16:01:31 <evrardjp> #startmeeting openstack_ansible_meeting
16:01:33 <openstack> Meeting started Tue Aug 29 16:01:31 2017 UTC and is due to finish in 60 minutes.  The chair is evrardjp. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:35 <cloudnull> woot!
16:01:37 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:01:37 <evrardjp> #topic rollcall
16:01:40 <spotz> \o/
16:01:42 <cloudnull> presente
16:01:47 <cloudnull> :P
16:02:10 <evrardjp> wow, I didn't know you were a transgender (in french)
16:02:32 <evrardjp> anyway...
16:03:32 <openstackgerrit> Major Hayden proposed openstack/openstack-ansible-tests master: [TEST] Test dnf in host/containers  https://review.openstack.org/495371
16:03:34 <evrardjp> #topic this week's bugs
16:03:36 <evrardjp> 20 of them!
16:03:46 <evrardjp> let's go go go
16:03:46 <evrardjp> https://bugs.launchpad.net/openstack-ansible/+bug/1713663
16:03:48 <openstack> Launchpad bug 1713663 in openstack-ansible "Set enable_proxy_headers_parsing = True when HAProxy is used" [Undecided,New]
16:04:01 <evrardjp> Adri2000: submitted that bug
16:04:17 <evrardjp> it sounds legit, and I agree that this should be something in oslo if possible
16:04:25 <evrardjp> however that bug in oslo is from 2016...
16:04:36 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron stable/ocata: Correct the "dns_domain" option by default  https://review.openstack.org/495896
16:04:40 <evrardjp> should we implement a workaround?
16:04:47 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron stable/ocata: Simplify the string check for offline db migrations  https://review.openstack.org/494614
16:04:58 <andymccr> hmm
16:06:45 <mhayden> wow, centos tempest tests take 25 minutes, only 7 for ubuntu
16:06:49 <mhayden> i wonder if somethin' broke in there
16:06:54 <andymccr> im not too sure what the best approach is there.
16:07:00 <evrardjp> mhayden: please, it's bug triage :p
16:07:11 <evrardjp> andymccr: yes it's a nice bug.
16:07:12 * mhayden ducks
16:07:34 <evrardjp> Adri2000: do you have time to change that upstream in oslo?
16:07:41 <andymccr> it makes sense with haproxy - but then i dont think we should check.
16:08:05 <evrardjp> so you mean we should override it in all cases?
16:08:23 <evrardjp> and let the deployer have an override to not do it, in some cases?
16:08:28 <andymccr> im not sure - it makes sense with haproxy enabled, but i dnt think we should have an "haproxy_enabled" check or anything
16:08:48 <andymccr> maybe we could check if the public endpoint is https?
16:08:56 <andymccr> not sure if that makes sense though either]
16:09:15 <evrardjp> well it could work in this case
16:09:34 <evrardjp> but it's not necessary that this will be the solutions, it's just fixing in one case
16:09:49 <andymccr> well if https is setup does that not imply you either have haproxy or your own LB setup? since we only setup those https bits in the haproxy role
16:09:51 <odyssey4me> I thought we already had a var for this?
16:10:09 <evrardjp> for oslo?
16:10:36 * odyssey4me is hunting
16:10:37 <evrardjp> yeah, but you could have the same kind of headers to parse if not http
16:10:43 <evrardjp> if http
16:10:45 <odyssey4me> openstack_external_ssl
16:11:05 <odyssey4me> which then gets used in group vars for each role which has it: nova_external_ssl
16:11:07 <evrardjp> I don't know what oslo parses to be honest
16:11:39 <odyssey4me> https://github.com/openstack/openstack-ansible-os_nova/blob/d2841072ef899de80b30a34d724aea3998ea9e1b/templates/nova.conf.j2#L287-L289
16:11:42 <andymccr> so under what examples do we not want to pass headers?
16:12:14 <odyssey4me> ok, it seems that's a different var?
16:12:20 <andymccr> yeah
16:12:32 <odyssey4me> anyway, I don't see why we can't do the same thing - but we have a group_var set which uses the inventory understanding to set it
16:12:32 <evrardjp> for me, like Adri2000 said, if we set these headers, it's for something.
16:12:36 <odyssey4me> which can be overridden
16:13:02 <odyssey4me> it's a way of doing a 'sensible default'
16:13:20 <odyssey4me> like this: https://github.com/openstack/openstack-ansible/blob/master/group_vars/cinder_all.yml#L50
16:13:48 <evrardjp> https://github.com/openstack/oslo.middleware/blob/master/oslo_middleware/http_proxy_to_wsgi.py#L56-L76
16:14:27 <evrardjp> I think for the triage of this bug, we should ack it
16:14:39 <evrardjp> how we fix it, is up to the committer indeed.
16:14:54 <evrardjp> I think an oslo change would be advisable, but apparently I am the only one to think that :p
16:15:06 <evrardjp> confirmed and high?
16:15:27 <evrardjp> or medium, because it "only" impacts designate?
16:15:56 <odyssey4me> well, there is a workaround - just use the config override
16:16:01 <evrardjp> it's workaroundable by setting a conf override
16:16:03 <evrardjp> ahaha
16:16:04 <odyssey4me> easy enough to implement
16:16:07 <evrardjp> yeah
16:16:10 <odyssey4me> so yeah, medium
16:16:59 <evrardjp> ok next
16:17:00 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1712889
16:17:01 <openstack> Launchpad bug 1712889 in openstack-ansible "lbaasv2 ports not updated on liberty to mikata upgrade" [Undecided,New]
16:17:10 <evrardjp> won't fix, too old?
16:18:54 <evrardjp> next
16:18:55 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1712741
16:18:56 <openstack> Launchpad bug 1712741 in openstack-ansible "[lxc_container_create] sysctl-container.service: Failed at step CGROUP spawning /sbin/sysctl: No such device" [Undecided,New]
16:19:16 <evrardjp> in process so I'll mark it in progress.
16:19:19 <evrardjp> assign to markos
16:19:43 <evrardjp> nobody said no, so I move on
16:19:51 <evrardjp> #next https://bugs.launchpad.net/openstack-ansible/+bug/1712619
16:19:52 <openstack> Launchpad bug 1712619 in openstack-ansible "neutron plugins : - dns in user_variables causes neutron to stop working" [Undecided,New]
16:20:34 <evrardjp> DimGR: are you there?
16:20:38 <DimGR> yes
16:20:52 <evrardjp> It's your bug, is that an instruction problem?
16:20:56 <evrardjp> I am not sure right there
16:21:21 <evrardjp> because of https://github.com/openstack/openstack-ansible-os_neutron/blob/master/defaults/main.yml#L252 ?
16:22:08 <DimGR> the thing is with ocata 15.1.0 if you enable dns plugin in neutron and rerun the neutron install role , the role will finish with no errors but neutron will be useless . nothing related to that will work , i tested from horizon and from client .  i did a minor upgrade to stable/ocata and everything is fine
16:23:50 <evrardjp> is that due to the docs?
16:24:45 <DimGR> i dont think this is a doc error  , i think its a neutron bug ?
16:25:14 <DimGR> or should there be a warning saying - dns plugin is not for ocata 15.1.0 ?
16:25:20 <evrardjp> Adri2000: are you there?
16:25:31 <evrardjp> you tested designate, so I guess you tested this
16:25:53 <DimGR> cloudnull  can also confirm this evrardjp  we were looking at this together
16:26:03 <jmccrory> are neutron services not starting? are there traceback messages or errors in logs that you could add to the bug report?
16:27:16 <evrardjp> oh so, stable ocata is fine?
16:27:22 <openstackgerrit> Major Hayden proposed openstack/openstack-ansible-os_tempest master: [TEST] Gate check  https://review.openstack.org/498859
16:28:59 <evrardjp> ok let's move on
16:29:03 <DimGR> i will now jmccrory
16:29:09 <evrardjp> if stable/ocata is fine
16:29:10 <DimGR> stable ocata is fine evrardjp
16:29:27 <evrardjp> ok then it's a temp bug, so let's forget it
16:29:29 <DimGR> should i paste some log errors or not
16:29:30 <DimGR> ok
16:29:42 <evrardjp> nah if it's fixed already it's fine :)
16:30:27 <evrardjp> next
16:30:28 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711376
16:30:29 <openstack> Launchpad bug 1711376 in openstack-ansible "intermittent AIO error: Timeout (7s) waiting for privilege escalation prompt" [Undecided,New]
16:31:03 <evrardjp> I will wait for Markos update, to know if that should fix it
16:31:11 <evrardjp> ok everyone?
16:31:50 <evrardjp> next
16:31:52 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711347
16:31:54 <openstack> Launchpad bug 1711347 in openstack-ansible "Config file override mechanism broken" [Undecided,New]
16:32:03 <evrardjp> we discussed this last week
16:32:26 <evrardjp> http://eavesdrop.openstack.org/meetings/openstack_ansible_meeting/2017/openstack_ansible_meeting.2017-08-22-16.03.log.html#l-198
16:33:27 <openstackgerrit> Matt Thompson proposed openstack/openstack-ansible-ops master: [WIP] Use separate interfaces for lb traffic  https://review.openstack.org/498862
16:33:29 <evrardjp> andymccr: what's your decision about that?
16:33:47 <evrardjp> Should we revert python3, ripping it out?
16:34:01 <andymccr> hmm yeah we spoke about that i think it might be best
16:34:06 <andymccr> all we have right now is some bootstrap shenanigans
16:34:10 <andymccr> and if its causing issues then we should bail
16:34:14 <evrardjp> it reduces the importance of the bug
16:34:34 <evrardjp> ok
16:34:48 <andymccr> but im not too sure why thats actually happening aside from "its related to py3 in some way"
16:35:39 <evrardjp> it makes sense to me to revert to python2 only, introduce tests for python3 in our plugins repo
16:35:51 <andymccr> agreed
16:35:53 <evrardjp> then going a little further with that later
16:35:54 <evrardjp> ok
16:36:07 <evrardjp> andymccr: would you be ok at doing the revert?
16:36:16 <andymccr> evrardjp: i will look into reverting that sure
16:36:20 <evrardjp> ok
16:36:40 <evrardjp> I will evaluate the tests for the plugins repo then
16:36:49 <evrardjp> we can mark that into wishlist then
16:36:54 <evrardjp> ok?
16:37:02 <andymccr> ok cool
16:37:15 <evrardjp> I'll mark in the bug that python3 is on our wishlist
16:38:06 <odyssey4me> if we change this, and only this, it should do the trick? https://github.com/openstack/openstack-ansible/blob/master/scripts/bootstrap-ansible.sh#L102
16:38:33 <andymccr> jamesdenton: if we put a patch in re https://bugs.launchpad.net/openstack-ansible/+bug/1711347 - would you be able to test?
16:38:34 <openstack> Launchpad bug 1711347 in openstack-ansible "Config file override mechanism broken" [Undecided,New]
16:38:40 <andymccr> and confirm
16:39:05 <evrardjp> that would be great.
16:40:48 <evrardjp> ok let's continue then
16:40:50 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711341
16:40:51 <openstack> Launchpad bug 1711341 in openstack-ansible "Client Error: Not Found for url" [Undecided,New]
16:41:43 <openstackgerrit> Andy McCrae proposed openstack/openstack-ansible stable/pike: Remove python3 usage in bootstrap-ansible.sh  https://review.openstack.org/498863
16:41:48 <nomatics> I changed /etc/openstack_deploy/env.d/cinder-volume.yml to is_metal: false-- however,  setup-openstack.yml is still trying trying to install cinder on the controller. Do I need to make additional changes?
16:42:24 <nomatics> My storage settings in openstack_user_config are under storage-infra_hosts
16:42:29 <evrardjp> I'll mark this as incomplete
16:42:51 <evrardjp> If we can't reproduce and the bug disappeared
16:43:32 <andymccr> ok ive not seent hat personally
16:43:37 <nomatics> Do i need to change it in /opt/openstack_deploy/playbooks/inventory/env.d/cinder.yml as well?
16:43:51 <odyssey4me> nomatics no
16:44:06 <evrardjp> ok let's move on
16:44:13 <odyssey4me> nomatics we're busy with bug triage right now, please wait until afterwards
16:44:19 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711335
16:44:21 <openstack> Launchpad bug 1711335 in openstack-ansible "Playbook fails to create required directory" [Undecided,New]
16:44:47 <nomatics> thanks, i'll check back
16:47:30 <evrardjp> confirmed, low and low-hanging-fruit
16:47:32 <evrardjp> ok?
16:49:37 <evrardjp> next
16:49:38 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711309
16:49:39 <openstack> Launchpad bug 1711309 in openstack-ansible "Add documentation for rolling upgrades" [Undecided,New]
16:49:40 <evrardjp> docs!
16:49:54 <evrardjp> odyssey4me: again?
16:50:20 <evrardjp> that would be great if we discuss that a working session during ptg?
16:50:39 <evrardjp> because it seems interesting to express what we did, but also improve our documentation coverage
16:51:24 <evrardjp> in the meantime, I'd like to confirm, and set to medium or high
16:51:46 <evrardjp> what do you think andymccr?
16:52:52 <andymccr> yeah it would be good to get docs on that if somebody has the time
16:53:00 <evrardjp> spotz: ?
16:53:04 <andymccr> we have done quite some work on that, so its worthwhile
16:53:11 <evrardjp> agreed.
16:53:21 <evrardjp> What about doing that at the PTG?
16:53:25 <spotz> evrardjp: Just put notes in I'll work on it around the dog show this weekend
16:53:33 <evrardjp> as a part of a retrospective session?
16:53:40 <evrardjp> haha ok
16:54:02 <evrardjp> why not, but it's maybe more than a few notes in odyssey4me's brain
16:54:39 <spotz> heheh
16:54:57 <openstackgerrit> Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/newton: migrate_openstack_vars.py: Correct folder name  https://review.openstack.org/498871
16:55:50 <evrardjp> odyssey4me: would you rather draft notes in the bug for spotz to understand what's in your brain, or a work session?
16:56:11 <spotz> hehe
16:56:14 <evrardjp> (the former may be more efficient)
16:56:19 <odyssey4me> maybe I should push up a first draft, then let spotz make it pretty
16:56:28 <evrardjp> that would work too.
16:56:29 <spotz> that'll work
16:56:44 <evrardjp> I'll assign it to you then
16:56:50 <odyssey4me> I haven't yet figured out what I think it should contain or look like and I need a little time to sit down and stare at it for a while.
16:56:54 <odyssey4me> so yeah, assign to me
16:57:01 <odyssey4me> thanks spotz :)
16:57:18 <evrardjp> last one for today I guess:
16:57:19 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1711108
16:57:20 <openstack> Launchpad bug 1711108 in openstack-ansible "Ocata designate service role documentation 404" [Undecided,New]
16:57:51 <evrardjp> spotz: ?
16:57:51 <andymccr> hmm
16:58:00 <odyssey4me> hm, no patches to designate role for ocata... so it hasn't been republished after the move
16:58:02 <evrardjp> could you have a look at the docs wiring there?
16:58:17 <odyssey4me> if we merge *any* patches to the roles it should republish
16:58:23 <andymccr> that applies to all roles
16:58:25 <spotz> evrardjp: Yeah
16:58:27 <andymccr> im not sure that is the correct link
16:58:34 <odyssey4me> oh hang on
16:58:55 <odyssey4me> https://docs.openstack.org/openstack-ansible-os_designate/ocata/
16:59:35 <odyssey4me> this is actually a redirect bug
16:59:36 <evrardjp> docs redirect bug indeed
16:59:41 <odyssey4me> for the manuals repo
17:01:04 <evrardjp> ok let's wrap up for today
17:01:09 <evrardjp> thanks everyone!
17:01:14 <evrardjp> #endmeeting