16:02:55 <evrardjp> #startmeeting openstack_ansible_meeting 16:02:56 <andymccr> love the influx of +2's ont hat patch :P 16:02:56 <openstack> Meeting started Tue Mar 14 16:02:55 2017 UTC and is due to finish in 60 minutes. The chair is evrardjp. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:59 <openstack> The meeting name has been set to 'openstack_ansible_meeting' 16:03:00 <asettle> yo yo 16:03:29 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672728 16:03:29 <openstack> Launchpad bug 1672728 in openstack-ansible "keystone fails to install, missing libgsasl-devel on CentOS 7" [Undecided,New] 16:03:29 <stevelle> hoorah 16:03:30 <cloudnull> yipie! 16:03:33 <admin0> if i want to check telemetry, shall i build a nuwton aio or ocata 16:03:40 <admin0> newton* 16:03:41 <evrardjp> centos is broken right now, right? 16:03:50 <andymccr> evrardjp: nope 16:03:52 <evrardjp> oh 16:04:01 <cloudnull> evrardjp: i just saw a review where it was working 16:04:02 <evrardjp> but at some point it could have been? 16:04:19 <andymccr> it was broken because of upstream repos 16:04:23 <evrardjp> it's 15.0.0 16:04:25 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/ocata: Provide example of using veth pairs for br-vlan https://review.openstack.org/445551 16:04:28 <cloudnull> in the fullness of time i believe yes, it has been broken 16:04:42 <odyssey4me> hahaha 16:04:46 <evrardjp> thanks cloudnull. 16:04:48 <evrardjp> so 16:04:50 * cloudnull is a helper 16:04:51 <odyssey4me> in the fullness of time, we have also been broken 16:04:52 <evrardjp> how do we triage this? 16:05:02 <evrardjp> odyssey4me: talk about yourself bro 16:05:19 <andymccr> evrardjp: hmm - i wonder if it was because of timing - the mess around with the RDO/EPEL work 16:05:30 <evrardjp> that's what I think too. 16:05:41 <evrardjp> let's ask if it's still broken and move on 16:05:43 <andymccr> either way that is a required patch 16:05:46 <andymccr> so our gates should fail if it wasnt there 16:05:57 <mgariepy> fixed by 3 backports 16:05:58 <andymccr> yeah i'll do a SHA bump this week 16:06:02 <andymccr> mgariepy: hahaa yes! 16:06:04 <mgariepy> https://review.openstack.org/#/c/443682/ 16:06:04 <mgariepy> https://review.openstack.org/#/c/444282/ 16:06:04 <mgariepy> https://review.openstack.org/#/c/443681/ 16:06:06 <mgariepy> those. 16:06:12 <evrardjp> ok 16:06:13 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/newton: Provide example of using veth pairs for br-vlan https://review.openstack.org/445556 16:06:21 <andymccr> which reminds me i want to add decent testing to those repos :P 16:06:53 <evrardjp> :) 16:06:59 <evrardjp> andymccr: action? 16:07:12 <andymccr> anyway for this bug - lets ask for more info and say we'll SHA bump this week 16:07:15 <andymccr> or i'll update it 16:07:25 <andymccr> next! 16:07:27 <evrardjp> ok 16:07:37 <evrardjp> cool thanks 16:07:45 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672662 16:07:45 <openstack> Launchpad bug 1672662 in openstack-ansible "Documentation bug: incorrect link to securing services with SSL page" [Low,New] - Assigned to Alexandra Settle (alexandra-settle) 16:08:00 <evrardjp> asettle: ? 16:08:07 <asettle> IM BACK 16:08:07 <asettle> YEs 16:08:08 <asettle> HELLO 16:08:09 <asettle> I did 16:08:12 <asettle> I assigned that to me 16:08:12 <evrardjp> if a phrase is incorrect in english, I can't do anything. 16:08:13 <asettle> Yes that's valid 16:08:20 <asettle> IT's a broken link 16:08:31 <asettle> S'all good 16:08:31 <evrardjp> ok. 16:08:32 <asettle> I'l do it now 16:08:37 <evrardjp> confirmed low 16:08:38 <andymccr> i thought we fixed all those. gah 16:08:39 <asettle> I've confirmed 16:08:43 <asettle> Yeah nah, this one is brand new 16:08:45 <asettle> So fresh 16:08:46 <asettle> Wonderful 16:08:47 <asettle> Brb 16:08:56 <evrardjp> thanks for the confirmation and triage 16:08:57 <evrardjp> next 16:09:00 <asettle> No sweat 16:09:02 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672648 16:09:02 <openstack> Launchpad bug 1672648 in openstack-ansible "Documented test setup fails due to HTTPS usage in proxy" [Undecided,New] 16:09:07 <evrardjp> rah 16:09:09 <evrardjp> pasting. 16:09:21 <evrardjp> oh no it's correct link 16:09:26 <evrardjp> 2 documentation bugs! 16:09:29 <asettle> Yep 16:09:32 <asettle> There's 3 actually 16:09:33 <asettle> Anyway 16:09:36 <asettle> This is number 2 16:09:48 <asettle> I want to see if this could be tested a little bit more before we cry wolf 16:09:51 <asettle> But you're already on it evrardjp 16:10:25 <evrardjp> ahah. I just asked for questions, but that's fine I can take it to confirm if it's real documentation bug or misunderstanding 16:10:28 <evrardjp> next 16:10:39 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672632 16:10:39 <openstack> Launchpad bug 1672632 in openstack-ansible "EPEL is missing from repo container. lsyncd requires EPEL. " [Undecided,New] 16:10:49 <asettle> Thanks evrardjp :) 16:10:50 <evrardjp> Temporary issue I guess. 16:10:54 <mgariepy> same as keystone one. 16:11:18 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-galera_server master: Rename vars/common.yml to vars/main.yml https://review.openstack.org/445041 16:11:23 <evrardjp> Ok I linked to the first one 16:11:52 <andymccr> sweet 16:12:08 <evrardjp> next 16:12:10 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672604 16:12:10 <openstack> Launchpad bug 1672604 in openstack-ansible "physical_skel containers group has unexpected hosts" [Undecided,New] 16:12:57 <openstackgerrit> Merged openstack/openstack-ansible-os_barbican master: Remove git_dest var and extras folder https://review.openstack.org/445370 16:13:19 <evrardjp> if possible, I'd like to leave from this conversation and let jmccrory alone on this one. Or calling for help @palendae or cloudnull 16:13:22 <jmccrory> not sure how long it's been like that 16:13:26 <evrardjp> :p 16:13:53 <openstackgerrit> Alexandra Settle proposed openstack/openstack-ansible master: [DOCS] Fixes broken link https://review.openstack.org/445561 16:14:05 * cloudnull reading the bug 16:14:06 <evrardjp> that is weird though 16:14:23 <andymccr> jmccrory: there is some funkyness, i had issues with ironic and getting that to work - otherwise it creates odd groups. personally i still hate the way that whole thing works. 16:15:03 <jmccrory> andymccr: yeah, not a fan either. tried troubleshooting with octavia patch, but it's all very confusing 16:15:10 <palendae> Hrm, yeah, not exactly sure why that's happening 16:15:19 <evrardjp> ok if someone can just confirm that would be great 16:15:38 <cloudnull> me neither. 16:15:41 <palendae> I'd love to redo the whole thing, but don't think a one-fell-swoop approach will work 16:15:44 <evrardjp> I never doubt jmccrory but the first step is to reproduce the bug :) 16:15:48 <stevelle> seems like it was introduced by a refactor? 16:15:51 <andymccr> jmccrory: i ended up having to create like a "ironic-infra_hosts" or "memcaching_hosts" group in this case. i think it does some funny business with the "_containers" groups 16:16:02 <odyssey4me> with the vars plugin that just merged we can have host_vars and group_vars in user space, so maybe it's time that we moved away from env.d files 16:16:14 <evrardjp> andymccr: oh so you got that same consistency issue? 16:16:29 <evrardjp> can we mark it as confirmed? 16:16:32 <andymccr> evrardjp: to be fair that was like liberty/mitaka cycle so quote some time ago 16:16:40 <evrardjp> yeah the inventory changed 16:16:41 <odyssey4me> I think this is a different issue to the conflicting host group bug (another bug) 16:16:46 <palendae> Yeah 16:16:50 <evrardjp> jmccrory: is that master? 16:16:50 <andymccr> well let me confirm on an AIO 16:17:00 <jmccrory> evrardjp : yes, that was against master 16:17:07 <evrardjp> ok. 16:17:12 <palendae> I think what andymccr's tallking about has been very long standing 16:17:16 <evrardjp> thanks andymccr for the confirmation attempt and classification :) 16:17:29 <palendae> There's some clash with naming groups directly after their service 16:17:29 <jmccrory> i can go back through branches later today and see if or when behavior changed 16:17:32 <evrardjp> let's move on, worse case we discuss this next week. 16:17:46 <evrardjp> well that could be interesting jmccrory 16:17:52 <evrardjp> At least for the bug report 16:18:12 <odyssey4me> jmccrory the one andymccr is talking about is this one, which I think is different to yours: https://bugs.launchpad.net/openstack-ansible/+bug/1623584 16:18:12 <openstack> Launchpad bug 1623584 in openstack-ansible "Dynamic inventory clash between host group name and container group name" [Low,Confirmed] - Assigned to Nolan Brubaker (nolan-brubaker) 16:18:12 <jmccrory> yeah, think we can move on for now. something to look further into later 16:18:13 <evrardjp> let's continue to fight against the clock 16:18:14 <andymccr> how do we want to prioritize this 16:18:22 <evrardjp> when confirmed? 16:18:24 <evrardjp> andymccr: ^ 16:18:26 <andymccr> yeah 16:18:29 <andymccr> i can confirm it 16:18:33 <andymccr> i'll paste into the issue but im not sure priority 16:18:45 <andymccr> we dont really use those groups so im thinking its not so much a big deal? 16:18:50 <evrardjp> let's do low, because we don't fully rely on this for our playbooks right? 16:18:56 <evrardjp> agreed andymccr 16:19:05 <evrardjp> still painfull so not wishlist :p 16:19:07 <evrardjp> -l 16:19:25 <evrardjp> let's continue the fight 16:19:34 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672565 16:19:34 <openstack> Launchpad bug 1672565 in openstack-ansible "Heat SSL issues when internal and external VIPs are https" [Undecided,New] - Assigned to Basil Baby (basilbaby) 16:20:48 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_barbican stable/ocata: Remove git_dest var and extras folder https://review.openstack.org/445562 16:21:14 <openstackgerrit> Merged openstack/openstack-ansible-os_keystone master: Revert "Install EPEL for keystone role" https://review.openstack.org/444361 16:21:19 <evrardjp> not sure what's going on there 16:21:31 <evrardjp> I don't know the haproxy_server role well :( 16:22:19 <odyssey4me> it probably requires the implementation of the insecure flag if not real certs are used 16:23:16 <evrardjp> #link http://git.openstack.org/cgit/openstack/puppet-heat/commit/?id=a60abccabbbba8aff6e53b2c64d98fb8a1a6739c 16:23:48 <cloudnull> its also entirely possible that heat has broken self signed ssl ( it would not be the first time an OpenStack service did so ). 16:23:50 <evrardjp> maybe worth setting a group var to define this 16:24:16 <evrardjp> interesting 16:24:23 <odyssey4me> we don't have a test for ssl internal, nor do I think the services will all work like that anyway as they're not setup for ssl 16:24:24 <evrardjp> could someone confirm this? 16:24:36 <odyssey4me> so I expect that's a very deep hole of misery 16:24:54 <odyssey4me> we could perhaps implement a periodic test for ssl all around at some point 16:25:36 <evrardjp> confirmed wishlist right now? 16:25:55 <evrardjp> that's maybe too low 16:26:00 <cloudnull> odyssey4me: i think internal ssl for everything will become a lot easier once we get our uwsgi all the things story complete. 16:26:17 <evrardjp> yup 16:26:29 <evrardjp> so how do we classify this? 16:26:42 <cloudnull> wishlist imo 16:27:02 <evrardjp> because we have this 16:27:05 <evrardjp> #link https://github.com/openstack/openstack-ansible-os_heat/blob/92e38306e9898c799a432ac3b3b8720bad502604/templates/heat.conf.j2#L69-L72 16:27:26 <evrardjp> so for me it would be simple to fix it with and no https :) 16:29:00 <openstackgerrit> Merged openstack/openstack-ansible-galera_server master: Force clean yum metadata when repo config changes https://review.openstack.org/445533 16:29:39 <evrardjp> ok let's move on, there is no more conversation about this 16:29:42 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1672452 16:29:42 <openstack> Launchpad bug 1672452 in openstack-ansible "Trove nova proxy unable to authenticate to keystone v3" [Undecided,New] 16:30:48 <evrardjp> It may be a duplicate of this: https://bugs.launchpad.net/openstack-ansible/+bug/1647681 16:30:48 <openstack> Launchpad bug 1647681 in openstack-ansible "trove error during publish_exists_event" [Undecided,In progress] - Assigned to Donovan Francesco (donovan-francesco) 16:31:21 <evrardjp> do you agree? 16:31:50 <jmccrory> https://github.com/openstack/openstack-ansible-os_trove/blob/master/defaults/main.yml#L47 16:31:51 <jmccrory> https://github.com/openstack/openstack-ansible-os_trove/blob/master/templates/trove.conf.j2#L4 16:31:54 <jmccrory> maybe that 16:32:01 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-galera_server stable/ocata: Force clean yum metadata when repo config changes https://review.openstack.org/445564 16:32:14 <jmccrory> look like same problem in both bugs 16:33:28 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Remove "tenant_id" from the nova endpoint https://review.openstack.org/444270 16:33:40 <evrardjp> Ok I'm linking them 16:34:11 <evrardjp> next 16:34:14 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1671958 16:34:14 <openstack> Launchpad bug 1671958 in openstack-ansible "Newton: repo build still broken with optional components (swift)" [Undecided,New] 16:35:55 <evrardjp> I don't think we are forcing swift by default 16:36:19 <evrardjp> I always test with an AIO that has it, so hard to know 16:36:33 <andymccr> evrardjp: i think the issue seems to be if you are using an external swift (using glance as an example) 16:36:55 <andymccr> you need the swift libs but htey didnt get built because there are no swift hosts. 16:36:56 <evrardjp> oh yes, it could make sense 16:36:59 <odyssey4me> so any client components can be added to the utility container - the only reason for the failure here is because 'swift' is in the package list 16:37:04 <evrardjp> or anything that requires swift tooling installed 16:37:18 <odyssey4me> swift is not downloadable via pypi 16:37:29 <admin0> scripts/scripts-library.sh: line 195: tracepath: command not found .. https://docs.openstack.org/developer/openstack-ansible/developer-docs/quickstart-aio.html - where did the minimal packages required section go ? 16:37:49 <admin0> ./scripts/bootstrap-ansible.sh: line 108: virtualenv: command not found 16:37:53 <andymccr> i guess the real issue is there is a disconnect between what is defined as being required and not skipped, and what we actually require/will install at some point 16:37:55 <odyssey4me> quite honestly, turn off selective repo building if you have this situation - or improve the way your requirements work 16:38:02 <andymccr> odyssey4me: agree 16:38:07 <andymccr> there is no way for us to know all your package requirements 16:38:08 <admin0> i recall seeing a minimal packages that i used to blindly copy-paste :D 16:38:24 <evrardjp> admin0: we are triaging, we can come back to your issue later. 16:38:25 <andymccr> if you want to improve the way it sets up the repo - for efficiency, you can use the selective build, if not - don't. 16:38:31 <admin0> oops ! 16:38:42 <admin0> sorry 16:38:45 <evrardjp> np 16:39:01 <odyssey4me> andymccr yeah, unless someone comes up with a creative way to implement user space rules for the selective build 16:39:02 <evrardjp> So it's invalid, dependant of the use case 16:39:08 <andymccr> i think its also impossible for us to know all packages you may require to install at some point? 16:39:28 <evrardjp> well I guess the user can add the package in its group vars or whatever... 16:39:30 <evrardjp> WAIT 16:39:32 <evrardjp> :p 16:39:35 <andymccr> wait! 16:39:35 <odyssey4me> as far as I'm concerned, the issue is specific to this deployers environment - I'm inclined to say won't fix 16:39:49 <andymccr> odyssey4me: hold up - evrardjp is offering to fix it 16:39:54 <andymccr> so lets mark critical confirmed and assign to evrardjp 16:40:04 <odyssey4me> lol 16:40:07 <evrardjp> that's a stab in the back. 16:40:12 <andymccr> :D 16:40:20 <odyssey4me> the issue is actually that the git source isn't added to the list of things to compile 16:40:39 <odyssey4me> the git clone is selective, as is the requirements list 16:40:43 <andymccr> perhaps we could have a specific "additional packages to install" var that could be set? i dunno. 16:40:48 <andymccr> white-list 16:41:02 <odyssey4me> it may be plausible to build something which picks up if you have a package in the list which is only available via git source 16:41:07 <odyssey4me> but this is an edge case of note 16:41:09 <andymccr> still wishlist. 16:41:41 <odyssey4me> the issue is not the package in the list, it's the git source not being cloned, and even if it was cloned, it's not going to be compiled 16:41:44 <evrardjp> I'll mark it as confirmed wishlist (I'm not sure if that's the case), and link to our conversation 16:43:12 <evrardjp> let's move on then 16:43:27 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1671795 16:43:27 <openstack> Launchpad bug 1671795 in openstack-ansible "setting rabbitmq_policy failes during initial deployment" [Undecided,New] 16:44:20 <evrardjp> oh 16:44:23 <evrardjp> fun with hostnames 16:45:17 <evrardjp> the master shouldn't have ansible_hostname set with _ 16:45:31 <andymccr> evrardjp: isnt it more that the task happens in a bad order? 16:45:45 <andymccr> e.g. it tries to do a task that only works on a running system which wont be up yet on a new build 16:45:46 <evrardjp> nah 16:45:52 <evrardjp> node name: 'rabbitmq-cli-49@infra1-rabbit-mq-container-0e654ad7' 16:45:57 <evrardjp> attempted to contact: ['rabbit@infra1_rabbit_mq_container-0e654ad7'] 16:46:31 <andymccr> he fixed it by moving the order - so the node_name wouldnt have changed in that? 16:48:25 <evrardjp> I doubt it has anything to do with that 16:49:00 <evrardjp> it's not even restarting rabbit 16:49:20 <andymccr> hmm 16:49:23 <evrardjp> I think because of the multiple attempts maybe something has changed in the env 16:49:29 <andymccr> we fixed the hostname stuff ages ago didnt we? 16:49:32 <andymccr> i wonder what version that is 16:49:52 <andymccr> because the container name seems correct 16:50:30 <evrardjp> I think that the command used is wrong 16:50:42 <openstackgerrit> Merged openstack/openstack-ansible master: Remove git_dest variables https://review.openstack.org/445243 16:50:53 <evrardjp> so it's normal that it answers "don't do this" 16:51:03 <andymccr> hmm 16:51:11 <openstackgerrit> Merged openstack/openstack-ansible master: Introduce userspace group_vars and host_vars https://review.openstack.org/445437 16:51:17 <openstackgerrit> Merged openstack/openstack-ansible master: [DOCS] Fixes broken link https://review.openstack.org/445561 16:52:02 <andymccr> there is something weird htere because the hostname on the container seems to change? 16:52:37 <evrardjp> or maybe because HOSTNAME is set in the env 16:53:20 <andymccr> ask for more info i guess 16:53:30 <evrardjp> I'd leave it as is and I already asked some questions 16:53:31 <andymccr> im not too sure what the bug is attempting to do and how the hostname has changed like that 16:53:34 <andymccr> oh ok 16:53:36 <evrardjp> let's see. 16:53:37 <andymccr> excellent! 16:53:54 <evrardjp> just right now, it's not like if it was done 5 minutes ago :p 16:53:57 <evrardjp> next! 16:54:12 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1671525 16:54:12 <openstack> Launchpad bug 1671525 in openstack-ansible "Support different container_vars when affinity > 1" [Undecided,New] 16:54:30 <evrardjp> Done. 16:54:33 <evrardjp> let's move on 16:54:42 <andymccr> wishlist on that anyway 16:54:47 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1671496 16:54:47 <openstack> Launchpad bug 1671496 in openstack-ansible "bridge interface added to physical_interface_mappings, if infra = compute node" [Undecided,New] 16:54:49 <evrardjp> andymccr: yes 16:55:36 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/ocata: [DOCS] Fixes broken link https://review.openstack.org/445576 16:56:37 <evrardjp> I think we can already do it with our system 16:56:48 <evrardjp> we just have to use a veth 16:57:07 <evrardjp> odyssey4me: cloudnull: didn't we talk about documenting this recently? 16:58:12 <evrardjp> guys it's aweful, it's been two weeks we can't catch up on bugs 16:58:14 <odyssey4me> evrardjp you're thinking of https://review.openstack.org/444971 16:58:31 <evrardjp> yes! 16:58:38 <andymccr> evrardjp: yeah i will try take some time to at least triage some of the remaining bugs this week so we can keep on top 16:59:07 <evrardjp> ok 16:59:32 <evrardjp> well I can continue this, and I will probably try to spend more time too, you don't have to carry that burden alone andymccr 16:59:46 <evrardjp> it's just a reminder 16:59:52 <andymccr> evrardjp: i think its less bad than it looks because we ahve looked at some of those bugs already - but we always leave the same final 10 or so :P 17:00:03 <openstackgerrit> Merged openstack/openstack-ansible-galera_server master: Rename vars/common.yml to vars/main.yml https://review.openstack.org/445041 17:00:49 <evrardjp> ok let's stop for today 17:00:51 <evrardjp> thanks everyone 17:00:54 <evrardjp> #endmeeting