| opendevreview | Farbod Saghafi proposed openstack/openstack-ansible-lxc_container_create master: Solving naming issue of OVS ports for lxc https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/938424 | 07:06 |
|---|---|---|
| jrosser | o/ morning | 08:18 |
| noonedeadpunk | o/ | 08:25 |
| jrosser | i was just looking at your molecule tests for networkd - which bits were troublesome? | 08:27 |
| noonedeadpunk | so networkd doesn't bring interfaces up in docker | 08:28 |
| jrosser | locally here it seems also to need https://review.opendev.org/c/openstack/ansible-config_template/+/938507 | 08:28 |
| jrosser | `ERROR! Unexpected Exception, this is probably a bug: No module named 'distutils'` | 08:28 |
| noonedeadpunk | it feels that networkd misses some requirement for that... as `ip link` works nicely | 08:29 |
| noonedeadpunk | and it creates interfaces... | 08:29 |
| jrosser | i was trying to reproduce that https://paste.opendev.org/show/blq6VfiPvDUv1Rjlu42U/ | 08:30 |
| noonedeadpunk | um... I didn't have such issue | 08:32 |
| noonedeadpunk | it passes nicely for me up to tests | 08:33 |
| noonedeadpunk | * to verify | 08:33 |
| noonedeadpunk | and fails on first assert, as DOWN interfaces do not end up in facts | 08:34 |
| jrosser | i am on a noble vm if that makes any difference | 08:35 |
| noonedeadpunk | I'm on noble laptop.... | 08:35 |
| noonedeadpunk | and you run `tox -e molecule` ? | 08:36 |
| jrosser | i had exactly the same trouble with molecule in the config_template repo too | 08:36 |
| jrosser | yes | 08:36 |
| noonedeadpunk | oh, well. I think I know - I have venv with python3.11 for tox | 08:37 |
| noonedeadpunk | :D | 08:37 |
| jrosser | the cleanup in config_template feels legit - its for very very old stuff | 08:38 |
| noonedeadpunk | python3.11 is available on noble and as I was upgrading from jammy - venv left on it... | 08:38 |
| noonedeadpunk | yeah. true | 08:38 |
| noonedeadpunk | but then the role is pulling master config_template | 08:38 |
| noonedeadpunk | as part of requirements | 08:39 |
| jrosser | so thats an interesting thing - we can't really depends-on the config-template change | 08:39 |
| noonedeadpunk | nope.. | 08:41 |
| noonedeadpunk | but also there's a problem with noble docker image kind of | 08:42 |
| jrosser | that two seperate things, host OS vs image? | 08:45 |
| noonedeadpunk | yeah | 08:45 |
| jrosser | hrrm so to make working on any of this less painful we need to patch to openstack-ansible and config_template landing | 08:46 |
| jrosser | this https://review.opendev.org/c/openstack/openstack-ansible/+/938220 | 08:47 |
| jrosser | and https://review.opendev.org/c/openstack/ansible-config_template/+/938507 | 08:47 |
| noonedeadpunk | ok, so networkd working on ubuntu - not debian... doh | 08:52 |
| noonedeadpunk | ubuntu jammy though | 08:52 |
| noonedeadpunk | so it's obviously missing some parts | 08:53 |
| noonedeadpunk | which are shipped with normal image | 08:53 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Do not try to configure resolved when it's not available https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938512 | 08:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Don't run systemd task in async https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938515 | 08:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 08:57 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Remove support for python2 and ansible < 2.9 https://review.opendev.org/c/openstack/ansible-config_template/+/938507 | 08:59 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 08:59 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Do not try to configure resolved when it's not available https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938512 | 09:19 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Don't run systemd task in async https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938515 | 09:19 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 09:19 |
| noonedeadpunk | centos also fails.... so only ubuntu jammy is passing networkd right now | 09:32 |
| noonedeadpunk | jrosser: have you seen that? https://www.phoronix.com/news/Fedora-Maybe-Retire-GlusterFS | 09:40 |
| jrosser | oh, no i had not | 09:41 |
| noonedeadpunk | huge part is - `CentOS Storage SIG decided to not build GlusterFS packages for CentOS Stream 10` | 09:41 |
| noonedeadpunk | sounds like time to replace gluster.... | 09:42 |
| jrosser | that is indeed an interesting choice of what to use | 09:43 |
| noonedeadpunk | the only idea I have right now - is minio.... | 09:46 |
| noonedeadpunk | and it could replace apache as well I guess | 09:46 |
| jrosser | i wonder if thats sufficient to make pip install from it | 09:47 |
| jrosser | as S3 is not exactly the same as web browser -> directory of files | 09:47 |
| noonedeadpunk | it's s3 kinda... | 09:47 |
| noonedeadpunk | but it very close to it? as it also can allow listings if bucket is public? | 09:48 |
| jrosser | but yes minio probably has good life in it as it's used a lot as stand-in for real S3 in lots of cases | 09:48 |
| noonedeadpunk | and it seems to have bucket replication in it | 09:50 |
| noonedeadpunk | well, we can do s3fs mount at worst :D | 09:50 |
| noonedeadpunk | I wonder how bad is lustre.... | 09:59 |
| noonedeadpunk | git looks alive at least: https://git.whamcloud.com/?p=fs/lustre-release.git;a=shortlog | 10:00 |
| noonedeadpunk | but probably minio is easier, as at least it stores data as regaular files on the storage, so we likely can even preserve current repo server layout completely | 10:09 |
| noonedeadpunk | got systemd debug logs for networkd: https://paste.openstack.org/show/bGrUfpjnV2vyQ8QhCH3r/ | 11:36 |
| noonedeadpunk | I wonder if that's requiring udev | 11:38 |
| noonedeadpunk | "link pending udev initialization" | 11:40 |
| noonedeadpunk | yeah. it's udev... | 11:58 |
| jrosser | we probably have clues to this in the lxc hosts role when preparing the lxc images | 12:01 |
| jrosser | huh https://github.com/openstack/openstack-ansible-lxc_hosts/commit/ee28a80de0ff81ddc96b9bb8a82806bce417cf4f | 12:02 |
| noonedeadpunk | well. it somehow feels that container restart is required on udev installation | 12:07 |
| noonedeadpunk | or maybe not | 12:07 |
| noonedeadpunk | ok, we'd need to discuss if we want to do all that in-role or specifically in molecule script... | 12:10 |
| noonedeadpunk | I'll push an update now... | 12:10 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 12:27 |
| noonedeadpunk | ok, so basically question - what we do with these? https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517/4/tests/prepare.yml | 12:29 |
| noonedeadpunk | and I think it's better to change in a follow-up anyway | 12:30 |
| jrosser | thats not obvious somehow | 12:36 |
| jrosser | as one way we add quite some more tasks to a role thats used many times | 12:36 |
| jrosser | and how much we actually support the role inside docker for real use is another question | 12:37 |
| noonedeadpunk | the thing is that these requirements of role are not obvious at all either | 12:37 |
| noonedeadpunk | policycoreutils needed only for this for instance https://opendev.org/openstack/ansible-role-systemd_networkd/src/branch/master/tasks/main.yml#L48 | 12:38 |
| noonedeadpunk | and could be that better to add condition `ansible_facts['selinux']['status'] == 'enabled'` instead | 12:39 |
| jrosser | we already have systemd_networkd_distro_packages tbh, so maybe extending those lists with the mandatory dependancies is not so bad | 12:40 |
| noonedeadpunk | now when I have mlecule passing - I can play with this | 12:43 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 12:59 |
| opendevreview | Merged openstack/ansible-config_template master: Remove support for python2 and ansible < 2.9 https://review.opendev.org/c/openstack/ansible-config_template/+/938507 | 13:03 |
| jrosser | noonedeadpunk: i think you need ansible.utils in the systemd_networkd requirements.yml | 13:18 |
| jrosser | (for ansible.utils.ipaddr) | 13:19 |
| noonedeadpunk | ah, yes, good catch | 14:18 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 14:19 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 14:20 |
| opendevreview | Merged openstack/openstack-ansible master: Add centralized defenition for molecule jobs https://review.opendev.org/c/openstack/openstack-ansible/+/938220 | 14:27 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 14:39 |
| jrosser | there is some unusual behaviour with diff: true for config_template ^ | 14:41 |
| jrosser | that passes locally but i'm not sure if the tests i've "fixed" are particularly valid | 14:42 |
| noonedeadpunk | iirc diff doesn't really work for config_template | 14:42 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Install role pre-requisite packages https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938560 | 14:48 |
| noonedeadpunk | ok, so installation packages is not too bad ^ | 14:48 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 14:50 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Do not try to configure resolved when it's not available https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938512 | 14:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Don't run systemd task in async https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938515 | 14:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 14:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Install role pre-requisite packages https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938560 | 14:51 |
| jrosser | small typo in the commit message "hanled" | 14:52 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Install role pre-requisite packages https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938560 | 14:52 |
| * noonedeadpunk looking at config_template | 14:59 | |
| spatel | noonedeadpunk hey! morning | 14:59 |
| spatel | did you ever use GPU with Magnum k8s stuff? | 14:59 |
| noonedeadpunk | nope, I did not | 14:59 |
| spatel | My developer wants to do some AI stuff on k8s and asking for GPU :) | 14:59 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 15:00 |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 15:00 |
| opendevmeet | Meeting started Tue Jan 7 15:00:41 2025 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_ansible_meeting' | 15:00 |
| noonedeadpunk | #topic rollcall | 15:00 |
| jrosser | o/ | 15:01 |
| noonedeadpunk | o/ | 15:01 |
| * jrosser will be 5 mins | 15:01 | |
| noonedeadpunk | #topic office hours | 15:04 |
| noonedeadpunk | so we're having a good progress for migration from tests repo to molecule testing | 15:05 |
| noonedeadpunk | #link https://review.opendev.org/q/topic:%22osa/molecule%22 | 15:05 |
| noonedeadpunk | I've just issued re-checks for failing ones, as patch to integrsated repo was merged | 15:06 |
| jrosser | it looks pretty reasonable tbh | 15:06 |
| noonedeadpunk | there're were hiccups, like I have no idea how to run swapon inside docker, so commented out that test at all | 15:07 |
| jrosser | though some things will be harder/impossible to test in docker | 15:07 |
| noonedeadpunk | yeah, quite some time went down the pipe for systemd-networkd | 15:07 |
| noonedeadpunk | from other side - we can run multiple containers and connect them with multiple networks | 15:07 |
| noonedeadpunk | so technically - we can do some kind of mnaio in the future.... | 15:08 |
| noonedeadpunk | another thing that had a progress - httpd role | 15:09 |
| noonedeadpunk | #link https://review.opendev.org/q/topic:%22osa/httpd_role%22 | 15:09 |
| NeilHanlon | o/ | 15:10 |
| noonedeadpunk | repo server does have a circular dependency with integrated repo | 15:10 |
| noonedeadpunk | and some small portion of love needed for skyline and centos/rocky | 15:10 |
| noonedeadpunk | NeilHanlon: o/ - good that you've joined :) | 15:11 |
| NeilHanlon | heyo :) happy new year | 15:11 |
| noonedeadpunk | I didn't look for httpd role wrt horizon and keystone yet. But I think a base is relatively solid now | 15:11 |
| noonedeadpunk | challanges there would be installing extra packages which provides extra modules (federation) | 15:12 |
| noonedeadpunk | and next thing we need to start thinking about is EL10 | 15:13 |
| noonedeadpunk | just today I've found out, that Storage SIG decided not to build gluster for centos, which is kind of a biggie for us | 15:14 |
| noonedeadpunk | with that I was thinking, if NeilHanlon maybe you can make some kind of copr repo for gluster for now?:) | 15:14 |
| NeilHanlon | ah yes.. i saw that as well. i think I saw a fedora packager that I am friendly with picked it up in Fedora (as RH disbanded from it there, too...) | 15:15 |
| NeilHanlon | maybe i am misremembering though | 15:15 |
| NeilHanlon | my brain still waking up | 15:15 |
| NeilHanlon | nope... https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/D4QN2OFKQWN66HVKUG4CG4XPRW5ON62Z/#7JR2A5XTBFMQU2B6TW7XBEORF5O6OCXA | 15:15 |
| NeilHanlon | anyways. yes, i will come up with a plan :) | 15:15 |
| noonedeadpunk | while we're thinking on a more long-term solution... like minio or smth.... | 15:16 |
| NeilHanlon | i feel like it's probably not going away any time soon | 15:17 |
| noonedeadpunk | but I'd really prefer not to do that this cycle while having EL10 support in | 15:17 |
| NeilHanlon | yeah we have enough with modular libvirt and such | 15:18 |
| noonedeadpunk | indeed | 15:18 |
| noonedeadpunk | btw, any idea about rocky timeline for 10? | 15:19 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-repo_server master: Use FQCN for modules https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/938272 | 15:20 |
| NeilHanlon | right now the only blockers are that we have to redo some stuff on how we sign secureboot artifacts because EL10 ripped out OpenSSL "Engine" support | 15:20 |
| NeilHanlon | and our signing relies on pkcs11 stuff | 15:20 |
| noonedeadpunk | wait, what? | 15:21 |
| NeilHanlon | https://discussion.fedoraproject.org/t/f41-change-proposal-disable-openssl-engine-support-system-wide/107511 | 15:21 |
| noonedeadpunk | I kind of wonder if that affects our pki role | 15:23 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 15:23 |
| NeilHanlon | i don't think it should | 15:23 |
| NeilHanlon | but, maybe | 15:24 |
| noonedeadpunk | but probably not... not sure about things like barbican with HSM though | 15:24 |
| noonedeadpunk | as it;s mainly relying on PKCS#11 as well | 15:24 |
| NeilHanlon | yeah I think that is more the possible problematic points. we're at the intersection of HSM for signing which I don't even know that RH has an answer for at this point | 15:24 |
| noonedeadpunk | but ok, I see. thanks for the update! | 15:24 |
| noonedeadpunk | and I don't think they're willing to help either... | 15:25 |
| NeilHanlon | of course! and I'm also still wanting to get mirrors setup of the content for opendev this year.. i took on too much the end of last year | 15:25 |
| noonedeadpunk | btw, as we've switched from mirrorlist to the baseurl - we don't see broken mirrors anymore :) | 15:26 |
| jrosser | fwiw barbican will talk directly to the vendor pkcs11 .so file, there should not be any openssl involved there | 15:29 |
| jrosser | its a python <> C interface | 15:30 |
| noonedeadpunk | yeah, right | 15:30 |
| noonedeadpunk | ok, then I think it's good | 15:30 |
| noonedeadpunk | NeilHanlon: btw, another question related to our molecule efforts. you publish rocky docker images only on dockerhub? | 15:33 |
| noonedeadpunk | ah, I found on quay already, lol | 15:34 |
| noonedeadpunk | #link https://quay.io/repository/rockylinux/rockylinux?tab=tags | 15:34 |
| noonedeadpunk | I wonder if we actually can just install systemd there at prepare step.... | 15:36 |
| jrosser | doesnt it need to be the entrypoint | 15:36 |
| * jrosser handwaves | 15:36 | |
| noonedeadpunk | well, I'm doing docker describe right now, and it seems that entrypath is `/sbin/init` | 15:38 |
| noonedeadpunk | but I think you're right and regular one may just not startup | 15:41 |
| jrosser | https://github.com/eniocarboni/docker-rockylinux-systemd/blob/main/Dockerfile | 15:43 |
| noonedeadpunk | yeah, I see | 15:43 |
| NeilHanlon | if you need systemd you can maybe use the ubi image? | 15:43 |
| NeilHanlon | https://github.com/rocky-linux/sig-cloud-instance-images/blob/main/Containerfile-init | 15:44 |
| noonedeadpunk | well, we want to test debian, ubuntu and some el | 15:44 |
| noonedeadpunk | aha | 15:44 |
| noonedeadpunk | let me check that :) | 15:44 |
| NeilHanlon | interesting, looks like we aren't publishing a 9.5-ubi-init. i will have to check on that... | 15:45 |
| NeilHanlon | or maybe we don't, i forget. need to redo all this soon lol | 15:45 |
| noonedeadpunk | there is 9-ubi-init | 15:46 |
| NeilHanlon | yeah i think we just don't publish the ubi variants with a minor rev | 15:47 |
| NeilHanlon | because they're descendants of the 9.5.20241118-ubi | 15:47 |
| NeilHanlon | anyways yea that image should work alright for applications | 15:49 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-config_template master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-config_template/+/938513 | 15:51 |
| noonedeadpunk | doh, there's no sudo even inside :D | 15:51 |
| noonedeadpunk | but yes, otherwise it's working nicely! | 15:52 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-pki master: Add molecule testing https://review.opendev.org/c/openstack/ansible-role-pki/+/831236 | 15:53 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add RockyLinux to molecule testing https://review.opendev.org/c/openstack/openstack-ansible/+/938571 | 15:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 15:59 |
| noonedeadpunk | #nedmeeting | 16:00 |
| noonedeadpunk | #endmeeting | 16:00 |
| opendevmeet | Meeting ended Tue Jan 7 16:00:48 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:00 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-07-15.00.html | 16:00 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-07-15.00.txt | 16:00 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-07-15.00.log.html | 16:00 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_mount master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_mount/+/938489 | 16:02 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_service master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_service/+/938492 | 16:04 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Install role pre-requisite packages https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938560 | 16:05 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-frrouting master: Use centralized requirements for molecule testing https://review.opendev.org/c/openstack/ansible-role-frrouting/+/938222 | 16:07 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-frrouting master: Use centralized requirements for molecule testing https://review.opendev.org/c/openstack/ansible-role-frrouting/+/938222 | 16:39 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-pki master: Add molecule testing https://review.opendev.org/c/openstack/ansible-role-pki/+/831236 | 16:40 |
| noonedeadpunk | seems like config_template is fixed? | 16:43 |
| jrosser | yes i had quite a big fight with it :) | 16:45 |
| noonedeadpunk | would be interesting to look into diff part kinda | 16:45 |
| jrosser | it might be a bit rough in places as a pushed what i had to avoid some accident and losing it | 16:45 |
| jrosser | my suspicion is that the old tests ran localhost/connection: local and that made some special behaviour | 16:46 |
| jrosser | that does not occur with an actual remote target | 16:46 |
| noonedeadpunk | yeah, could be, but we also do trigger service restarts on config changes, so idempotency there kind of important | 16:47 |
| jrosser | the config_template code has some conditional open('file') vs. calling slurp which would certainly behave differently | 16:47 |
| jrosser | and i think that is the root cause of the failure to evaulate {{ baz }} | 16:47 |
| jrosser | with -vvvvv it was failing in the use of slurp rather than in config_template itself but it is exceedingly hard to debug it | 16:48 |
| jrosser | the other one was this https://github.com/openstack/ansible-config_template/blob/master/tests/templates/test.ini#L6 | 16:50 |
| jrosser | that is descrbied as being removed by the diff output, making the task be changed | 16:50 |
| jrosser | but without diff it is not changed | 16:50 |
| jrosser | so there is something odd happening there, but i did not investigate further yet | 16:51 |
| noonedeadpunk | > with -vvvvv it was failing in the use of slurp rather than in config_template - ah, that would make sense, yes | 16:51 |
| jrosser | ^ yes so I think that test failure is triggered by the not local connection | 16:51 |
| noonedeadpunk | but test.ini#L6 s kind of more concerning... but I think it can be followed-up later on | 16:52 |
| jrosser | right - i did not know what the expectation was - if the empty section should be preserved or removed | 16:53 |
| noonedeadpunk | as we don't see anything too obvious being wrong wrt idempotency with config_template right now anyway | 16:53 |
| noonedeadpunk | that's a good question... Not sure either, but it should not result in changed task if nothing is actually changed | 16:54 |
| noonedeadpunk | and if it always removing section, then result is not changed consistently | 16:54 |
| jrosser | we can WIP this for now and take a closer look? | 16:55 |
| noonedeadpunk | any fixes to module still should be in follow-ups, imo | 16:55 |
| jrosser | though tbh we do not change the code/behaviour, just that the test before did not test idempotence at all | 16:55 |
| noonedeadpunk | yeah | 16:56 |
| noonedeadpunk | so it's not getting different from current state. just indicator that we have potenitally fishy behaviour in module | 16:56 |
| jrosser | yep | 16:56 |
| noonedeadpunk | so we can land this and interate over | 16:56 |
| mgariepy | kleini, https://github.com/prometheus-community/ansible/pull/474 look at : https://github.com/prometheus-community/ansible/pull/507 | 19:13 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!