15:00:36 <noonedeadpunk> #startmeeting openstack_ansible_meeting
15:00:36 <opendevmeet> Meeting started Tue Mar 29 15:00:36 2022 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:36 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:36 <opendevmeet> The meeting name has been set to 'openstack_ansible_meeting'
15:00:44 <noonedeadpunk> #topic rollcall
15:00:47 <jrosser> o/ hello
15:00:53 <noonedeadpunk> \o/
15:00:55 <noonedeadpunk> hey there
15:05:06 <mgariepy> o/ half there as usual ;)
15:05:11 <jrosser> i kind of got fed up trying to fix centos things :(
15:05:50 <noonedeadpunk> #topic office hours
15:06:10 <noonedeadpunk> ok, so what has happened to centos again?
15:06:30 <jrosser> this during the lxc cache prep https://paste.opendev.org/show/bCn7QIEJ0VvYwbNofYhd/
15:06:55 <jrosser> and as far as i can see, centos-gpg-keys-1:8-5.el8.noarch is only in EPEL
15:07:07 <jrosser> so why it tries to install it at that point, i don't really know
15:07:37 <jrosser> though we did get a tip to fix it anskiy> jrosser: you can "workaround" it with "--nobest"
15:08:02 <noonedeadpunk> not sure you can provide that to ansible module...
15:08:27 <jrosser> i think we template out a shell script for this
15:09:07 <noonedeadpunk> ah, for preparing lxc cache yeah, you did that with dnf install or smth I believe
15:09:11 <jrosser> https://github.com/openstack/openstack-ansible-lxc_hosts/blob/master/templates/prep-scripts/redhat_prep.sh.j2#L15-L19
15:09:21 <jrosser> this is once the chroot is made
15:09:31 <noonedeadpunk> but I wonder if that could be jsut some infra mirror issue? or you was able to reproduce that?
15:09:53 <jrosser> i have been in a bunch of meetings today so it's not been possible to spend proper time on it
15:10:28 <jrosser> i had a quick look in the infra mirror and compared it to rpm-find and couldnt see much problem
15:10:41 <jrosser> other than notice it only being in EPEL which i still don't understand
15:10:49 * noonedeadpunk in new country since monday so was accomodating
15:10:56 <jrosser> unless we cross over centos-8 and centos-8-stream still somewhere
15:11:51 <noonedeadpunk> or rocky/with stream
15:13:49 <jrosser> i did wonder if there is still room to really simplify away a lot more stuff from lxc_hosts / lxc_container_create
15:14:18 <jrosser> anyway, so there is that for centos
15:14:49 <jrosser> and the other difficult problem we have is with the latest virtualenv(bundled setuptools) for openstack-tox-docs jobs
15:15:40 <jrosser> the only proper solution is this https://github.com/pypa/virtualenv/pull/2324
15:15:50 <noonedeadpunk> jrosser: hm, isn't there even 8.6 now? https://centos.pkgs.org/8-stream/centos-baseos-aarch64/centos-gpg-keys-8-6.el8.noarch.rpm.html
15:16:21 <noonedeadpunk> maybe they released smth and then dropped?
15:16:56 <noonedeadpunk> * https://centos.pkgs.org/8-stream/centos-baseos-x86_64/centos-gpg-keys-8-6.el8.noarch.rpm.html
15:16:57 <jrosser> it's 8-6 though, not 8.6
15:17:08 <jrosser> 6th rev. of 8?
15:17:31 <jrosser> but yes that does look later
15:17:34 <noonedeadpunk> but it was complaining about 8-4 vs 8-5 in paste
15:17:39 <noonedeadpunk> and 8-5 is not even available
15:18:07 <noonedeadpunk> so maybe we should jsut recheck tomorrow :)
15:18:34 <jrosser> huh 6 is here now https://mirror.dfw.rax.opendev.org/centos/8-stream/BaseOS/x86_64/os/Packages/
15:18:42 <jrosser> maybe the mirrors have updated recently
15:18:48 <noonedeadpunk> yup
15:18:59 <noonedeadpunk> so let's just wait then.....
15:19:17 <noonedeadpunk> regarding setuptools - that sounded to me like u-c update should kind of solve issue?
15:19:37 <noonedeadpunk> but likely for stable branches it would be still broken
15:20:23 <jrosser> u-c isnt helpful here
15:20:41 <jrosser> in the openstack-tox-docs job it doesnt use any of our code to prepare the environment
15:20:59 <jrosser> so it takes some zuul-jobs stuff to install python/virtualenv/whatever
15:21:15 <jrosser> so the bundled setuptools gets used regardless of what u-c says
15:21:57 <jrosser> but then u-c prevents a different version of setuptools being installed in the tox environment
15:22:00 <jrosser> so complete catch-22
15:26:43 <jrosser> there is a hacky fix https://review.opendev.org/c/openstack/openstack-ansible-tests/+/835468
15:40:42 <noonedeadpunk> I guess I just didn't get at what stage issue happens. So once venv is being created - first thing we do we upgrade pip, setuptools and wheels
15:40:50 <noonedeadpunk> and that's what failing, right?
15:41:11 <noonedeadpunk> oh, wait, it's for tox only I guess?
15:47:59 <noonedeadpunk> another thing - next week is PTG one _already_ and we have super empty etherpad as of today...
15:52:31 <noonedeadpunk> oh, damn, it just tries to install local setup.py regardless...
15:53:34 <noonedeadpunk> and we get python3-venv for distro....
15:53:37 <noonedeadpunk> *from
15:53:45 <jrosser> yes
15:53:49 <jrosser> it's not really our job
15:54:09 <jrosser> i think the only place we have to adjust things is inside tox.ini in the tests repo
15:55:08 <noonedeadpunk> well, we could create our by parenting, but dunno if that will help
15:55:32 <noonedeadpunk> anyway I don't think this workaround will hurt unless we need to set it everywhere
16:00:44 <noonedeadpunk> #endmeeting