| jamesdenton_ | good morning, all | 13:21 |
|---|---|---|
| *** jamesdenton_ is now known as jamesdenton | 13:21 | |
| mgariepy | good morning jamesdenton | 13:21 |
| jamesdenton | /wave | 13:23 |
| jrosser | hello | 14:38 |
| jamesdenton | howdy | 14:43 |
| mgariepy | hmm.. is it possible to override the horizon config ? like adding the manila-ui config in there with an override ? | 17:49 |
| mgariepy | like : https://github.com/openstack/manila-ui/blob/master/doc/source/configuration/index.rst | 17:50 |
| jamesdenton | i am not aware | 18:22 |
| jrosser | I wonder if you can put that in the regular local_settings rather than the Manila specific one | 18:24 |
| mgariepy | yes you can | 18:29 |
| mgariepy | https://github.com/openstack/openstack-ansible-os_horizon/blob/master/releasenotes/notes/horizon-arbitrary-config-8a36e4bd6818afe1.yaml | 18:31 |
| mgariepy | not quite clear in my head how you put a dict in there :/ | 18:31 |
| opendevreview | Merged openstack/openstack-ansible-os_cinder master: Reduce memory consumption in Cinder services https://review.opendev.org/c/openstack/openstack-ansible-os_cinder/+/888027 | 18:59 |
| jrosser | rabbitmq/erlang repos fail again https://zuul.opendev.org/t/openstack/build/970f7c441b724e9c8d3cb03f70e37234/log/job-output.txt#13229 | 21:11 |
| spatel | https://ibb.co/k61dSh3 Did you see this in octavia? I am not able fill port info | 21:31 |
| johnsom | It was an issue related to newer versions of chrome, see this bug: https://bugs.launchpad.net/octavia/+bug/2024188 | 21:46 |
| tuggernuts | is there a solution for https://pastebin.com/P46dnubh issue? if I rerun setup-openstack.yml it seems to move past this spot | 22:17 |
| tuggernuts | this happens during the keystone setup and is from the antelope deployment | 22:22 |
| jrosser | tuggernuts: i think that the question really is how your private key file ended up with 0640 permissiosn | 22:45 |
| tuggernuts | yeah | 22:46 |
| tuggernuts | somehow the deploy isn't resolving this on the first run | 22:47 |
| jrosser | well it should be defined here https://github.com/openstack/openstack-ansible-plugins/blob/master/roles/ssh_keypairs/vars/standalone_keypair.yml#L36 | 22:48 |
| jrosser | that the private key file is 0600 | 22:48 |
| jrosser | and if this were a more general bug then it would be picked up in our CI | 22:49 |
| jrosser | those keys are generated on the deploy host then copied to the keystone hosts, so i think it's ok to move/rename them and re-run the os_keystone playbook to see what happens | 22:53 |
| tuggernuts | ok, does your CI stuff happen to test any of the ubuntu repo distro install path? | 22:53 |
| jrosser | oh :( | 22:53 |
| jrosser | distron | 22:53 |
| tuggernuts | yeah not source | 22:53 |
| jrosser | why? :) | 22:54 |
| tuggernuts | umm long story but short story is faster and I can mirror the repos | 22:54 |
| jrosser | yes we do test distro installation in CI | 22:55 |
| tuggernuts | ok | 22:55 |
| jrosser | but only for a single host, not H/A | 22:55 |
| tuggernuts | ok | 22:55 |
| tuggernuts | must be something my environment is doing then | 22:55 |
| jrosser | there is a CI for 3x keystone contaniners and key distribution for the source install | 22:55 |
| jrosser | but there should be no differnce at all for this | 22:55 |
| tuggernuts | what's annoying is it only happens first run | 22:56 |
| jrosser | just FYI i think i have run patched versions of openstack services for so many releases now | 22:56 |
| jrosser | this is the superpower of the source install | 22:56 |
| jrosser | the distro install is (imho) becoming more experimental over time and we have repeatedly discussed removing that option totally | 22:57 |
| tuggernuts | ah I see | 22:58 |
| tuggernuts | I guess I don't know enough about how the source install gets pinned | 22:58 |
| tuggernuts | I need to have 9 of these environments deployed globally and I like to have them all based on the same code | 22:59 |
| tuggernuts | technically I have 18 of them =P | 22:59 |
| tuggernuts | I'm doing Blue/Green deployments | 23:00 |
| jrosser | but they will be | 23:00 |
| jrosser | look here https://github.com/openstack/openstack-ansible/blob/master/inventory/group_vars/keystone_all/source_git.yml | 23:00 |
| tuggernuts | gotcha ok | 23:00 |
| jrosser | each tagged release of OSA specifies the git SHA of the openstack services | 23:00 |
| jrosser | and the openstack python upper-constraints defines the depenancies | 23:00 |
| jrosser | so you can trivially override those vars to point to your own forks, should you need to | 23:01 |
| tuggernuts | I see | 23:01 |
| tuggernuts | that does sound fine then | 23:01 |
| jrosser | and if you want a mirror of the all yourself, define openstack_opendev_base_url to be your mirror server | 23:01 |
| jrosser | theres a matching one for the things we grab from github | 23:01 |
| tuggernuts | is there a full list somewhere of all the repos? | 23:02 |
| tuggernuts | or they just al lin source_git.yml? | 23:02 |
| jrosser | on ubuntu distro install, you'll get "whatever happens to be in the UCA repo on the day you install" so arguably thats less well defined than the source install | 23:02 |
| jrosser | in the past we had one large file with all of those source_git vars together | 23:03 |
| jrosser | but for $reasons thats been split out into the indivdual group vars very recently | 23:03 |
| tuggernuts | I run an aptly instance where I snapshot these repos during releases basically | 23:03 |
| tuggernuts | then I move that snapshot around | 23:03 |
| jrosser | it's fine, if you're happy with whatever UCA happens to provide | 23:04 |
| tuggernuts | I'll dig on this source install more because if you think source is more stable then that's what I want | 23:04 |
| jrosser | but the distro install is vastly less tested, and less used than the source install | 23:05 |
| jrosser | i think some of the roles may not support it at all, though i'm not quite sure about that | 23:05 |
| tuggernuts | I just don't understand it enough to know how I'm getting the same code everywhere and how to mirror things so I don't have to rely on external repos | 23:05 |
| jrosser | you mirror the git repos | 23:05 |
| tuggernuts | yeah | 23:05 |
| tuggernuts | I just need to understand how to do it | 23:06 |
| jrosser | but the SHA you use everywhere are defines | 23:06 |
| jrosser | *defined | 23:06 |
| jrosser | so it has to be the same code, by virtue of the way git hashes work | 23:06 |
| tuggernuts | right | 23:06 |
| jrosser | there is possibly some variance on the things you get from pypi | 23:07 |
| tuggernuts | years ago when we decided to use this openstack-ansible code we just thought the source was too slow and we moved on without much more thought than that | 23:07 |
| jrosser | it's true the distro install is faster | 23:07 |
| tuggernuts | yeah ok this is what I didn't know about inventory/group_vars/*/source_git.yml | 23:08 |
| jrosser | right - when we make a release there is a script which updates those all to the most recent SHA | 23:09 |
| tuggernuts | I had in my mind that each environment I deployed would get a different HEAD, I had no data to back this up lol | 23:09 |
| jrosser | nooooo | 23:10 |
| jrosser | for example here is the patch which defines what the 27.0.1 release is for a source install https://github.com/openstack/openstack-ansible/commit/9921c23ea463c8cd16f709198e7602157848e41a | 23:13 |
| jrosser | it is completely impossible to define what the equivalent is for a distro install without doing something additional like your aptly repo snapshot | 23:13 |
| tuggernuts | ok I'll probably delete and recreate this deployment tomorrow with a source install and see what's different | 23:13 |
| jrosser | and like i say this really becomes powerful when you need to apply a local patch or bugfix | 23:14 |
| jrosser | because you can override a couple of vars and point to your own github/mirror and deploy a fix pretty quickly | 23:14 |
| tuggernuts | yeah for sure because our old stein deployment had at least 30 patches we applied post setup | 23:15 |
| jrosser | have a go in an all-in-one, it should be pretty easy to dig around and see how it works | 23:16 |
| tuggernuts | for sure thanks for all the info | 23:19 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!