13:01:18 #startmeeting tripleo 13:01:19 Meeting started Thu Jul 18 13:01:18 2019 UTC and is due to finish in 60 minutes. The chair is cloudnull. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:20 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:22 The meeting name has been set to 'tripleo' 13:01:29 #topic rollcall 13:01:42 who's not here :D 13:02:11 i have some news regarding molecule and rhel8 13:02:59 very cool, lets make sure to raise that in open discussion. 13:03:07 sure 13:03:11 for a reminder 13:03:14 #link https://storyboard.openstack.org/#!/project/openstack/tripleo-ansible 13:03:15 #link https://etherpad.openstack.org/p/tripleo-ansible-agenda 13:03:38 etherpad and storyboard project 13:03:52 weshay: o/ 13:04:00 #link https://storyboard.openstack.org/#!/board/174 13:04:10 and our general work board. 13:04:19 give folks a few more minutes to trickle in 13:05:21 oh didn't notice that storyboard could replace trello! :D 13:05:46 ++ 13:05:59 shameless plug https://cloudnull.io/2019/06/trello-to-storyboard-migration/ 13:06:01 weshay: yep (sorry, on mtg until now). let's see how it goes with designate. 13:06:07 gchamoul ^ 13:06:24 anyway, getting on with it 13:06:41 #topic recap from past weeks 13:06:55 Tengu you can add some services as well that may be better for validations 13:07:03 previous ACTION: ekultails to work on getting tripleo-persist into tripleo-ansible 13:07:06 cloudnull: impressive! 13:07:09 ekultails hows that going? 13:08:12 looks like there is a review for it 13:08:14 #link https://review.opendev.org/#/c/670362/ 13:08:28 though its in merge conflict, is there anything we can do to help that along ? 13:08:56 previous ACTION: cloudnull to start working on porting the hiera roles to tripleo-ansible 13:09:42 ^ I've held back becuase the hiera roles are being worked on in tripleo-common. once we've stabalized those existing roles I'll work on porting them -cc EmilienM slagle 13:09:56 Jose Luis Franco proposed openstack/tripleo-heat-templates master: Force removal of docker container in tripleo-docker-rm. https://review.opendev.org/670796 13:10:02 Guys, pretty please merge this: https://review.opendev.org/#/c/624325/ 13:10:05 panda, ^ 13:10:24 previous ACTION: sshnaidm to look into ARA. 13:10:27 cloudnull: feel free to move it 13:10:28 is that still on-hold at this point? 13:10:38 cloudnull: I haven't been able to address slagle's review yet 13:10:39 cloudnull, started to play with it in molecule jobs, but no review still 13:11:01 cloudnull: if you move it, I'll rework my patch on top of yours, so no biggies 13:11:16 EmilienM ++ I just didn't want to impede progress, however I can get on that ASAP. 13:11:32 +1 13:11:32 #ACTION cloudnull to start porting hiera roles, for real this time 13:11:48 Gael Chamoulaud proposed openstack/python-tripleoclient master: Add plan parameter to run_ansible_playbook function https://review.opendev.org/671289 13:12:16 once these three roles are moved over, we're out of tripleo-common 13:12:17 #link https://github.com/openstack/tripleo-common/tree/master/roles 13:12:27 gchamoul: ok, I can open an issue against libpod "filter". Could prove the docker version does work with "foo|bar". 13:12:42 and ekultails has taken care of the persist role. 13:12:45 so good progress. 13:12:55 sshnaidm cool 13:13:00 anything we can help with on that front ? 13:13:09 Tengu: like I said, we use it since Rocky so with docker. 13:13:12 cloudnull, not right now 13:13:40 ++ 13:13:53 cloudnull, I think we need to consider a standard way to add plugins/modules/whatever to ansible run in tripleo, not sure if it exists.. 13:14:19 cloudnull, ara is not last plugin we'll want to use 13:14:39 we have that here https://github.com/openstack/tripleo-ansible/tree/master/tripleo_ansible/ansible_plugins for plugins we make 13:15:13 I think sshnaidm means a place where we control which ones we enable in a centralized way? 13:15:26 what EmilienM said ^ 13:16:12 currently ara is installed in CI in mistral container in hacky way 13:16:19 it should change 13:16:32 ok I see, we need an interface to configure which plugins we want to enable when deploying tripleo 13:16:40 no matter it's standalone, undercloud or overcloud 13:16:51 yeah, install them, their dependencies and enable 13:17:09 because plugin have also dependencies we should care about them 13:17:09 so for undercloud it could be a new parameter in undercloud.conf, and in overcloud/standalone, a heat parameter 13:17:17 which would lay down the config 13:17:33 we already expose being able to provide the ansible.cfg, so what's missing? 13:17:45 gchamoul: https://github.com/containers/libpod/issues/3596 13:17:46 also we don't expect customers to add plugins 13:17:55 mwhahaha, I think installation and dependencies 13:18:03 should be rpms in standard locations 13:18:04 we could edit the ansible action script to do that ? 13:18:07 #link https://github.com/openstack/tripleo-common/blob/master/tripleo_common/actions/ansible.py#L52 13:18:09 if they aren't, we're doing it wrong 13:18:14 mwhahaha, ara doesn't have rpm afaik 13:18:23 ara won't be shipped 13:18:32 Tengu: ++ 13:18:32 so that's something specific to us and not part of tripleo proper 13:18:53 also ara 1.x is problematic so i don't really want to add ara support right now 13:19:00 mwhahaha, yes, but we still want it 13:19:05 i think that's an ara problem not tripleo 13:19:22 -cc dmsimard 13:19:43 mwhahaha, it can work with <1 version like we have now in CI 13:20:13 mwhahaha, and maybe to enable it for undercloud eventually 13:20:29 after we get rid of mistral maybe 13:20:31 and then this is moot 13:20:41 i wouldnt' spend time trying to fix the mistral integration 13:20:45 poor mistral. nobody's loving it 13:20:49 lol 13:21:03 which is why it's such a hack to enable 13:21:10 when do we plan to get rid of it? 13:21:15 * d0ugal ducks 13:21:18 soonish (tm) 13:21:19 Giulio Fidente proposed openstack/tripleo-common master: Bump ceph/daemon container image to v4.0.1 https://review.opendev.org/671482 13:21:30 we're looking into it now 13:21:42 never liked it 13:22:08 point being, we should spend the resources to fix why we have this problem rather than working around it :D 13:23:01 ok, so leaving the old way of injecting ara in CI? 13:23:06 for now 13:23:11 ack 13:23:29 and what with other possible fancy plugins? 13:23:41 we revisit when we have new fancy plugin 13:24:09 sshnaidm if its something we're shipping with tripleo I think we would just need to add them to 13:24:12 #link https://github.com/openstack/tripleo-ansible/tree/master/tripleo_ansible/ansible_plugins 13:24:29 unless its something that has to be specifically enabled, like a call back 13:24:39 then we'd have to edit the tripleo action config too 13:25:03 cloudnull, ok 13:25:06 things like action plugins, libraries, etc would get automatically picked up if we put them in that path 13:25:51 we do have a few plugins, and they're doc'd 13:25:55 #link https://docs.openstack.org/tripleo-ansible/latest/modules.html 13:26:10 cloudnull, well, we can't just take plugin of somebody and put in our repo 13:26:14 I think 13:26:30 ++ correct, I think that would be bad form 13:26:32 is it's just python package 13:26:42 in order for it to be in tripleo it needs to be packageds 13:26:50 and should use standard paths that we can add 13:26:58 mwhahaha, in case we ship it 13:27:04 and it's part of the product 13:27:16 the plugins in tripleo-ansible were all written by us. 13:27:21 unless you have a specific example i think we're going to keep going around in circles on this 13:27:29 ++ 13:27:33 yeah, agree, let's move on 13:27:48 we'll talk about it when I find next fancy callback plugin 13:27:52 :) 13:27:54 #topic THT 2 TripleO-Ansible 13:27:59 Here is our current work-lists 13:27:59 #link https://storyboard.openstack.org/#!/worklist/685 13:27:59 #link https://storyboard.openstack.org/#!/worklist/711 13:28:07 I started on a couple services/roles, which could use some reviews 13:28:07 #link https://review.opendev.org/#/q/topic:podman-role+(status:open+OR+status:merged) 13:28:07 #link https://review.opendev.org/#/q/topic:time-role+(status:open+OR+status:merged) 13:28:18 but there are A LOT of things in those lists 13:28:25 so help would be greatly appreciated. 13:29:02 unless folks want to talk more about that, I'd like to move right in to Open Discussion 13:29:16 #topic Open Discussion 13:29:27 zbr what to kick us off with the RHEL 8 bits ? 13:30:28 EmilienM new role for config things ? 13:30:34 o/ 13:30:49 I'm currently exploring the replacement of Puppet by Ansible to generate config files 13:31:08 we started the process to import ansible-config_template in RDO 13:31:28 and I plan to create a basic role in tripleo-ansible to consume that plugin 13:31:46 that module* 13:32:05 #link https://review.rdoproject.org/r/#/q/(topic:ansible_config_template+OR+topic:pkg-config_template)+(status:open+OR+status:merged) 13:32:33 this role would be the interface for THT to generate the configuration files 13:32:33 thank you chandankumar and others for helping to move that along 13:33:00 I'll give more updates when I have actual progress. I just wanted to share the plan 13:33:35 cool! 13:33:43 * EmilienM done 13:34:09 Martin Schuppert proposed openstack/tripleo-heat-templates master: Only run cellv2 host discovery on default cell https://review.opendev.org/671486 13:34:16 #link https://review.opendev.org/#/c/669672/ 13:34:18 cloudnull: re rhel8, I added some comments on the document. if we write molecule.yml files with overridable vars inside, we can test on alternative platforms. 13:34:20 to EmilienM role 13:35:09 zbr mind walking us through some of what you were thinking ? 13:36:06 sadly i do not have something ready, but you can get an idea from https://review.rdoproject.org/r/#/c/21489/11/molecule/default/molecule.yml 13:36:29 in tripleo-ansible we do have a couple roles running with the delegated driver, is there something we need to change to make what your thinking go ? 13:36:44 if we put a single platform at platforms, and use envvars, we can create additional jobs that override default platform. 13:37:11 oic 13:37:34 so then we'd have 1 job per OS? 13:37:51 Alex Schultz proposed openstack/tripleo-common master: Skip standalone deploys if tox jobs fail https://review.opendev.org/671071 13:38:01 in most cases, at this point, that would be CENT7 and FED28 (for tripleo-ansible) 13:38:47 folks zbr is currently on the phone with tripleo-ci squad for sprint planning he might be a bit slow to respond for few mins 13:39:02 * mnaser lurks 13:39:15 np 13:39:31 sshnaidm want to talk about meeting times? 13:39:48 cloudnull, yeah :D we're double booked 13:40:09 cloudnull, I thought about voting on doodle or kind of 13:40:35 there was a vote here in the last IRC meeting, however, nobody, except me, voted. 13:40:39 Alex Schultz proposed openstack/tripleo-common master: Revert "Linting hardening via pre-commit" https://review.opendev.org/671056 13:41:03 #link http://eavesdrop.openstack.org/meetings/tripleo/2019/tripleo.2019-07-11-13.00.html 13:41:07 item 6. 13:41:24 however we can do that again? or do we want to send something to the ML 13:42:50 Alex Schultz proposed openstack/tripleo-common master: Set container=oci for our containers https://review.opendev.org/670574 13:43:35 cloudnull, whatever works, but not changing this after voting like we did :) 13:44:09 hahaha. 13:44:48 Gael Chamoulaud proposed openstack/python-tripleoclient master: Don't use local connection by default for the validations https://review.opendev.org/671491 13:44:49 EmilienM, cloudnull: just fyi i'm going to be reviewing and clearing up all open ansible-config_template reviews, feel free to ping me if you need any review work 13:44:53 cloudnull, we can submit a patches with times and people can +1/-1 them 13:45:54 #action cloudnull to send a message to the ML w/ a doodle so we can nail down an appropriate meeting time that works for more folks. 13:45:56 * mnaser has a small thing to bring up after the time discussion 13:46:25 sshnaidm I'll take it to the ML so everyone has a chance to participate 13:46:38 * cloudnull hands mic to mnaser 13:46:42 cloudnull, ack, maybe using like that: https://doodle.com/poll/2hb42hhq678m9f5k 13:46:45 ++ 13:46:48 sshnaidm ++ 13:47:19 so one of the things that's awesome is that we're sharing a lot of stuff together (yay), but i feel like an odd thing seems like one group or another has 'ownership' over the stuff we collaborate one 13:47:52 so triploe consumes ansible-config_template and os_tempest but its technically "openstack-ansible" yet no one from tripleo team is in openstack-ansible so logistically i feel like that might be limiting from time to time 13:48:33 i'm wondering if there maybe a good idea in possibly taking s ome openstack-ansible repos and having them with a shared core-wnership from tripleo-ansible and openstack-ansible 13:48:58 we don't consume ansible-config_template yet btw, we're in the process to look at it :D 13:49:10 but yeah 13:49:44 well just in general, yeah, as an overall idea 13:49:57 +1 for shared ownership 13:49:59 EmilienM, we use it on os_tempest 13:50:01 mnaser that's a good idea 13:50:07 Definitely +1 13:50:48 Alex Schultz proposed openstack/tripleo-heat-templates stable/stein: Add facter cache for container configurations https://review.opendev.org/671492 13:51:02 totally ok with any other proposals but given the "openstack-ansible" seems like a pretty .. good namespace for shared workspace, i think it might be good to keep it and maybe create an acl called openstack-ansible-collab and have both OSA cores and tripleo-ansible cores in there? 13:51:08 arxcruz ++ hopefully we can start using the shiny new package for the config_template plugin soon :) 13:52:29 is still OSA core on all the things 13:52:32 * cloudnull hides 13:52:49 :P 13:52:50 mnaser, I think we don't have "tripleo-ansible" cores, tripleo has a shared cores space 13:53:15 mnaser, but could be a tripleo-ansible squad 13:53:31 I think there is a project config for it, though im not sure anyone is in the group though 13:53:33 we can create a tripleo-ansible group 13:53:35 no big deal 13:53:46 i guess as long as we try to 'share' ownership (i.e. like at least have an OSA core 'ok' a patch alongside a tripleo one) that would be nice 13:53:49 mwhahaha, sshnaidm, cloudnull: let's take a few mins to sync about the status of ara 1.x after meeting :) 13:53:49 might help with some of the ansible review thingies 13:54:01 but i think we all have been around enough to trust each other not to break eachother :D 13:54:04 dmsimard, sure 13:54:17 (and hopefully that encourages use of more cross-project jobs so we dont accidentally break each other anyways) 13:54:50 mwhahaha https://review.opendev.org/#/admin/projects/openstack/tripleo-ansible,access - so maybe we just need to add people to that group? 13:55:06 dmsimard ++ 13:55:09 will need to check if that's actually in use 13:55:10 but yea 13:55:23 tripleo-ansible-core seems to have all of tripleo inside of it 13:55:32 mnaser ++ 13:55:32 which tbh, i dont really mind as long as people arent just merging random stuff anyways :p 13:55:46 * cloudnull broke the gate yesterday with my config_template patch .... 13:55:58 so not to complicate things, ill push up a patch for openstack-ansible-collab and add both tripleo-ansible-core and openstack-ansible-core to it 13:55:58 we'll just blame cloudnull 13:56:14 a cross project gate would have been super useful 13:56:21 mwhahaha that's fair 13:56:23 cloudnull, https://www.youtube.com/watch?v=SrDSqODtEFM 13:56:26 and then maybe ansible-config_template or os_tempest could be the first target to add 13:56:32 does that sound good? :> 13:56:45 arxcruz https://twitter.com/cloudnull/status/1151588948608802816 13:56:58 cloudnull, i see that :) 13:57:02 :D 13:57:08 mnaser ++ 13:57:22 ok ill try to do this today 13:57:26 any takers to work on that ? 13:57:28 cool ! 13:57:32 ill do the acl-y stuff 13:58:02 #action mnaser to see about setting up acl-y stuff for cross project thing-a-me's 13:58:18 ok welp, we're just about out of time. anything else ? 13:58:53 OK well, great meeting all! 13:58:59 #endmeeting