16:00:07 <evrardjp> #startmeeting openstack_ansible_meeting 16:00:08 <openstack> Meeting started Tue Oct 3 16:00:07 2017 UTC and is due to finish in 60 minutes. The chair is evrardjp. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:12 <openstack> The meeting name has been set to 'openstack_ansible_meeting' 16:00:15 <andymccr> o/ 16:00:16 <evrardjp> we do have a lot of bugs. 16:00:18 <hwoarang> o/ 16:00:32 <evrardjp> let's start directly 16:00:33 <evrardjp> # topic new bugs 16:00:39 <evrardjp> #topic new bugs 16:00:43 <spotz> o? 16:00:50 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1721037 16:00:51 <openstack> Launchpad bug 1721037 in openstack-ansible "cinder-api in master role test fails" [Undecided,New] 16:00:57 <evrardjp> this one pained me a lot 16:01:15 <evrardjp> Either I am too tired to find why, or I am not clever enough. 16:01:44 <andymccr> im taking a look 16:01:58 <evrardjp> it's on master, so it means it could also be coming from a bug in upstream cinder 16:02:05 <andymccr> i think it might be 16:02:46 <evrardjp> if I want to override the sha for testing 16:03:08 <evrardjp> I should replace in the tests/whatever_overrides.yml the cinder_git_install_branch , right? 16:03:21 <andymccr> yeah that'd work 16:03:58 <evrardjp> that's what I did, re-using a 1 month old reference from cinder. It still was broken. At that time we had a job in os_cinder passing. 16:04:09 <evrardjp> I am very confused. 16:04:24 <evrardjp> has someone seen that before? 16:04:35 <evrardjp> if not let's not mark it critical yet 16:04:37 <andymccr> you mean https://review.openstack.org/#/c/504015/ 16:04:48 <andymccr> which passed like 2 weeks ago 16:05:23 <logan-> does the reference to "key_manager" there imply barbican integration of some sort is active? 16:05:37 <logan-> maybe we're not seeing it in some places because we don't have it enabled everywhere? 16:05:47 <andymccr> hmm 16:05:52 <andymccr> wasnt sure we set that up in cinder tests 16:05:55 <evrardjp> I took a reference from even older 16:06:27 <evrardjp> mmm that is maybe in api-paste 16:07:36 <evrardjp> nope no ref of that 16:07:37 <evrardjp> anyway 16:07:43 <andymccr> hmm 16:08:02 <evrardjp> if you reproduce that, we should classify this as critical I guess 16:08:13 <evrardjp> if that's the case, and I am not insane, I'll speak with the cinder team 16:08:28 <evrardjp> ok for everyone? 16:08:51 <logan-> "cls = importutils.import_class(conf.key_manager.backend)" -> ValueError: Empty module name 16:08:52 <logan-> https://github.com/openstack/openstack-ansible-os_cinder/blob/a4a190b907643a437a480f2a0cec6c2de388c8e6/templates/cinder.conf.j2#L125-L131 16:08:53 <evrardjp> andymccr: could you attempt to triage it on your side, see if I'm right? 16:09:00 <andymccr> evrardjp: yeah taking a look now i have the same problem 16:09:04 <logan-> it is looking for a key called "backend" and we are giving it as "api_class" 16:09:12 <andymccr> ok 16:09:13 <andymccr> so maybe 16:09:17 <andymccr> we need to adjust our version of castellan or w/e 16:09:29 <evrardjp> I don't see that in my generated .conf 16:09:35 <andymccr> https://github.com/openstack/cinder/commit/e75be5d90519094fca3ee475b906e7c2fe1d09fd 16:09:52 <evrardjp> ahah good catch andymccr! 16:09:54 <logan-> nice 16:09:55 <logan-> :) 16:10:14 <andymccr> trying something quick 16:10:41 <andymccr> hmm nope 16:10:45 <andymccr> using master of all those projects didnt really fix it 16:11:11 <andymccr> but that seems like it is related 16:11:22 <evrardjp> nope but we may have found a first path to it 16:11:23 <evrardjp> thanks 16:11:26 <jmccrory> yeah, option was renamed https://github.com/openstack/cinder/blob/master/releasenotes/notes/castellan-backend-0c49591a54821c45.yaml#L11-L13 16:11:54 <evrardjp> well how does that impact us if we're not following that code path is another topic ... 16:12:06 <andymccr> im not convinced we even use that opt 16:12:07 <andymccr> so its confusing 16:12:14 <logan-> yeah 16:12:24 <andymccr> unless they set a default somewhere 16:12:27 <evrardjp> we are not. so I am not the only one finding this weird. 16:12:36 <evrardjp> oh... yeah code defaults 16:12:41 <evrardjp> completely forgot that 16:12:56 <evrardjp> that is very weird though 16:13:02 <evrardjp> ok how do we triage this? 16:13:09 <evrardjp> confirmed/critical? 16:13:31 <evrardjp> (it will block everything..., but currently, weirdly, doesn't seem to block the integrated aio) 16:13:45 <evrardjp> (probably because we haven't bumped pins of cinder in a month) 16:13:59 <andymccr> evrardjp: there must be some depdenency that hasnt changed yet 16:14:07 <evrardjp> yeah 16:14:15 <evrardjp> so triage? 16:14:24 <andymccr> still critical imo 16:14:35 <andymccr> we cant move forward with cinder if we sha bump it will break id say 16:14:46 <evrardjp> ok next 16:14:49 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1721014 16:14:50 <openstack> Launchpad bug 1721014 in openstack-ansible "Include bind9utils during designate installation when needed" [Undecided,New] 16:15:25 <evrardjp> sounds valid and low 16:15:34 <hwoarang> i agree 16:15:42 <Tahvok> That's mine :P 16:15:53 <evrardjp> haha 16:16:02 <evrardjp> welcome Tahvok! 16:16:07 <evrardjp> yeah sounds valid 16:16:29 <evrardjp> if you want we can help you on the way to fix it after the bug triage 16:16:39 <evrardjp> I suggest we continue in the meantime, many bugs 16:16:45 <evrardjp> next! 16:16:46 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1720535 16:16:47 <openstack> Launchpad bug 1720535 in openstack-ansible "lxd.filters out of date" [Undecided,New] 16:16:54 <evrardjp> confirmed 16:17:03 <Tahvok> Is it possible to check a list of hashes for a specific variable in ansible? 16:17:26 <evrardjp> not sure what you mean but we can discuss that after triage Tahvok 16:17:29 <Tahvok> Nvm, I'll ask later about it 16:17:31 <evrardjp> the answer is probably yes 16:17:32 <evrardjp> :D 16:17:35 <evrardjp> so 16:17:36 <evrardjp> for the lxd 16:18:07 <evrardjp> I discussed with James Page at the PTG and he suggested to wait a little for the bump. I don't remember why. I think we can go ahead now 16:18:18 <evrardjp> confirmed medium? 16:18:41 <andymccr> yeah sounds fine 16:18:55 <evrardjp> ok next 16:18:57 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1719743 16:18:58 <openstack> Launchpad bug 1719743 in openstack-ansible "Fail when l2pop + L3 HA are enabled" [Undecided,New] 16:19:08 <evrardjp> that's IMO confirmed wishlist 16:19:24 <evrardjp> it's not a bug by itself, it's for helping others not doing mistakes 16:19:40 <evrardjp> maybe we can do pre-flight checks in the os_neutron role and fail at the beggining ? 16:19:47 <evrardjp> what do you think of the importance? 16:20:12 <andymccr> hmm 16:20:24 <andymccr> well 16:20:26 <andymccr> its wishlist for sure 16:20:51 <andymccr> i mean it would be nice to have that 16:21:24 <evrardjp> ok we agree then 16:21:25 <evrardjp> next 16:21:27 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1719741 16:21:28 <openstack> Launchpad bug 1719741 in openstack-ansible "repo_server role enables epel-testing which causes package conflicts" [Undecided,New] - Assigned to Major Hayden (rackerhacker) 16:21:40 <andymccr> EPEL - im out :P 16:22:03 <evrardjp> sounds ok to install per package 16:22:07 <evrardjp> andymccr: haha 16:22:09 <evrardjp> I understand 16:22:10 <evrardjp> mgariepy: ? 16:22:19 <evrardjp> what do you think of mhayden proposition? 16:22:40 <evrardjp> for me we could mark it as confirmed and low 16:22:51 <evrardjp> hwoarang: an opinion for opensuse side of things? 16:23:10 <hwoarang> hmm 16:23:23 <evrardjp> you're not impacted, it was just to see if you were still alive :p 16:23:28 <hwoarang> it shouldn't affect opensuse 16:23:29 <evrardjp> <-- evil 16:23:45 <hwoarang> i was just double checking in case that repo was enabled for all rpm distros 16:24:02 <mgariepy> yes should limit to needed pkg 16:24:10 <evrardjp> mgariepy: great 16:24:15 <evrardjp> could you handle that part? 16:24:23 <mgariepy> sorry i was almost out :) 16:24:26 <mgariepy> yeah sure 16:24:32 <evrardjp> I'm marking it as confirmed and wishlist, assigning to you 16:24:40 <mgariepy> perfect thanks :) 16:24:56 <evrardjp> thanks to you, sir! 16:25:02 <evrardjp> next 16:25:05 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1719517 16:25:06 <openstack> Launchpad bug 1719517 in openstack-ansible "Ceilometer policy.json includes python unicode encoding" [Undecided,New] 16:25:09 <evrardjp> ... 16:25:11 <evrardjp> ceilometer 16:25:46 <evrardjp> I don't really know 16:26:07 <evrardjp> because the role deprecation phase is still not merged, we cannot deprecate it right away 16:26:25 <evrardjp> (see https://review.openstack.org/#/c/504279/ ) 16:26:39 <evrardjp> spotz: if you could vote on this doc change ^ 16:26:52 <evrardjp> so I suggest we leave the bug as is? 16:26:56 <andymccr> ceilometer fun 16:26:57 <evrardjp> ok for everyone? 16:26:58 <andymccr> yeah 16:26:59 <andymccr> basically 16:27:04 <andymccr> i mean if somebody wants to go maintain it cool 16:27:05 <hwoarang> ok... 16:27:25 <hwoarang> maybe we can clarify that one bug in case the reporter wants to step up? 16:27:28 <andymccr> i mean it suggests ppl are trying it out? 16:27:29 <hwoarang> if he cares about it... 16:27:33 <andymccr> yeah agree with that 16:27:39 <evrardjp> what do you mean? 16:27:45 <andymccr> if we can explain to the submitter whats happening with ceilometer 16:27:58 <Tahvok> what's up with ceilometer? 16:27:59 <andymccr> that way they have a choice to say "i care about ceilometer and will try fix the role up" or "oh ok..." 16:28:00 <evrardjp> ok I will do that. 16:28:06 <Tahvok> I was about to deploy it, so asking.. 16:28:16 <evrardjp> Tahvok: terrible support :p 16:28:22 <hwoarang> do we have a deprecated list of roles? 16:28:25 <andymccr> Tahvok: its basically been unmaintained - so it is in a working state for like newton maybe not even. and hasnt been maintained 16:28:27 <Tahvok> Are there alternative services? 16:28:32 <evrardjp> hwoarang: not yet. 16:28:40 <evrardjp> hwoarang: that's the review I pinged you. 16:28:50 <andymccr> the service itself is fine, the OSA roles just hant been maintained, so if you fix it up it'll be fine imo. 16:29:07 <Tahvok> So the problem is not of not mainteining the role, but of not mainteining the ceilometer itself? 16:29:11 <hwoarang> evrardjp: yeah but i mean how do you know that ceilometer is going to be deprecated? which other role is in a similar situation? 16:29:27 <andymccr> Tahvok: no - its about maintaining the role, not ceilometer itself. 16:29:34 <andymccr> which is an easier problem to solve 16:29:39 <andymccr> but we just haven't had anybody step up to do that 16:29:56 <evrardjp> hwoarang: trove 16:30:06 <Tahvok> If our company decides finally that we want to deploy it, I could step up as mainteiner, but it will might take a couple of months 16:30:32 <andymccr> sounds good :) 16:30:38 <hwoarang> evrardjp: ok 16:30:54 <evrardjp> sounds great Tahvok! 16:31:12 <evrardjp> but when the docs patch is in, we can apply the policy 16:31:29 <evrardjp> I'll draft the maturity table as a next step 16:31:46 <evrardjp> ok I guess we can move on to a different bug right now, we know what to say in this bug 16:31:52 <evrardjp> next 16:31:53 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1718356 16:31:54 <openstack> Launchpad bug 1718356 in Ironic "Include default config files in python wheel" [Wishlist,In progress] - Assigned to Jesse Pretorius (jesse-pretorius) 16:32:08 <evrardjp> that's a wishlist, I'd mark it as confirmed 16:32:33 <evrardjp> next 16:32:35 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1716908 16:32:36 <openstack> Launchpad bug 1716908 in openstack-ansible "Unable to create new instances after upgrade." [Undecided,New] 16:33:41 <evrardjp> that sounds like an important bug 16:33:49 <andymccr> yeah 16:34:04 <andymccr> we have upgrade jobs for this no? 16:34:04 <evrardjp> I'll take this based on previous week's conversation after the triage 16:34:16 <evrardjp> from N to O? 16:34:42 <andymccr> yeah i thought we did 16:35:07 <evrardjp> well right now every periodic is busted with zuulv3 16:35:23 <evrardjp> but we triaged it offline with logan- last week 16:35:30 <evrardjp> we should just list all the packages explicitly 16:35:34 <andymccr> well that is unhelpful :P 16:35:36 <evrardjp> I will fix that 16:35:49 <evrardjp> yes, sorry I didn't update the bug 16:35:50 <evrardjp> my bad 16:35:54 <andymccr> http://logs.openstack.org/periodic/periodic-openstack-ansible-upgrade-aio-ocata-ubuntu-xenial/46b029e/console.html worked ok 16:36:01 <andymccr> well you got it under control so no worries 16:36:38 <evrardjp> it's very much a race condition too 16:38:14 <evrardjp> anyway 16:38:16 <evrardjp> next 16:38:24 <andymccr> next 16:38:25 <andymccr> ! 16:38:34 <evrardjp> admin0: https://bugs.launchpad.net/openstack-ansible/+bug/1715416 16:38:34 <openstack> Launchpad bug 1715416 in openstack-ansible "libvirtd override" [Undecided,New] 16:38:38 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1715416 16:38:42 <evrardjp> that's admin0's bug 16:38:57 <evrardjp> so basically he suggest that, for live migration, we have different defaults 16:39:09 <evrardjp> or a boolean to flip 16:40:02 <evrardjp> tcp and no auth seems a security issue... but I understand the willingness to disable it for speed of migration 16:40:18 <hwoarang> i am not sure. since you can configure them on your deployment why do defaults matter 16:40:19 <hwoarang> ? 16:40:37 <andymccr> can we put config_template on that setup? 16:40:40 <logan-> our default configuration doesn't use any of that, I think that bug was from before we added back live_migration_uri to pike 16:40:59 <Tahvok> Where does he take the needed conf for livemigration? 16:41:06 <logan-> https://github.com/openstack/openstack-ansible-os_nova/commit/12e09a3402cb810c53188a94ad1c820086d8e302 16:41:19 <logan-> our normal config uses ssh to migrate, not libvirt listeners 16:41:41 <Tahvok> Because I have tested live migration myself with only changing vncserver_listen=0.0.0.0 and it's working for me... 16:43:04 <evrardjp> logan-: that sounds more secure and working by default 16:43:12 <evrardjp> ok 16:43:19 <evrardjp> so we should mark this as invalid 16:43:24 <evrardjp> I am glad we have ppl at triage! 16:43:39 <andymccr> haha yeah 16:43:41 <andymccr> makes it a lot easier :P 16:43:58 <Tahvok> Where did he read of this libvirt configuration? 16:44:01 <logan-> i don't think it is unreasonable to request libvirt config overrides, but its just not necessary to get LM working by default. only if you wanted a custom config :) 16:44:23 <andymccr> if we can expose options to override the config if wanted that'd be fine 16:44:34 <andymccr> config_template might not work since it doesnt have [section] s 16:44:41 <evrardjp> andymccr: yes indeed 16:44:45 <logan-> right 16:44:51 <evrardjp> but all the sections talked about there are already overridable 16:44:57 <evrardjp> so I think we can mark this as invalid 16:44:59 <logan-> gotcha 16:45:34 <evrardjp> ok next 16:45:46 <evrardjp> # topic previous bugs 16:45:55 <evrardjp> dammit 16:45:58 <evrardjp> #topic previous bugs 16:46:01 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1716785 16:46:02 <openstack> Launchpad bug 1716785 in openstack-ansible "Optimize how we install and configure OS package repositories" [Wishlist,New] - Assigned to Major Hayden (rackerhacker) 16:46:49 <andymccr> major is out - i like the idea though ;D 16:47:14 <evrardjp> I emitted that idea long ago too, never formalized in a bug, so I'd say I agree. :p 16:47:26 <hwoarang> it sounds good in principle :) 16:47:40 <evrardjp> well that's how ppl ansible generally :D 16:47:41 <evrardjp> :p 16:48:12 <evrardjp> ok I'll mark it as confirmed 16:48:34 <evrardjp> next 16:48:36 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1716411 16:48:36 <openstack> Launchpad bug 1716411 in openstack-ansible "lxd driver broken due to upstream changes in nova" [Undecided,New] 16:49:14 <evrardjp> yeah so confirmed 16:49:28 <evrardjp> it is linked to previous bug to bump our files for nova-lxd 16:49:46 <evrardjp> with both of those in, it should improve stuff. 16:49:59 <evrardjp> I will bump master role when the dust settles on infra for master 16:50:02 <evrardjp> and releases are out :p 16:50:37 <andymccr> yeah 16:51:15 <evrardjp> ok next 16:51:26 <evrardjp> FYI, I've assigned that to me, confirmed and high 16:51:32 <evrardjp> next 16:51:34 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1715178 16:51:35 <openstack> Launchpad bug 1715178 in openstack-ansible "Git clone of openstack-ansible guide is wrong and not working" [Low,New] - Assigned to Alexandra Settle (alexandra-settle) 16:52:59 <evrardjp> that sounds invalid to me 16:53:05 <evrardjp> the clone seems ok to me 16:54:10 <evrardjp> ok next 16:54:13 <evrardjp> #link https://bugs.launchpad.net/openstack-ansible/+bug/1714597 16:54:15 <openstack> Launchpad bug 1714597 in openstack-ansible "multi OS repo-build needs to be more intuitive " [Undecided,New] 16:55:24 <evrardjp> for this I think the code has been updated since then 16:55:43 <evrardjp> so maybe there is no need for that 16:55:51 <andymccr> i thought that got fixed up 16:55:54 <andymccr> end of last cycle 16:55:55 <andymccr> could be wrong though 16:55:56 <evrardjp> yes 16:56:10 <evrardjp> well this is an older bug, and there are still remnants in some places 16:56:28 <evrardjp> someone said it wasn't reproducible 16:56:40 <evrardjp> so I guess we can say let's leave it open, and let it expire 16:56:47 <andymccr> yeah sure 16:57:13 <andymccr> or well cloudnull created the bug 16:57:23 <cloudnull> what? 16:57:23 <andymccr> so maybe cloudnull can confirm if its still an issue/what needs to get done 16:57:29 <andymccr> https://bugs.launchpad.net/openstack-ansible/+bug/1714597 this guy 16:57:30 <openstack> Launchpad bug 1714597 in openstack-ansible "multi OS repo-build needs to be more intuitive " [Undecided,New] 16:57:33 <cloudnull> oh yea. 16:57:53 <andymccr> oh yeah :P 16:57:55 <cloudnull> its the passing of all the extra ops to rebuild 16:58:08 <evrardjp> it should be smart enough 16:58:13 <cloudnull> it should be 16:58:21 <evrardjp> odyssey4me: 's gonna change the process in this cycle anyway :p 16:58:32 <andymccr> somebody run the smarterator against it and we should be good to go 16:58:40 <evrardjp> andymccr: twice. 16:58:41 <evrardjp> :p 16:58:57 <evrardjp> ok I guess that's all for today 16:59:01 <evrardjp> we are running out of time 16:59:05 <andymccr> yeah 16:59:09 <evrardjp> thanks to everyone involved! 16:59:10 <andymccr> think we. got through a lot 16:59:11 <andymccr> so good job 16:59:27 <evrardjp> indeed! good job everyone! 16:59:32 <evrardjp> closing... 16:59:34 <evrardjp> #endmeeting