that we put in the constraint on stable seems highly failure-prone to me | 18:33 |
| mriedem | that hurts my brain | 18:34 |
| lifeless | so thats why we're going to have two files | 18:35 |
| lifeless | upper-constraints will be latest known good, released-constraints will be stable/liberty versions and hand curated | 18:35 |
| dansmith | um | 18:35 |
| mriedem | latest known good* where *is based on what test coverage we have | 18:36 |
| mriedem | so g-r has the min, | 18:36 |
| lifeless | mriedem: yes | 18:36 |
| mriedem | released-constraints has the max at the time of release, | 18:37 |
| mriedem | and u-c is the max today | 18:37 |
| lifeless | line 204 in https://review.openstack.org/#/c/226157/12/specs/backwards-compat-libs-clients.rst | 18:37 |
| mriedem | and you test new changes against released-constraints to make sure you don't break people that would have fall <= that version | 18:37 |
| dansmith | so that's three points.. the minimum we think will work, the maximum we think will work and .. a third one | 18:37 |
| lifeless | we may want to have a separate bot run that only accepts changes to non-openstack projects, or some such, targeting release-constraints | 18:37 |
| lifeless | mriedem: right. we want to know that nova changes in stable work with the set of libraries it was released with. we also want to know it works with the current highest version folk can run | 18:38 |
| mriedem | "Then we test with that rather than 'upper-constraints.txt'" | 18:39 |
| mriedem | that == release-constraints which is the max at the time of release | 18:39 |
| mriedem | which is another way of doing what we did with capping previous stable branches in g-r | 18:39 |
| lifeless | yes | 18:39 |
| mriedem | except we f'ed that up a few times | 18:39 |
| dansmith | so to be honest, | 18:40 |
| lifeless | yes, because the capping is basically a distributed database and thats hard :) | 18:40 |
| dansmith | I know I'm not a smart person, but I really don't understand how I'm going to explain this to our packaging people :/ | 18:40 |
| mriedem | well, the capping failed for 2 big reasons | 18:40 |
| lifeless | dansmith: what are they trying to do ? | 18:40 |
| lifeless | dansmith: and what got you coming down this path ? | 18:41 |
| dansmith | lifeless: they're trying to know what to ship and when. right now I can tell them that it should be the version in upper-constraints because that's what is tested, but going forward I don't think I can tell them how to decipher things for themselves | 18:41 |
| dansmith | lifeless: what got me here is that o.vo's stable/liberty is not what is in upper-constraints, not what is tested, and is missing a couple fixes to make it work with liberty nova at all | 18:42 |
| lifeless | dansmith: ok, so I think going forward we're going to want release-constraints.txt to be the golden rule for stable branches. | 18:42 |
| lifeless | dansmith: known good, branch specific | 18:43 |
| lifeless | dansmith: (or, upper-constraints, for less risk-averse distributors) | 18:43 |
| dansmith | that would be fine with me, but keeping the stable/$release branch in sync with that file is what concerns me | 18:43 |
| dansmith | and because I don't see the point in doing that | 18:43 |
| lifeless | dansmith: basically we're going to have two known-good sets: one with latest versions of deps, and one with stable release versions of deps | 18:43 |
| dansmith | yeah, I get that | 18:44 |
| lifeless | dansmith: and we're going to have to have that to be able to test stable changes for backwards compat issues | 18:44 |
| lifeless | oop, ECHILD again - sorry, its breakfast time etc | 18:45 |
| mriedem | so rather than knowing what your cap is in the project's own requirements.txt, you have to check in the release-constraints.txt file in another repo | 18:46 |
| dansmith | dims: what are your thoughts on all this? | 18:46 |
| mriedem | i've commented again in https://review.openstack.org/#/c/226157/12/specs/backwards-compat-libs-clients.rst | 18:46 |
| dansmith | I would guess that mriedem and dims need to have a high level of comfort in whatever solution we come up with | 18:46 |
| *** ihrachys has quit IRC | 18:47 | |
| mriedem | i'm not really comfortable with this :) i've expressed that a few times, but people want what people want | 18:47 |
| dims | dansmith : i agree that if we have stable branches then they should work with the stable release! | 18:47 |
| mriedem | i think it makes packaging more confusing, at least on stable | 18:48 |
| mriedem | i think it makes testing more confusing on stable | 18:48 |
| dims | dansmith : so we seem to end up with 2 sets of libraries (one set of say oslo libs from master and another set reflecting oslo libs from stable branches) | 18:48 |
| dims | and both should work | 18:49 |
| dims | don't know if we'll end up there | 18:49 |
| dansmith | yeah, I don't feel like we're on a path to end up with that | 18:49 |
| mriedem | if i'm a distro, i don't really know why i'd want to pick up oslo libs from master when i'm supporting a kilo release | 18:49 |
| mriedem | or liberty | 18:49 |
| mriedem | b/c that exposes me to knew untested territory | 18:50 |
| dims | dansmith : the other way to look at it is that oslo libs are just like other libs from pypi so we don't have stable branches | 18:50 |
| mriedem | when all i want on stable fix pack releases is bug fixes | 18:50 |
| dansmith | I do feel like we're on a path to end up with some very confusing failures because the version in one of the 12 requirements files reflects slightly different code than what is in the stable branch for that repo | 18:50 |
| dansmith | dims: right, unless we need to release a .z for a specific reason, right? | 18:50 |
| dims | right | 18:50 |
| dansmith | dims: and then we create a branch for it, fix it, and then release the .z | 18:50 |
| dansmith | dims: thats +1000 from me | 18:50 |
| *** vilobhmm11 has joined #openstack-oslo | 18:51 | |
| * dhellmann notes the very long scrollback | 18:52 | |
| dims | dhellmann : so we started off by dansmith exploring what their packagers should use for o.vo with stable/liberty nova | 18:53 |
| dims | dhellmann : stable/liberty branch of o.vo will not work with stable/liberty nova | 18:53 |
| dhellmann | I see a question about branch naming, I can answer that: Branches are tested with things with the same name, or master if nothing else with that name exists. So if we had stable/0.10 of something it would be tested against master instead of whatever stable series it was part of | 18:54 |
| dhellmann | the other reason for using series names is to only need to support one stable branch for a given series at a time | 18:54 |
| dhellmann | same as the servers | 18:54 |
| openstackgerrit | Merged openstack/oslo.i18n: add versionadded designations to newer functions https://review.openstack.org/262022 | 18:54 |
| dansmith | dhellmann: well, that's demonstrably not how it is today | 18:54 |
| dansmith | dhellmann: and not how lifeless is describing the future, AFAICT | 18:54 |
| dhellmann | dansmith : ok, that was the intent, I'm not sure how we got away from that | 18:55 |
| dansmith | dhellmann: keeping the stable branch correct with what the stable branches are tested against would require humans | 18:55 |
| dhellmann | I'm not sure how the future will be different. I don't think we want to support a bunch of separate stable branches though | 18:55 |
| *** rlrossit has quit IRC | 18:56 | |
| dhellmann | dansmith : the devstack gate machinery that checks out code does the testing. every cycle we have to add a name to the list of things it knows how to check out | 18:56 |
| dhellmann | if that has broken down, I wonder if we don't have the right jobs defined against the stable branch of the lib somehow | 18:56 |
| *** rlrossit has joined #openstack-oslo | 18:56 | |
| dansmith | dhellmann: but today we test nova from oslo packages from pip on stable and master, not from the branches of those projects | 18:56 |
| dansmith | dhellmann: the reason it's not right today is that we're using the pip version of the package from upper-constraints, which does not match what is actually in stable/liberty | 18:57 |
| dhellmann | dansmith : right, but the constraints should be set to use the released versions of things from the branches for nova and changes to oslo.vo stable/foo are tested against nova's stable/foo | 18:57 |
| dhellmann | dansmith : ok, so that's where we went wrong, that shouldn't have been allowed | 18:57 |
| dhellmann | we should dial that back down | 18:57 |
| dansmith | dhellmann: that's how it's been for a long time, AFAIK | 18:57 |
| dansmith | dhellmann: the only time we test (say) nova against o.vo's master is in the o.vo test with a different job.. nova itself is tested against the released version, | 18:58 |
| dansmith | which means we don't get things like the upgrade jobs using the branches, only the packages | 18:58 |
| dhellmann | the upgrade jobs should be defined to run against changes in the libraries, right? | 18:59 |
| dansmith | and same for all the other ones, AFAIK | 18:59 |
| dhellmann | but you're right about where the source comes from in both cases | 18:59 |
| dansmith | dhellmann: sure, we could, that's a different topic I'm just pointing out how the tests are run now and from what, yeah | 19:00 |
| dhellmann | ah, the constraints bot updated that entry for us | 19:00 |
| dhellmann | nice | 19:00 |
| dhellmann | so we're in this weird transition period between the capping we used to do and the "no stable branches" era lifeless wants to usher in | 19:00 |
| dhellmann | where we're not capping, but we have stable branches, but they don't match | 19:00 |
| dhellmann | :-( | 19:00 |
| dansmith | I know we are, and that's understandable | 19:01 |
| dansmith | what I care about is the future, and this actually getting better | 19:01 |
| dhellmann | oh, I'm just talking outloud while I come up to speed on the existing conversation | 19:01 |
| dansmith | what I'm concerned about is that we won't actually fix it | 19:01 |
| *** rlrossit has quit IRC | 19:01 | |
| *** rlrossit has joined #openstack-oslo | 19:01 | |
| dhellmann | well, lifeless' spec is up for discussion at the tc meeting in about an hour. after that, I don't know exactly what has to happen off the top of my head, unfortunately. | 19:01 |
| *** rlrossit has quit IRC | 19:02 | |
| dhellmann | if we go ahead with his plan, I'm not sure how we clearly indicate to anyone packaging the stable branch what version of a lib they should use -- I don't recall if that was covered in the spec | 19:02 |
| dhellmann | that's https://review.openstack.org/#/c/226157/ for reference | 19:02 |
| mriedem | dhellmann: basically i'd recommend to packagers that they use the proposed release-constraints.txt file as a cap | 19:02 |
| dansmith | I feel like in lifeless' post-apocalyptic future, we can still have this case where we bump the version in some requirements file that we're testing as the bottom end of something for stable/liberty, | 19:03 |
| dansmith | and that will not match what is in that branch for the oslo library | 19:03 |
| dansmith | basically, exactly what has happened here with o.vo | 19:03 |
| dhellmann | mriedem : ah, right, I forgot about release-constraints.txt, that's how we're supposed to do it | 19:03 |
| mriedem | dhellmann: which i'd argue is just another way of capping the requirements in g-r | 19:03 |
| mriedem | but fancier | 19:03 |
| mriedem | and more complicated | 19:03 |
| dhellmann | mriedem : I think it's less complicated to change the cap, but more files to keep track of | 19:04 |
| *** rlrossit has joined #openstack-oslo | 19:04 | |
| *** rlrossit has quit IRC | 19:04 | |
| dhellmann | that is, I think we don't run into the wedging issues using the constraints system that we had with upper bounds | 19:04 |
| mriedem | in release-constraints or g-r? | 19:04 |
| *** rlrossit has joined #openstack-oslo | 19:04 | |
| openstackgerrit | amrith proposed openstack/oslo.utils: Add a mechanism to mask passwords in dictionaries https://review.openstack.org/257561 | 19:04 |
| dhellmann | in release-constraints | 19:04 |
| dhellmann | dansmith : again, I'm still catching up: are you familiar with that spec I linked? | 19:04 |
| dims | mriedem : dhellmann : release-constraints reflect current stable branches? | 19:04 |
| dhellmann | it sounds like you'd have good input on it | 19:04 |
| dhellmann | dims : IIRC, that's the intent | 19:05 |
| mriedem | dims: release-constraints is meant to be upper-constraints (a snapshot of) at the time of the releease | 19:05 |
| dansmith | dhellmann: I'm familiar with it as described by folks here.. I haven't read/digested it myself and I won't by the time the TC meeting rolls around | 19:05 |
| dims | dhellmann : so we do end up with 2 sets of working libs (one set of oslo libs from master and one set of oslo libs from stable branches) | 19:05 |
| mriedem | dims: so g-r stable/liberty has the min when liberty was released, | 19:05 |
| *** rlrossit_ has joined #openstack-oslo | 19:05 | |
| mriedem | release-constraints in stable/liberty would be u-c at the time of liberty release | 19:05 |
| dhellmann | dansmith : sure. if you can be around to discuss it in the meeting that would help, but I'll raise the concern if you can't | 19:05 |
| mriedem | and u-c stable/liberty is whatever is bleeding edge and passes tempest today | 19:05 |
| dims | mriedem : that does not imply release-constraints tracks stable branches or is even working against it | 19:06 |
| dhellmann | mriedem : I would expect release-constraints to be updated if we backport fixes to older versions of the libs, is that what the spec says? | 19:06 |
| dansmith | dhellmann: that's the part that requires humans that concerns me | 19:07 |
| mriedem | dhellmann: the spec doesn't say when we'd raise caps in release-constraints on stable | 19:07 |
| dhellmann | dansmith : yeah, we'd want to write some validation to ensure only patch-release changes can be made to that file | 19:07 |
| dhellmann | mriedem : ok, I can't remember if it's meant to be static or not, maybe it is | 19:08 |
| lifeless | dansmith: sp, to bump a version in a requirements file | 19:08 |
| lifeless | dansmith: that gets reviewed by stable folk | 19:08 |
| lifeless | dansmith: the constraints files are just points within the requirements space to test at | 19:09 |
| *** rlrossit has quit IRC | 19:09 | |
| dansmith | lifeless: yeah, I get that | 19:09 |
| lifeless | dansmith: the mismatch we have at the momemnt is because a) we only test one point and b) humans approved a change that isn't in line with policy | 19:09 |
| dansmith | lifeless: yeah, I get that too | 19:09 |
| dansmith | the current situation is fully understandable | 19:09 |
| lifeless | dansmith: once we have two points, and no bot suggesting changes to the release-branch one | 19:09 |
| lifeless | dansmith: it seems like humans would have to go out of their way to cause a policy break ? | 19:10 |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Remove deprecated use-syslog-rfc-format option https://review.openstack.org/263785 | 19:11 |
| dims | lifeless : do the versions in the release-constraints reflect the stable/ branch of the library? | 19:11 |
| dansmith | lifeless: because they propose lifting the release version for something that isn't actually in the stable version? sounds like a thing a human could do pretty easily | 19:11 |
| dansmith | writing lots of checks to make sure we don't let humans do bad things will obviously eventually end up working, | 19:11 |
| *** harlowja_at_home has quit IRC | 19:11 | |
| dansmith | but this just seems like a large amount of complexity | 19:11 |
| dhellmann | lifeless : have we already turned off the bot that is proposing constraints changes to stable branches? | 19:12 |
| lifeless | dansmith: how are you measuring the complexity here? like - this has way fewer things in the air at once than actual capping | 19:12 |
| dansmith | lifeless: really? | 19:13 |
| dansmith | I don't understand that at all, so clearly I don't belong in this conversation | 19:13 |
| lifeless | dansmith: really. to do capping we have to do an expand-contract pattern across a hundred odd repositories at every release | 19:13 |
| dansmith | dims: what could I do to catch up the stable branch? do I have to backport everything between 0.10 and 1.1.0 and land them? | 19:13 |
| lifeless | dansmith: we didn't manage to do it right even *once* | 19:14 |
| *** harlowja_at_home has joined #openstack-oslo | 19:14 | |
| mriedem | the caps in stable/kilo have been much less of a problem than they were in icehouse and juno | 19:14 |
| lifeless | dansmith: see e.g. tony's mails about shared versions between releases, and the caps in those things break if you install the libs directly because they downgrade other componennts | 19:14 |
| dims | dansmith : if we do something we should do it for all the oslo libs and not just one. so don't know yet. need to think a bit more | 19:14 |
| mriedem | lifeless: right, the shared versions were really between juno and kilo | 19:15 |
| dansmith | dims: okay, I need to go back to my packagers and tell them to forget what I told them this morning, but they're going to need to do something about o.vo right away | 19:15 |
| mriedem | i remember unwinding that issue myself personally | 19:15 |
| lifeless | mriedem: fun times | 19:15 |
| dansmith | dims: so either I need to tell them to do a one-off for o.vo and hold on for the rest or ... something | 19:15 |
| dims | dansmith : understood :( | 19:15 |
| mriedem | a big reason for shared versions between stable branches was because back in juno/kilo, the PTLs were releasing libs/clients themselves at whatever version they thought was appropriate | 19:16 |
| mriedem | which usually mean not bumping minor versions | 19:16 |
| mriedem | the releases process has fixed a lot of that | 19:16 |
| lifeless | mriedem: also we just didn't release some things | 19:16 |
| lifeless | mriedem: capping means 'must release every time' because of the distributed redundancy in the caps | 19:16 |
| *** jschlueter has joined #openstack-oslo | 19:17 | |
| mriedem | i get that capping the first time is a pain in the ass | 19:17 |
| lifeless | we could do a pure semver thing everywhere, but distributors have pushed back because they don't trust new releases to actually not be subtly broken | 19:17 |
| mriedem | who are distributors? | 19:18 |
| lifeless | mriedem: helion, ibm, rackspace, redhat etc | 19:18 |
| mriedem | so i'm the only packaging guy ibm ever really had for openstack | 19:18 |
| lifeless | mriedem: I think I've had someone from all of them say 'please keep stable branches and don't require use to run latest releases ever' | 19:18 |
| mriedem | so i'm not sure what ibm is saying about this | 19:18 |
| lifeless | mriedem: Well, its your apparent needs I'm going off of :) | 19:19 |
| dims | mriedem : lifeless : my one thing here is that release-constraints should reflect a library version released from a stable branch..... or no stable branch exists for that library | 19:19 |
| lifeless | mriedem: in that, AIUI, you want stable branches | 19:19 |
| mriedem | sure, i also mostly want caps | 19:19 |
| mriedem | i want my stable/kilo nova to work with the versions of it's dependencies that were out there at the time of kilo GA | 19:20 |
| dansmith | mriedem: you prefer stable branches of stable/$release or 0.X.Y? | 19:20 |
| lifeless | mriedem: so caps completely stop us testing for backwards compat, for instance. I don't think you get how intrusive they are | 19:20 |
| mriedem | i don't care that nova kilo could pick up oslo.utils from mitaka | 19:20 |
| mriedem | caps allow libs to break backward compat with major version bumps | 19:20 |
| lifeless | mriedem: I too want stable/kilo nova to work in that case | 19:21 |
| dansmith | lifeless: I can't believe you just said that to mriedem | 19:21 |
| mriedem | dims' life would be a lot easier if we capped on stable so oslo libs could drop things in major versions | 19:21 |
| dims | ++ mriedem | 19:21 |
| lifeless | mriedem: but we can drop things in major versions | 19:21 |
| mriedem | we can't w/o the caps | 19:21 |
| mriedem | or until -eol | 19:21 |
| lifeless | dansmith: sorry, why? | 19:21 |
| mriedem | i assume for my lack of knowledge on gate wedge issues on stable | 19:22 |
| mriedem | i have some experience there | 19:22 |
| dansmith | lifeless: because it's mriedem.. I think he's pretty well versed in the actual mechanics of fixing these issues | 19:22 |
| lifeless | mriedem: that is true. One part of the problem is that caps presume all breaks will affect all users | 19:22 |
| mriedem | i don't get that statement | 19:23 |
| lifeless | dansmith: if nova liberty caps on o.vo < 1, and we're releasing 1.2 of o.vo, how can we test it with nova liberty? | 19:24 |
| mriedem | e.g. if nova kilo works with let's say lib foo at 1.2.1, we cap foo<1.3 in kilo g-r b/c we assume foo follows semver | 19:24 |
| lifeless | dansmith: we don't know its breaking nova liberty - we might be deleting something that was deprecated in juno | 19:24 |
| mriedem | and won't release backward incompatible changes in 1.2.2 | 19:24 |
| dansmith | lifeless: by pinning nova liberty at a version? | 19:24 |
| mriedem | lifeless: why would we want to test o.vo 1.2 with nova liberty if it's capped | 19:24 |
| mriedem | ? | 19:24 |
| dansmith | yeah, that | 19:24 |
| lifeless | mriedem: so that we know if we're breaking things we didn't mean to? | 19:25 |
| dansmith | the reason we actually broke the released version was because we had to make stable/liberty work with 0.10 and 1.1.0, IIRC | 19:25 |
| mriedem | if some packagers nulls out the nova liberty requirements.txt and wants to ship o.vo 1.2, that's their problem | 19:25 |
| lifeless | ECHILD, sorry but this is basically the worst time of day for discussion for me | 19:25 |
| mriedem | let's break, i want to get coffee before the TC meeting | 19:26 |
| mriedem | in 34 min | 19:26 |
| * dansmith looks for a rusty fork | 19:26 | |
| *** salv-orlando has joined #openstack-oslo | 19:33 | |
| *** salv-orlando has quit IRC | 19:38 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary and/or non-optimized job priorities (WIP) https://review.openstack.org/263512 | 19:41 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary and/or non-optimized job priorities (WIP) https://review.openstack.org/263512 | 19:52 |
| *** devananda has quit IRC | 19:54 | |
| *** harlowja_at_home has quit IRC | 19:54 | |
| *** lucasagomes is now known as lucas-dinner | 19:59 | |
| *** devananda has joined #openstack-oslo | 20:00 | |
| openstackgerrit | Gauvain Pocentek proposed openstack/oslo.messaging: list_opts: update the notification options group https://review.openstack.org/263818 | 20:01 |
| lifeless | and back | 20:02 |
| *** yassine__ has quit IRC | 20:15 | |
| *** e0ne has quit IRC | 20:28 | |
| openstackgerrit | Ronald Bradford proposed openstack/oslo.log: Remove deprecated log-format option https://review.openstack.org/263903 | 20:57 |
| *** kgiusti has left #openstack-oslo | 21:18 | |
| *** zzzeek has quit IRC | 21:22 | |
| *** zzzeek has joined #openstack-oslo | 21:22 | |
| *** rockyg has joined #openstack-oslo | 21:39 | |
| *** stpierre has joined #openstack-oslo | 21:55 | |
| stpierre | can someone help me understand the oslo (specifically oslo.log) release cycle? it seems like it's mostly separate from the openstack release cycle, but i'm confused about the purpose of the stable/* branches (e.g., stable/liberty, stable/kilo, etc.) | 21:56 |
| *** pratikmallya has joined #openstack-oslo | 21:57 | |
| rockyg | The oslo releases *are* separate from the integrated release cycle. They are done on an as needed basis. But, the stable branches are there because when the integrated release happens, it works with the current (and sometimes earlier) versions of the lib, but may break on later ones | 21:59 |
| rockyg | So, the stable branch has a "guaranteed to work with" version of the library | 21:59 |
| *** vilobhmm11 has quit IRC | 22:00 | |
| *** vilobhmm11 has joined #openstack-oslo | 22:01 | |
| *** vilobhmm11 has quit IRC | 22:01 | |
| rockyg | stpierre ^^ | 22:01 |
| *** vilobhmm11 has joined #openstack-oslo | 22:02 | |
| stpierre | ok, that makes sense | 22:02 |
| *** vilobhmm11 has quit IRC | 22:02 | |
| *** vilobhmm11 has joined #openstack-oslo | 22:02 | |
| stpierre | does the stable branch ever change? are fixes backported to it? does it follow the latest release unless there's a known breaking change? | 22:02 |
| *** vilobhmm11 has quit IRC | 22:03 | |
| *** vilobhmm11 has joined #openstack-oslo | 22:03 | |
| rockyg | Yes. fixes are backported if needed (and someone does it) | 22:03 |
| rockyg | Uh, the second point...I *think it isn't capped until there is breakage, but not positive. There has been a lot of churn trying to get various libraries and packages to not break over time. I believe that going forward, from Liberty, on, there will be caps on all requirements | 22:05 |
| rockyg | The person to ask definitively on this is lifelfess. He's the guru/expert. dims and dhellmann also know the specifics. | 22:06 |
| *** vilobhmm11 has quit IRC | 22:06 | |
| *** vilobhmm11 has joined #openstack-oslo | 22:06 | |
| *** vilobhmm11 has quit IRC | 22:07 | |
| *** vilobhmm11 has joined #openstack-oslo | 22:07 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Allow for alterations in decider 'area of influence' https://review.openstack.org/246051 | 22:09 |
| *** harlowja has quit IRC | 22:09 | |
| *** harlowja has joined #openstack-oslo | 22:10 | |
| *** pratikmallya has quit IRC | 22:14 | |
| *** ndipanov has quit IRC | 22:19 | |
| *** boris-42 has joined #openstack-oslo | 22:20 | |
| *** dims has quit IRC | 22:20 | |
| *** regXboi has quit IRC | 22:22 | |
| *** rlrossit_ has quit IRC | 22:23 | |
| *** dims has joined #openstack-oslo | 22:25 | |
| openstackgerrit | Eric Brown proposed openstack/oslo.config: Numerous corrections to the docstrings https://review.openstack.org/263934 | 22:32 |
| *** edmondsw has quit IRC | 22:33 | |
| *** stpierre has quit IRC | 22:34 | |
| *** edmondsw has joined #openstack-oslo | 22:37 | |
| *** edmondsw has quit IRC | 22:37 | |
| *** cdent has quit IRC | 22:41 | |
| *** jecarey_ has quit IRC | 22:45 | |
| *** jecarey has joined #openstack-oslo | 22:46 | |
| *** jecarey_ has joined #openstack-oslo | 22:49 | |
| *** jecarey has quit IRC | 22:52 | |
| *** jecarey_ has quit IRC | 22:53 | |
| *** zqfan has quit IRC | 23:01 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Add rundimentary and/or non-optimized job priorities https://review.openstack.org/263512 | 23:05 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Use newly minted notifier library and deprecate existing and alias to new https://review.openstack.org/256130 | 23:15 |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: Replace clear zookeeper python with clear zookeeper bash https://review.openstack.org/196528 | 23:18 |
| *** dims has quit IRC | 23:20 | |
| *** dims has joined #openstack-oslo | 23:25 | |
| *** dimtruck is now known as zz_dimtruck | 23:29 | |
| *** shardy has quit IRC | 23:30 | |
| *** gordc has quit IRC | 23:33 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 23:35 | |
| *** mriedem has quit IRC | 23:37 | |
| openstackgerrit | Joshua Harlow proposed openstack/taskflow: For taskflow patterns don't show taskflow.patterns prefix https://review.openstack.org/257538 | 23:38 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!