*** holser has quit IRC | 01:06 | |
*** holser has joined #oooq | 01:09 | |
*** holser has quit IRC | 02:16 | |
*** rfolco has joined #oooq | 02:50 | |
*** rfolco has quit IRC | 02:54 | |
*** holser has joined #oooq | 05:17 | |
*** ykarel has joined #oooq | 06:30 | |
*** ykarel has quit IRC | 06:41 | |
*** Goneri has quit IRC | 06:52 | |
*** Tengu has quit IRC | 07:39 | |
*** Tengu has joined #oooq | 07:41 | |
*** d0ugal has quit IRC | 07:49 | |
*** d0ugal has joined #oooq | 07:49 | |
*** ykarel has joined #oooq | 08:19 | |
*** ykarel has quit IRC | 09:10 | |
*** ykarel has joined #oooq | 09:11 | |
*** ykarel has quit IRC | 09:46 | |
zbr | sshnaidm: happy new year! nice to see you back. | 10:10 |
---|---|---|
sshnaidm | zbr, thanks | 10:19 |
zbr | sshnaidm: what do you think about https://github.com/containers/libpod/issues/4665 ? | 10:22 |
zbr | mainly my rhel-8 box cannot run either docker nor podman, lucky that centos-8 works fine. | 10:23 |
sshnaidm | zbr, do you build rpm from libpod master? | 10:27 |
zbr | sshnaidm: afaik not, but maybe I should reinstall the rpm. maybe i have done it while testing the rpmbuilding and I forgot since. | 10:28 |
sshnaidm | zbr, last comments in issue were about building from master and it worked | 10:28 |
zbr | i bet from master it will work as I wrote the rpm building jobs back in december, but that will be 1.7.x | 10:29 |
zbr | but now that I remember, the rpm building jobs tests only installation of podman, it does not run a podman info after. so there is a chance something could get broken. | 10:30 |
zbr | sshnaidm: btw, fixed https://github.com/ansible/molecule/pull/2482 | 10:32 |
sshnaidm | zbr, reviewed pr | 10:34 |
zbr | answered. | 10:37 |
zbr | import gives false positive results | 10:37 |
zbr | mainly not fixing the problem | 10:37 |
zbr | also the "cli" tests would fail without having the package installed because the script does look for package. | 10:38 |
zbr | import was my first attempt :D | 10:38 |
sshnaidm | ack | 10:42 |
zbr | sshnaidm: so lets merge https://github.com/ansible/molecule/pull/2482 | 11:13 |
zbr | sshnaidm: re using "podman info" vs "podman ps" to test if is working, you know for sure one case where ps woulf fail and info would not? | 11:17 |
zbr | i prefer info because we could also extract some info from its output, ps output is kinda useless. | 11:17 |
sshnaidm | zbr, afaik podman ps is more functional, like making mounts for example, not sure podman info does it | 11:18 |
sshnaidm | do you need this output? | 11:18 |
zbr | not at this moment, for example for docker I did but for this one not yet. i will switch to ps and we can improve later when we encounter the need for it. | 11:19 |
sshnaidm | zbr, maybe better to catch pkg_resources.DistributionNotFound and not general Exception | 11:19 |
zbr | in fact the best test would be to run a container | 11:19 |
sshnaidm | zbr, yeah, but this is already done in tests | 11:20 |
sshnaidm | zbr, if "info" fails with wrong conmon, maybe it's even enough | 11:21 |
zbr | sshnaidm: it does, both of them fail with conmon. reinstalling fixed it. | 11:21 |
*** tosky has joined #oooq | 11:24 | |
zbr | sshnaidm: it would be great if we can spend some time fixing the slow functional tests in molecule, is kinda insane how long the run for what they really do. don't you agree? | 11:26 |
sshnaidm | zbr, yeah, worth to check.. | 11:27 |
zbr | unit are quite ok, with few minor tunes I managed a record: 116.78 seconds to run all unit. | 11:28 |
zbr | sshnaidm: have you read https://www.reddit.com/r/ansible/comments/eji7ie/should_we_consider_use_of_tasks_from_with_ansible/ ? | 11:30 |
*** bogdando has joined #oooq | 11:56 | |
*** bogdando has quit IRC | 11:57 | |
sshnaidm | zbr, I'm not sure it's "antipattern", it exists intentionally there | 12:10 |
sshnaidm | zbr, maybe more atomic role would prevent using it, but it's not always justified | 12:11 |
sshnaidm | zbr, i.e. our undercloud-setup role, it includes a lot of different and not related stuff, using only part of it seems reasonable | 12:11 |
sshnaidm | zbr, and using variables/tags for enable/disable tasks in role seems like more complicated and difficult to debug | 12:12 |
zbr | i barely find a good reason to justify its use, mainly because i was not yet able to find a case where using a dispatcher would not work. | 12:13 |
zbr | probably is a gray-areas, likely bad practice but maybe not in all cases. | 12:13 |
sshnaidm | zbr, it can work in various ways, but some of them are less clear ans easy than others | 12:14 |
sshnaidm | zbr, usually having atomic and single-usage roles would not require tasks_from | 12:14 |
sshnaidm | zbr, maybe collections would be useful here, since they can contain multiple small roles | 12:17 |
sshnaidm | and hopefully will replace all our huge overloaded roles | 12:17 |
*** holser has quit IRC | 12:20 | |
zbr | sshnaidm: true about atomic but is not as easy to achieve in practice, especially where you need to share a lot of code/logic between sister roles. 2020 will show us where collections are sending us towards. | 12:20 |
*** holser has joined #oooq | 12:21 | |
zbr | sshnaidm: i rewrote https://github.com/ansible/molecule/pull/2483/files based on your feedback. | 12:36 |
zbr | lru_cache is for performance | 12:36 |
zbr | mentioned in commit message | 12:36 |
zbr | i think the new ifs order should make it easier to read | 12:37 |
sshnaidm | zbr, ack, +2 | 12:42 |
*** ykarel has joined #oooq | 12:53 | |
*** holser has quit IRC | 13:19 | |
zbr | sshnaidm: need bit of help around podman, see http://paste.openstack.org/show/788054/ | 13:46 |
zbr | mainly one simple ansible task fais apparently due to missing cp on podman | 13:46 |
zbr | i wasn't expecting to find a reason to check podman version so soon, but first lest double check if is not something else | 13:47 |
zbr | there is also a related bug https://bugzilla.redhat.com/show_bug.cgi?id=1759713 | 13:48 |
openstack | bugzilla.redhat.com bug 1759713 in podman "podman cp command not found, but man page says it should be there" [Unspecified,Closed: currentrelease] - Assigned to bbaude | 13:48 |
*** d0ugal has quit IRC | 13:51 | |
sshnaidm | zbr, what is ansible version? | 14:26 |
sshnaidm | zbr, package: podman-1.0.5-1. - is really old | 14:30 |
sshnaidm | it doesn't worth to run anything before 1.4 | 14:30 |
zbr | we will have to add test for it | 14:35 |
sshnaidm | test for what? | 15:05 |
*** ykarel has quit IRC | 15:11 | |
sshnaidm | zbr, do you know maybe what is going on with https://cloud.centos.org/centos/8/ ? | 15:40 |
*** ykarel has joined #oooq | 15:41 | |
zbr | sshnaidm: no but https://www.reddit.com/r/CentOS/comments/dtv0jv/centos_81_eta/ contains an interesting link | 15:42 |
sshnaidm | seems like some abandoned pet project of dropout | 15:42 |
zbr | we still get 403 | 15:42 |
zbr | so they are building it,... | 15:43 |
zbr | i also faced a podman-manpages-1.4.2-5.module_el8.1.0+237+63e26edc.noarch conflicts with file from package docker-ce-cli-1:19.03.5-3.el7.x86_64 | 15:44 |
sshnaidm | manpages? o_O | 15:44 |
zbr | which makes be unkeen to continue with the update, i may endup with the same situation i have on rhel 8.1, where I cannot used any of them. | 15:44 |
*** ykarel has quit IRC | 16:08 | |
*** ykarel has joined #oooq | 16:08 | |
*** ykarel has quit IRC | 16:13 | |
zbr | https://github.com/containers/libpod/issues/4791 -- is bit fun as I have an idea on how to change our jobs to reproduce the bug. | 16:36 |
zbr | the kind of bug where everyone will be happy to throw the dead cat to someone else | 16:37 |
zbr | as docker-ce is a 3rd party | 16:37 |
sshnaidm | it's all this weird idea to replace "docker" command with "podman" | 16:40 |
sshnaidm | it's fine maybe for deprecated tools | 16:40 |
sshnaidm | but for competing? | 16:41 |
zbr | i agree with you, i think that imagining that they are drop-in replacement was... a dream. | 17:02 |
zbr | drop-in for <5% of cases, mainly for hello-world :D | 17:03 |
zbr | anyway, we need to find someone to workflow the first two linter changes today | 17:03 |
zbr | i would be happy to help but i not have the +2 | 17:04 |
*** ykarel has joined #oooq | 17:39 | |
sshnaidm | zbr, are we good to +w https://review.opendev.org/#/c/701113 ? | 17:41 |
zbr | yes, do it. i added comment | 17:42 |
sshnaidm | zbr, left comments in https://review.opendev.org/#/c/701116/ I don't think we need -xe there | 17:48 |
sshnaidm | it's not like pure bash | 17:49 |
*** ykarel is now known as ykarel|away | 17:54 | |
*** ykarel|away has quit IRC | 18:07 | |
*** holser has joined #oooq | 18:10 | |
zbr | sshnaidm: guess, what... https://review.rdoproject.org/r/#/c/24349/ | 18:38 |
zbr | i wonder if I should make them NV to merge that change. | 18:38 |
*** holser has quit IRC | 18:38 | |
zbr | that change breaks the jobs, but that is a good thing. | 18:39 |
zbr | mainly what we builded so far was not so usable after all. | 18:40 |
*** holser has joined #oooq | 18:44 | |
*** holser has quit IRC | 18:46 | |
*** holser has joined #oooq | 18:50 | |
*** Goneri has joined #oooq | 20:38 | |
*** tosky has quit IRC | 20:49 | |
*** holser has quit IRC | 20:51 | |
*** holser has joined #oooq | 20:52 | |
*** holser has quit IRC | 20:57 | |
ssbarnea | help me merge https://review.rdoproject.org/r/#/c/24349/ -- is OK for these jobs to fail. | 20:58 |
*** holser has joined #oooq | 21:24 | |
*** tosky has joined #oooq | 22:00 | |
*** holser has quit IRC | 22:59 | |
*** holser has joined #oooq | 23:47 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!