15:02:19 <noonedeadpunk> #startmeeting openstack_ansible_meeting 15:02:19 <opendevmeet> Meeting started Tue Feb 7 15:02:19 2023 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:19 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:19 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting' 15:02:35 <noonedeadpunk> #topic rollcall 15:02:38 <noonedeadpunk> o/ 15:02:39 <damiandabrowski> hi! 15:05:53 <jrosser> o/ hello 15:08:27 <noonedeadpunk> #topic office hours 15:08:48 <noonedeadpunk> sorry - was a bit distracted 15:09:20 <noonedeadpunk> I actually think we had some new bugs during previous week, but let's do bug triage later 15:10:22 <noonedeadpunk> So it seems that our gates are broken now 15:10:34 <jrosser> andrewbonney is away for several months so we are going to be low on reviewers for a while 15:11:03 <damiandabrowski> and curl issue is more complicated than i thought 15:11:15 <damiandabrowski> - on centos 'curl' is installed, so we can't install 'curl-minimal' during bootstrap-ansible.sh 15:11:15 <damiandabrowski> - on rocky 'curl-minimal' is installed, so we can't install 'curl' during bootstrap-ansible.sh 15:11:16 <noonedeadpunk> damiandabrowski: yeah, I know where issue is I guess 15:11:21 <damiandabrowski> so i see few possible options: 15:11:24 <damiandabrowski> a) do not try to install curl on dnf-based systems and just assume they are present by default 15:11:27 <damiandabrowski> b) define a different set of packages installed during bootstrap-ansible.sh(rocky with curl-minimal and centos/rhel with curl) 15:11:31 <damiandabrowski> c) pass --allowerasing to dnf install 15:11:33 <damiandabrowski> d) pass --skip-broken to dnf install 15:12:15 <noonedeadpunk> I think we can install curl-minimal everywhere on rhel? 15:12:37 <noonedeadpunk> jrosser: gotcha, thanks for updating 15:13:05 <noonedeadpunk> The question about curl - is where we try to install it or better - how much work is to replace 15:14:05 <noonedeadpunk> I assume integrated repo and openstack_hosts are the only? 15:14:19 <noonedeadpunk> Which is already a circular dependency, but not that bad, right? 15:15:27 <jrosser> we need to make this select between curl/curl-minimal then? 15:18:59 <noonedeadpunk> I'd try to just do culr-minimal everywhere for rhel9 15:20:50 <noonedeadpunk> Also I've seen jammy timeouting couple of times... 15:21:00 <noonedeadpunk> and timeout was quite weird 15:21:42 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Replace curl with curl-minimal for DNF https://review.opendev.org/c/openstack/openstack-ansible/+/872896 15:21:59 <noonedeadpunk> I do hope it's just slow provider rather then smth off with lxc images (again) 15:24:09 <damiandabrowski> ahh so it's not that curl/curl-minimal packages are available in images by default. We install them during bootstrap-host 15:24:11 <damiandabrowski> I missed that before 15:24:54 <noonedeadpunk> we also have that in openstack_hosts on a later step 15:26:58 <damiandabrowski> do you have an idea how to solve this circular dependency? 15:28:55 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Replace curl with curl-minimal for DNF https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/872973 15:29:12 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Replace curl with curl-minimal for DNF https://review.opendev.org/c/openstack/openstack-ansible/+/872896 15:29:21 <noonedeadpunk> yup, as usual 15:29:23 <damiandabrowski> ahh, zuul.d/project.yaml :D 15:30:04 <opendevreview> Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Replace curl with curl-minimal for DNF https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/872973 15:30:05 <noonedeadpunk> damn 15:30:39 <noonedeadpunk> ok, anything else interesting to discuss? 15:30:56 <noonedeadpunk> I've seen discussions going on around refactoring of haproxy_services 15:31:11 <noonedeadpunk> I don't have much to comment as didn't read up on these 15:32:18 <damiandabrowski> no worries, let's discuss it next time 15:33:10 <jrosser> personally i would have preferred to see the haproxy role used a little more like we do system_service/pki/python_venv_build 15:33:58 <jrosser> i.e very obvious <this role> with <this data> in the service playbooks 15:35:01 <noonedeadpunk> I think it goes to on how important do we see reaching goal `1. Variables scope` 15:36:34 <damiandabrowski> btw. jrosser i fixed most of your comments yesterday 15:36:37 <noonedeadpunk> As all this complexity was to make that thing work from what I got. And I personally don't care much about it. I kind of agree it's overall confusing though, but from other side we mainly promote using user_variables and not group_vars as group_vars tended to slow ansible down if I'm not mistaken? 15:37:36 <noonedeadpunk> but yeah, things could be simplifed already :) 15:39:10 <noonedeadpunk> #topic bug triage 15:39:22 <noonedeadpunk> So after rejecting couple of bugs, this one left 15:39:27 <noonedeadpunk> #link https://bugs.launchpad.net/openstack-ansible/+bug/2004546 15:39:35 <noonedeadpunk> And I think it's quite plausible 15:40:16 <noonedeadpunk> And I guess it's due to having ternary right in variable 15:40:41 <jrosser> what an odd set of packages 15:42:11 <noonedeadpunk> I think main complain is on empty line at the end of the file 15:42:23 <noonedeadpunk> Which might change depending on the role 15:42:30 <noonedeadpunk> *role you run 15:43:54 <noonedeadpunk> though I think we might want to have a bit more details.... 15:44:03 <jrosser> the bug is very terse, is it talking about this? https://github.com/openstack/ansible-role-python_venv_build/blob/master/tasks/python_venv_wheel_build.yml#L176 15:44:58 <jrosser> #179 i mean 15:46:08 <noonedeadpunk> I think so... 15:47:22 <noonedeadpunk> or it could be global-constraints.txt as well 15:47:44 <noonedeadpunk> I'd say it's more about this https://github.com/openstack/ansible-role-python_venv_build/blob/master/tasks/python_venv_install.yml#L109-L122 15:47:48 <jrosser> perhaps just a little more info would be helpful 15:48:16 <noonedeadpunk> Yeah, will ask for this 15:48:18 <jrosser> but relatedly i did see a thing where you could so `{{ some_list | select }}` and it would remove falsy elements 15:48:27 <jrosser> like empty or null or whatever 15:48:55 <noonedeadpunk> yeah, I was also thinking to apply filter there to drop empty elements from the list 15:50:31 <jrosser> it might also be a {% vs {%- thing? 15:51:12 <jrosser> those confuse me sometimes particularly with multiline inline jinja like that 15:52:21 <noonedeadpunk> Oh, true, we have `|` vs `|-` 15:52:30 <noonedeadpunk> but why it would be different.... 15:53:04 <noonedeadpunk> Anyway the only way is likely to reproduce the issue and realize what the issue really is 15:55:27 <jrosser> there was another wheel build bug too i think? 16:08:19 <noonedeadpunk> Hm... I don't really see it... 16:08:47 <jrosser> https://bugs.launchpad.net/openstack-ansible/+bug/2004252 16:09:54 <noonedeadpunk> oh 16:10:04 <noonedeadpunk> that;s interesting 16:10:15 <noonedeadpunk> I though you meant this one tbh https://bugs.launchpad.net/openstack-ansible/+bug/2003294 16:11:34 <jrosser> omg the irony if it was rackspace dos'ing the infra git servers 16:11:53 <noonedeadpunk> lol 16:12:18 <noonedeadpunk> yeah, that would be "fun" 16:13:51 <noonedeadpunk> regarding 2004252 - I'm about to give up and hardcode `venv_wheel_build_enable: true` 16:14:04 <noonedeadpunk> With overriding it to false for aio 16:14:31 <noonedeadpunk> But that will make us testing what nobody outside of CI will use 16:14:34 <jrosser> i was not even really able to understand the bug 16:15:09 <jrosser> and regarding https://bugs.launchpad.net/openstack-ansible/+bug/2003294, watch this...... 16:15:16 <jrosser> jamesdenton: ^ can you take a look? 16:16:45 <noonedeadpunk> oh my, we're out of time 16:16:48 <noonedeadpunk> #endmeeting