*** tetsuro has joined #openstack-placement | 00:30 | |
*** tetsuro has quit IRC | 00:32 | |
*** tetsuro_ has joined #openstack-placement | 00:32 | |
*** tetsuro_ has quit IRC | 00:40 | |
*** tetsuro has joined #openstack-placement | 00:53 | |
openstackgerrit | Takashi NATSUME proposed openstack/os-resource-classes master: Update author-email in setup.cfg https://review.openstack.org/624885 | 01:38 |
---|---|---|
*** tetsuro has quit IRC | 03:30 | |
*** tetsuro has joined #openstack-placement | 04:25 | |
*** tetsuro has quit IRC | 05:36 | |
*** tetsuro has joined #openstack-placement | 06:53 | |
*** tetsuro has quit IRC | 07:33 | |
*** e0ne has joined #openstack-placement | 07:35 | |
*** e0ne has quit IRC | 07:39 | |
*** helenafm has joined #openstack-placement | 08:08 | |
*** takashin has left #openstack-placement | 08:31 | |
*** gibi has joined #openstack-placement | 08:52 | |
*** ttsiouts has joined #openstack-placement | 09:34 | |
openstackgerrit | Chris Dent proposed openstack/placement master: WIP: Use os-resource-classes in placement https://review.openstack.org/623556 | 09:40 |
openstackgerrit | Chris Dent proposed openstack/placement master: WIP: Use os-resource-classes in placement https://review.openstack.org/623556 | 10:02 |
*** e0ne has joined #openstack-placement | 10:02 | |
*** cdent has joined #openstack-placement | 11:07 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Use os-resource-classes in placement https://review.openstack.org/623556 | 11:30 |
*** e0ne has quit IRC | 11:31 | |
*** jaypipes has joined #openstack-placement | 11:33 | |
*** helenafm has quit IRC | 11:57 | |
*** diga has joined #openstack-placement | 12:20 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Use os-resource-classes in placement https://review.openstack.org/623556 | 12:26 |
openstackgerrit | Chris Dent proposed openstack/placement master: Add irrelevant files list to perfload job https://review.openstack.org/624047 | 12:37 |
*** e0ne has joined #openstack-placement | 12:47 | |
guilhermesp | hello! seems that we have greenfield for our placement role https://review.openstack.org/#/c/618820/ | 12:53 |
guilhermesp | but I'd like to know from you guys how are we testing placement. Do we have a set of tempest tests for it? | 12:53 |
guilhermesp | testing my role locally, seems that the placement installation is not 100 percent functional, for that reason I'd like to test it with tempest of other tool | 12:54 |
cdent | guilhermesp: what are you seeing that suggests its not 100%? | 13:00 |
cdent | there was discussion last week that a) no there aren't specific placement tests in tempest, b) the compute service tests will exercise placement well | 13:01 |
guilhermesp | for example, I can't see placement port up in my tests. It's like there's something missing in MY installation process | 13:01 |
cdent | are you installating it on a port, or on a prefix /placement? That is, what ends up in the service catalog? | 13:02 |
guilhermesp | actually I'm configuring a port ( same as was before when placement was within nova) https://review.openstack.org/#/c/618820/48/defaults/main.yml line 122 that ends up in the service catalog | 13:04 |
guilhermesp | I think I'm missing some step. I was out last week so I'm planning to come back to this work this week to figure out | 13:05 |
guilhermesp | that's why I was curious about testing | 13:05 |
cdent | guilhermesp: mnaser, mriedem and I talked about it last week and worked out that the best route for testing via tempest would be to turn on the compute tests | 13:07 |
cdent | but if the port isn't coming up, I wonder if the web server is being restarted/reloaded after the config changes have been done | 13:08 |
sean-k-mooney | why are you doing the db setup https://review.openstack.org/#/c/618820/48/tasks/main.yml@55 after you start the service https://review.openstack.org/#/c/618820/48/tasks/placement_install.yml@52 | 13:08 |
*** e0ne has quit IRC | 13:11 | |
*** e0ne has joined #openstack-placement | 13:14 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Document API error codes https://review.openstack.org/624705 | 13:17 |
guilhermesp | sean-k-mooney: actually, the systemd was introduced across all roles with the purpose of this commit https://github.com/openstack/openstack-ansible-os_glance/commit/9eb319c22b5329f78b2b6c3c4f4744793c633744#diff-512247f2612555c70f64504f7c3916a7 ( take as an example ) | 13:18 |
guilhermesp | it has not really a side effect to set up db after the execution of this piece of code | 13:18 |
guilhermesp | cdent: thanks for the hint, I'll take a look at this detail | 13:18 |
sean-k-mooney | guilhermesp: your assuming that placement will be happy to start without the db configured which may result in it falling back to sqlite | 13:20 |
cdent | sean-k-mooney: placement won't start without the db explicitly configured, there is no default connection. but I agree, it should be configured and set up before the service starts, if possible | 13:21 |
*** helenafm has joined #openstack-placement | 13:22 | |
guilhermesp | sean-k-mooney: actually, the task https://review.openstack.org/#/c/618820/48/tasks/placement_post_install.yml is responsible to set up the placement config https://review.openstack.org/#/c/618820/48/templates/placement.conf.j2 that has the correct db path | 13:22 |
sean-k-mooney | it look like https://review.openstack.org/#/c/618820/48/handlers/main.yml should restart the services atht he end of the main task | 13:23 |
guilhermesp | yes sean-k-mooney actually that piece of systemd code was placed to refactor a lot of code that was a bit messy before. That's why seems a bit confusing | 13:24 |
guilhermesp | but I appreciate your questions about it sean-k-mooney and cdent that's very important | 13:25 |
sean-k-mooney | well using handeler makses sense for restarting the services actully. | 13:25 |
sean-k-mooney | guilhermesp: i would maybe move https://review.openstack.org/#/c/618820/48/tasks/placement_install.yml@52 to https://review.openstack.org/#/c/618820/48/tasks/placement_service_setup.yml@100 but other then that it seam like it would fix it self when you flush the handeler at the end of the role | 13:28 |
guilhermesp | good point sean-k-mooney I'll bring that discussion into openstack-ansible channel, as it seems we are keep the pattern of code across all roles. Maybe the guys out there has a really good reason to keep this where it is and I'm not able to tell you now. Let me discuss with them and I will give you a feedback about it | 13:30 |
sean-k-mooney | cdent: could this be related to the the global db context in any way? | 13:31 |
sean-k-mooney | guilhermesp: do you run placment under apache? | 13:32 |
cdent | sean-k-mooney: which "this" do you mean? | 13:32 |
sean-k-mooney | cdent: placement not startign properly | 13:32 |
guilhermesp | sean-k-mooney: I think that could be the missing part. I didn't configure apache2 and uwsgi there I think | 13:32 |
cdent | sean-k-mooney: i'm unwilling to speculate. Until I see some logs or something it could be anything. | 13:33 |
sean-k-mooney | cdent: fair. | 13:33 |
cdent | but certainly if apache and uwsgi aren't set up, that would do it | 13:34 |
sean-k-mooney | look like uwsgi is being used https://review.openstack.org/#/c/618820/48/defaults/main.yml@155 | 13:35 |
guilhermesp | let me check locally sean-k-mooney and cdent | 13:38 |
guilhermesp | sean-k-mooney: cdent http://paste.openstack.org/show/737484/ that could be one of the reasons | 13:40 |
* cdent is having a slow internet day | 13:40 | |
sean-k-mooney | yep i was just going to point out that i dont see where you create /etc/uwsgi/placement-api.ini | 13:41 |
guilhermesp | yep also there is no placement stuff in there http://paste.openstack.org/show/737485/ | 13:41 |
*** e0ne has quit IRC | 13:42 | |
sean-k-mooney | guilhermesp: i woudl guess the systemd service has a file not found error or something similar | 13:42 |
guilhermesp | could be sean-k-mooney I'd try to find something in the logs locally | 13:43 |
sean-k-mooney | cdent: is https://github.com/openstack/placement/blob/master/setup.cfg#L47 ment to auto generate the wsgi script on install | 13:43 |
sean-k-mooney | cdent: specifically /etc/uwsgi/placement-api.ini | 13:44 |
cdent | yes, it's built into pbr | 13:44 |
cdent | no | 13:44 |
cdent | that's something different | 13:44 |
sean-k-mooney | ok it generate teh script not the config | 13:44 |
guilhermesp | sean-k-mooney: just a note about systemd piece of code | 13:44 |
guilhermesp | the service unit files are laid down, but they are not started until the handlers are fired once the role execution is complete - for example: http://logs.openstack.org/43/619343/1/check/openstack-ansible-functional-ubuntu-bionic/9ffd452/job-output.txt.gz#_2018-11-21_20_04_42_849229 | 13:44 |
*** e0ne has joined #openstack-placement | 13:45 | |
cdent | placement-api.ini is the uwsgi config script, which is created by the deployment tooling | 13:45 |
sean-k-mooney | cdent: right in that case guilhermesp if you create the /etc/uwsgi/placement-api.ini file that might fix your issue its a good place to start in anycase | 13:46 |
cdent | guilhermesp, sean-k-mooney : this is what devstack does, if that's a useful reference: https://github.com/openstack-dev/devstack/blob/master/lib/apache#L235 | 13:47 |
sean-k-mooney | cdent its proably simpler to use a jinja template in ansible but ya that has all the default wsgi options | 13:48 |
sean-k-mooney | do they differ from how you run it in you docker dev env for placement | 13:49 |
cdent | oh sure, I wasn't suggesting _using_ that, but rather pointing it out as the options used | 13:49 |
cdent | I would guess that osa already has a template/role/task for such things and it just hasn't bee included, as a simple mistake | 13:50 |
cdent | sean-k-mooney: quite a lot different | 13:50 |
guilhermesp | hum, so which contents should be placed in placement-api.ini? or is it a blank file? | 13:50 |
cdent | but the docker stuff I've set up is designed to use uwsgi without apache | 13:50 |
cdent | guilhermesp: I'd suggest you look at the keystone role and see what it's doing\ | 13:50 |
cdent | sean-k-mooney: so the uwsgi in the docker env is set up to be more robust/complete | 13:51 |
guilhermesp | nice. I will take a look, do the modification and test it locally in a few minutes | 13:51 |
sean-k-mooney | cdent: so would you recommend https://github.com/cdent/placedock/blob/master/shared/placement-uwsgi.ini as a base for osa | 13:52 |
cdent | sean-k-mooney: I don't know enough about how osa sets up the other services, but would think it would be important to be consistent with the others (assuming they are doing apache with mod_proxy_uwsgi, then my config is definitely not what they want) | 13:53 |
cdent | scheduler/placement meeting in #openstack-meeting-alt in 5 minutes if people want one | 13:54 |
guilhermesp | oh great, I will attend the meeting as a guest :) | 13:56 |
sean-k-mooney | cdent: thats a good point i have not looked at osa before now but ya if they are usign mod_wsgi then ya it should be consitent | 13:57 |
guilhermesp | I agree cdent and sean-k-mooney about keep things like osa does, but would be good just as a proof of concept to follow your suggestion locally, then adapt how osa does in general | 13:57 |
*** ttsiouts has quit IRC | 14:00 | |
*** ttsiouts has joined #openstack-placement | 14:00 | |
*** takashin has joined #openstack-placement | 14:01 | |
*** ttsiouts has quit IRC | 14:04 | |
*** ttsiouts has joined #openstack-placement | 14:04 | |
*** mriedem has joined #openstack-placement | 14:08 | |
*** takashin has left #openstack-placement | 14:08 | |
guilhermesp | I think this is one example how we set up uwsgi in keystone https://github.com/openstack/openstack-ansible-os_keystone/blob/master/templates/keystone-uwsgi.ini.j2 cdent sean-k-mooney | 14:12 |
cdent | guilhermesp: that looks very much like what placement would want, except of cource the template vars would be different | 14:14 |
cdent | and the virtualenv | 14:14 |
guilhermesp | yep, I'm taking this https://github.com/cdent/placedock/blob/master/shared/placement-uwsgi.ini that sean-k-mooney suggested | 14:14 |
cdent | edleafe: can you remember of the top of your head the tools you used for trimming nova history to create the placement repo? | 14:40 |
edleafe | cdent: I have a script I used lying around somewhere. Gimme a minute to dig it up | 14:41 |
cdent | thanks | 14:43 |
edleafe | cdent: ok, it was the `filter_git_histroy.sh` that did the extraction, and another script that ran the filter script with the list of files/directories to preserve. | 14:49 |
cdent | and that came from oslotools or something like, yes? | 14:50 |
* cdent looks | 14:50 | |
edleafe | cdent: after that I had a series of scripts to clean up the extracted files to remove the unwanted things | 14:50 |
edleafe | Yeah - I think you were the one who pointed me to it | 14:50 |
edleafe | I have a copy if you need it | 14:50 |
cdent | cool, thanks. i had some of the pieces in my head, but not the name of the script | 14:50 |
cdent | no that's enough, thank you | 14:51 |
edleafe | You starting over? :) | 14:51 |
*** e0ne has quit IRC | 15:02 | |
*** e0ne has joined #openstack-placement | 15:06 | |
*** efried has joined #openstack-placement | 15:06 | |
cdent | edleafe: I responded on your startswith suggestion in the "naw" tone. But I can be convinced otherwise if you disagree with my disagreeing | 15:47 |
cdent | edleafe: also, I take it your were happy with the locking changes? | 15:48 |
edleafe | Yeah, the locking stuff looked fine | 15:49 |
edleafe | I'm just curious about the "another namespace" idea. What would be another possible namespace? | 15:50 |
cdent | there was discussion in traits (ack they are not the same, but...as an example) (on a spec I can't remember) of using prefixes on traits to mean things | 15:51 |
cdent | so it just seems like a possibility (albeit one I would resist) | 15:52 |
cdent | but since it is conceivable, and the numeric-based solution works, why not? | 15:52 |
edleafe | It just is one other mental step to xlate "minimum id" to "able to be deleted" | 15:53 |
edleafe | Not a big deal at all | 15:53 |
cdent | k, cuz if it was I would want to direct you to some other mental steps that seem bigger | 15:55 |
openstackgerrit | Chris Dent proposed openstack/placement master: Use os-resource-classes in placement https://review.openstack.org/623556 | 15:58 |
edleafe | Heh, well, those other things aren't being changed in your patch :) | 15:58 |
*** diga has quit IRC | 16:12 | |
gibi | cdent: sorry I run out of time so I have to push the os-resource-classes to tomorrow | 16:14 |
cdent | gibi: no worries, at this time of year rushing isn't going to help much of anything | 16:15 |
*** N3l1x has joined #openstack-placement | 16:23 | |
*** N3l1x_ has joined #openstack-placement | 16:23 | |
* cdent goodnights | 16:54 | |
*** cdent has quit IRC | 16:54 | |
*** ttsiouts has quit IRC | 16:56 | |
*** ttsiouts has joined #openstack-placement | 16:56 | |
*** ttsiouts has quit IRC | 17:01 | |
*** efried has quit IRC | 17:01 | |
*** helenafm has quit IRC | 17:02 | |
*** e0ne has quit IRC | 17:46 | |
*** e0ne has joined #openstack-placement | 18:42 | |
*** e0ne has quit IRC | 18:46 | |
*** mriedem has quit IRC | 18:48 | |
*** mriedem has joined #openstack-placement | 18:51 | |
*** e0ne has joined #openstack-placement | 19:01 | |
*** N3l1x has quit IRC | 19:53 | |
*** e0ne has quit IRC | 22:25 | |
*** mriedem has quit IRC | 23:40 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!