Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-04-16

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:07 xingchao joined #puppet-openstack
00:24 rcallawa_ joined #puppet-openstack
00:26 rcallawa joined #puppet-openstack
00:27 rcallawa joined #puppet-openstack
00:32 rcallawa joined #puppet-openstack
00:35 ozialien joined #puppet-openstack
00:37 mwhahaha joined #puppet-openstack
00:37 rcallawa joined #puppet-openstack
00:42 xingchao joined #puppet-openstack
00:43 rcallawa joined #puppet-openstack
00:47 rcallawa_ joined #puppet-openstack
00:49 openstackgerrit Merged stackforge/puppet-keystone: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174099
00:55 xarses joined #puppet-openstack
00:55 rcallawa joined #puppet-openstack
01:00 rcallawa_ joined #puppet-openstack
01:03 rcallawa joined #puppet-openstack
01:07 markvoelker joined #puppet-openstack
01:13 BrianShang joined #puppet-openstack
01:15 EmilienM nibalizer: thanks
01:15 EmilienM nibalizer: was afk, sorry
01:24 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: Beaker tests  https://review.openstack.org/155873
01:25 openstackgerrit Merged stackforge/puppet-cinder: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174125
01:29 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-keystone: V3 support for Service provider only  https://review.openstack.org/173708
01:32 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Update Beaker tests to be consistent  https://review.openstack.org/174182
01:39 fedexo joined #puppet-openstack
01:47 ilbot3 joined #puppet-openstack
01:47 Topic for #puppet-openstack is now Puppet/OpenStack http://irclog.perlgeek.de/puppet-openstack/today - Weekly meeting (every Tuesday 13:00 UTC): https://wiki.openstack.org/wiki/Meetings/PuppetOpenStack - On going efforts: https://trello.com/puppetopenstack
02:34 ericpeterson joined #puppet-openstack
02:40 Vinsh joined #puppet-openstack
02:55 ericpeterson joined #puppet-openstack
02:59 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api  https://review.openstack.org/146926
03:11 britthouser joined #puppet-openstack
03:12 britthou_ joined #puppet-openstack
03:19 Vinsh joined #puppet-openstack
03:36 ericpeterson joined #puppet-openstack
03:48 ericpeterson joined #puppet-openstack
04:02 aimon joined #puppet-openstack
04:12 BrianShang joined #puppet-openstack
05:02 stevemar joined #puppet-openstack
05:08 fedexo joined #puppet-openstack
05:22 zerda joined #puppet-openstack
05:30 BrianShang joined #puppet-openstack
05:40 stevemar joined #puppet-openstack
05:44 zerda joined #puppet-openstack
05:47 saneax joined #puppet-openstack
06:01 ddmitriev1 joined #puppet-openstack
06:28 timfreund joined #puppet-openstack
06:33 hostranger left #puppet-openstack
06:51 saju_m joined #puppet-openstack
06:54 stevemar joined #puppet-openstack
07:08 stamak joined #puppet-openstack
07:32 tfz joined #puppet-openstack
07:43 jistr joined #puppet-openstack
07:47 trash Hmm, how do you handle automation for live migration requirements? As in, the nova user only having /bin/false instead of a shell and deploying the ssh keys?
07:48 trash I've written some puppet code which puts all compute nodes nova ssh keys into the other compute nodes nova user's authorized_keys, same for known_hosts, but the nova user seems to have /bin/false as shell... also having access from one node to all other nodes for live migration even though it has shared storage seems kind of... not nice.
08:00 mmagr joined #puppet-openstack
08:00 dgurtner joined #puppet-openstack
08:00 dgurtner joined #puppet-openstack
08:00 jpena joined #puppet-openstack
08:06 trash well, at least puppet-nova has a   nova_shell parameter
08:07 dalgaaf joined #puppet-openstack
08:12 gfidente joined #puppet-openstack
08:18 trash Well, the parameter is only being used when setting a UID also...
08:18 trash a wrapper it is then.
08:19 derekh joined #puppet-openstack
08:23 openstackgerrit Merged stackforge/puppet-horizon: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174130
08:23 openstackgerrit Sebastien Badia proposed stackforge/puppet-keystone: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174249
08:24 openstackgerrit Merged stackforge/puppet-swift: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174136
08:26 openstackgerrit Merged stackforge/puppet-heat: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174127
08:26 openstackgerrit Sebastien Badia proposed stackforge/puppet-swift: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174250
08:26 openstackgerrit Merged stackforge/puppet-glance: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174126
08:26 openstackgerrit Merged stackforge/puppet-manila: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174132
08:27 openstackgerrit Merged stackforge/puppet-openstacklib: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174140
08:27 openstackgerrit Merged stackforge/puppet-openstacklib: Unpin rabbitmq module  https://review.openstack.org/174142
08:27 tfz joined #puppet-openstack
08:27 openstackgerrit Merged stackforge/puppet-nova: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174120
08:28 openstackgerrit Merged stackforge/puppet-neutron: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174119
08:28 openstackgerrit Merged stackforge/puppet-ceilometer: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174123
08:34 skolekonov joined #puppet-openstack
08:34 openstackgerrit François Charlier proposed stackforge/puppet-ceilometer: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174252
08:38 openstackgerrit François Charlier proposed stackforge/puppet-neutron: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174254
08:43 openstackgerrit François Charlier proposed stackforge/puppet-nova: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174255
08:51 openstackgerrit Merged stackforge/puppet-glance: Separate api and registry packages for Red Hat  https://review.openstack.org/172440
08:51 openstackgerrit François Charlier proposed stackforge/puppet-heat: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174259
08:53 openstackgerrit François Charlier proposed stackforge/puppet-glance: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174260
09:12 skolekonov joined #puppet-openstack
09:36 openstackgerrit François Charlier proposed stackforge/puppet-horizon: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174276
09:38 openstackgerrit François Charlier proposed stackforge/puppet-cinder: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174277
09:40 openstackgerrit Gael Chamoulaud proposed stackforge/puppet-ceilometer: spec: Add Unit Tests for Ceilometer_config type/provider  https://review.openstack.org/170488
09:44 openstackgerrit Gael Chamoulaud proposed stackforge/puppet-manila: Move rabbit/kombu settings to oslo_messaging_rabbit section  https://review.openstack.org/173872
09:46 openstackgerrit Gael Chamoulaud proposed stackforge/puppet-ceilometer: Move rabbit/kombu settings to oslo_messaging_rabbit section  https://review.openstack.org/173481
09:46 openstackgerrit Gael Chamoulaud proposed stackforge/puppet-cinder: Move rabbit/kombu settings to oslo_messaging_rabbit section  https://review.openstack.org/173491
09:46 openstackgerrit Gael Chamoulaud proposed stackforge/puppet-heat: Move rabbit/kombu settings to oslo_messaging_rabbit section  https://review.openstack.org/173506
09:48 gchamoul sbadia: Salut! do you know what mfisch was meaning by 'see mailing list
09:48 gchamoul in https://review.openstack.org/#/c/173506/ for example
09:48 gchamoul sbadia: cannot see any thread related to this ...
09:49 sbadia huhu
09:49 sbadia no indeed…
09:49 sbadia no idea…
09:49 sbadia gchamoul: (hi :-))
09:50 gchamoul ok I think mfisch did eat too much seafood this week at Santa Clara ;-P
09:51 openstackgerrit Merged stackforge/puppet-ceilometer: spec: Add Unit Tests for Ceilometer_config type/provider  https://review.openstack.org/170488
09:51 sbadia :D
10:59 rcallawa joined #puppet-openstack
11:04 rcallawa joined #puppet-openstack
11:18 gildub joined #puppet-openstack
11:21 cdent joined #puppet-openstack
11:29 tfz joined #puppet-openstack
11:39 _joes_ joined #puppet-openstack
11:43 EmilienM good morning!
11:48 gchamoul social: puppet-neutron????
11:49 social O.o
11:49 * social hides
11:49 social gchamoul: puppet4 Error: Could not autoload puppet/provider/vs_port/ovs_redhat: Could not find parent provider ovs of ovs_redhat <, vswitch error
11:49 social from puppet neutron
11:50 social Error: Evaluation Error: Error while evaluating a Virtual Query, Could not autoload puppet/type/vs_port: Could not autoload puppet/provider/vs_port/ovs_redhat: Could not find parent provider ovs of ovs_redhat at /var/tmp/packstack/4ef44a1defe544a3b9313c0c22c70874/modules/vswitch/manifests/ovs.pp:57:29 on node controller-f21.default
11:50 gchamoul social: yes we forgot to prepare puppet-vswitch for puppet 4.x
11:51 gchamoul social: will do that right now !
11:51 social yes! and update master opm
11:52 social <,>
11:54 EmilienM sbadia: gchamoul: see maling list == the mfisher last email
11:55 gchamoul EmilienM: puppet-openstack ml?
11:55 EmilienM apparently, your patches break juno
11:55 gchamoul :S
11:55 EmilienM "Move rabbit/kombu settings to oslo_messaging_rabbit section" patches
11:56 gchamoul EmilienM: but all the patches are for Master ... so for Kilo ... not for juno :S
11:56 gchamoul i miss something!
11:56 EmilienM gchamoul: who said master == kilo ?
11:56 gchamoul EmilienM: oh ... just me! :D
11:56 EmilienM in my deployments, I consider that, but it's not a fact for everyone
11:56 EmilienM mfisch is running master for Juno.
11:57 EmilienM I think we should just decide of that and it will be ok.
11:57 EmilienM my opinion is to use stable branches and keep master for the OpenStack release in dev (currently Kilo)
11:57 gchamoul EmilienM: that will be logical!
11:58 EmilienM indeed, but again nothing has been decided
11:58 gchamoul EmilienM: ack
11:58 gchamoul of course!
11:58 gchamoul will find out the mfisch mail ...
12:00 social huh I think we have more juno incompatible changes in master
12:07 tfz joined #puppet-openstack
12:08 rcallawa joined #puppet-openstack
12:08 slagle joined #puppet-openstack
12:09 slagle EmilienM: jpena|lunch : https://bugs.launchpad.net/puppet-glance/+bug/1444974
12:09 rcallawa joined #puppet-openstack
12:09 EmilienM slagle: ack. I'll take care of it asap.
12:13 ericpeterson joined #puppet-openstack
12:20 dprince joined #puppet-openstack
12:21 stevemar joined #puppet-openstack
12:26 rzerda joined #puppet-openstack
12:31 EmilienM slagle: even if I revert the patch, glance won't work on RH
12:32 slagle EmilienM: ok, b/c of the actual packaging changes?
12:33 slagle EmilienM: a quick fix would work for me just as well :)
12:33 jpena my take on this is that it can (and should) be fixed on the packaging side
12:38 panda__ gchamoul: do you have a few minutes to chat ?
12:41 openstackgerrit Cédric LECOMTE proposed stackforge/puppet-neutron: Make neutron_plugin_ml2 before db-sync  https://review.openstack.org/173943
12:43 gchamoul ack
12:43 gchamoul panda: the same ... wrong channel :S
12:44 openstackgerrit Merged stackforge/puppet-keystone: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174249
12:46 openstackgerrit Merged stackforge/puppet-swift: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174250
12:48 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: Fix metadata.json lint issues  https://review.openstack.org/173578
12:48 openstackgerrit Merged stackforge/puppet-cinder: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174277
12:50 openstackgerrit Merged stackforge/puppet-nova: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174255
12:51 openstackgerrit Merged stackforge/puppet-horizon: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174276
12:51 openstackgerrit Merged stackforge/puppet-neutron: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174254
12:52 openstackgerrit Merged stackforge/puppet-glance: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174260
12:52 openstackgerrit Merged stackforge/puppet-heat: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174259
12:53 openstackgerrit Merged stackforge/puppet-ceilometer: Pin puppetlabs-concat to 1.2.1 in fixtures  https://review.openstack.org/174252
12:56 rcallawa_ joined #puppet-openstack
12:56 xingchao joined #puppet-openstack
12:58 rcallaw__ joined #puppet-openstack
13:02 rcallawa joined #puppet-openstack
13:11 openstackgerrit Cédric LECOMTE proposed stackforge/puppet-neutron: Make neutron_plugin_ml2 before db-sync  https://review.openstack.org/174369
13:14 richm joined #puppet-openstack
13:18 ferest joined #puppet-openstack
13:24 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: init: install openstack-glance on RH platforms  https://review.openstack.org/174375
13:24 EmilienM jpena: ^
13:24 EmilienM jpena: can you have a look ?
13:26 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: init: install openstack-glance on RH platforms  https://review.openstack.org/174375
13:27 jpena EmilienM: looking
13:29 jpena EmilienM: I'm afraid it won't fix packstack, because it calls class { '::glance::api' and class { '::glance::registry' directly. Depending on what is being done on tripleo, it will be the same
13:29 jpena EmilienM: for this fix, I'd revert the patch
13:29 EmilienM jpena: that's a design error
13:29 EmilienM you should call ::glance wherever you install glance.
13:29 EmilienM it's a puppet best practice
13:29 markvoelker joined #puppet-openstack
13:31 jpena EmilienM: ok, fixing packstack should be easy then. How is it done in tripleo?
13:31 EmilienM jpena: it's not correct I think
13:32 jpena EmilienM: which one is not correct? Not calling ::glance, or the way it is done in tripleo?
13:35 EmilienM jpena: https://review.openstack.org/174382
13:39 EmilienM jpena: ok so. Should I abandon my patch and wait packaging is updated?
13:40 jpena EmilienM: imho this is a good long-term fix, yes. It might take a day of instability until everything is tested, though
13:40 rcallawa joined #puppet-openstack
13:44 ibravo joined #puppet-openstack
13:45 dvorak EmilienM: is there a reason glance::api and glance::registry don't include ::glance ?
13:45 dvorak I noticed that's common, but I wasn't sure why
13:45 EmilienM well, I'm not sure we do that in other modules
13:46 dvorak I think most the modules work that way, I didn't know if it was for a reason, or just the way it turned out
13:52 stamak joined #puppet-openstack
13:56 openstackgerrit Merged stackforge/puppet-keystone: Fix metadata.json lint issues  https://review.openstack.org/173578
14:00 openstackgerrit Merged stackforge/puppet-glance: Change location to be a param, not a property  https://review.openstack.org/172579
14:02 ericpeterson joined #puppet-openstack
14:05 ferest joined #puppet-openstack
14:07 dgurtner joined #puppet-openstack
14:07 dgurtner joined #puppet-openstack
14:07 openstackgerrit Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
14:08 openstackgerrit Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
14:09 openstackgerrit Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
14:10 ibravo joined #puppet-openstack
14:14 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Update Beaker tests to be consistent  https://review.openstack.org/174182
14:16 markvoelker_ joined #puppet-openstack
14:18 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: LDAP: add support to configure credential driver  https://review.openstack.org/173565
14:20 mfisch my email should be on the ML
14:20 mfisch http://lists.openstack.org/pipermail/openstack-dev/2015-April/061640.html
14:21 EmilienM mfisch: can you add it to next meeting please? there is an etherpad: https://etherpad.openstack.org/p/puppet-openstack-weekly-meeting-20150421
14:22 EmilienM mfisch: we will make a choice and finally decide our policy.
14:22 mfisch will do
14:23 ddmitriev joined #puppet-openstack
14:23 markvoelker joined #puppet-openstack
14:23 ericpete_ joined #puppet-openstack
14:25 ferest joined #puppet-openstack
14:26 mfisch btw I was not trying to make anyone upset with my email so sorry if I did
14:28 Vinsh joined #puppet-openstack
14:29 dgurtner joined #puppet-openstack
14:29 dgurtner joined #puppet-openstack
14:30 alex_bh joined #puppet-openstack
14:36 ddmitriev joined #puppet-openstack
14:39 openstackgerrit Merged stackforge/puppet-neutron: Make neutron_plugin_ml2 before db-sync  https://review.openstack.org/174369
14:46 openstackgerrit Merged stackforge/puppet-neutron: Fix db-sync dependencies  https://review.openstack.org/173937
14:46 gchamoul mfisch: no worries about that mail ... :D
14:50 mdorman joined #puppet-openstack
14:53 ericpeterson joined #puppet-openstack
14:57 openstackgerrit François Charlier proposed stackforge/puppet-openstack: Fix unit-test in order to prepare module for deprecation  https://review.openstack.org/101704
14:59 salmankh joined #puppet-openstack
15:04 markvoelker joined #puppet-openstack
15:12 richm I have the first of my smaller patches as the result of splitting up the large keystone v3 patch
15:12 nosleep77 joined #puppet-openstack
15:12 richm The commit will have Implements: blueprint api-v3-support
15:12 richm however, it only implements part of that blueprint
15:13 richm Do I need to also split the blueprint into smaller blueprints, or is it ok if you have multiple commits for a single blueprint?
15:14 EmilienM richm: one blueprint is enough.
15:14 richm EmilienM: ok - thanks
15:15 EmilienM richm: if there is no change in BP, go ahead with multiple patchsets. Make sure you use the same topic for all patches. It would be easier to review
15:15 EmilienM also feel free to use dependencies
15:16 richm EmilienM: topic == commit message?
15:16 EmilienM no
15:16 EmilienM a gerrit topic
15:16 EmilienM or feature branch
15:19 openstackgerrit Matt Fischer proposed stackforge/puppet-keystone: Support notification_format  https://review.openstack.org/174426
15:28 britthouser joined #puppet-openstack
15:28 mdorman mfisch / EmilienM  you around?
15:29 mdorman obviously EmilienM is i guess :)
15:29 EmilienM mdorman: o/
15:29 xingchao joined #puppet-openstack
15:29 mdorman hey just wanted to apologize for my lack of work on the summit presentation so far.  i’m prepping for another conference i’m speaking at next week, then i’m on vacation for a few days.  but i will hit the YVR stuff pretty hard when i’m back the week of 4/27
15:30 dfisher joined #puppet-openstack
15:30 EmilienM cool np, we're all busy
15:31 mdorman yup yup.  trying to focus on one thing at a time, thrashing is bad :)
15:32 ozialien joined #puppet-openstack
15:34 salmankh Hey guys!
15:35 salmankh I have a question that stackforge/puppet-openstack is going to deprecate. What else is being introduced and is that on stackforge or puppetlabs forge
15:36 salmankh I see that puppetlabs/puppetlabs-openstack is provided at stackforge/puppet-openstack as a link for future work
15:37 salmankh So my question is, moving forward puppetlabs/puppetlabs-openstack is going to be used. Is that correct, or there is any other wrapper module as well.
15:38 ozialien left #puppet-openstack
15:39 salmankh Anyone to respond my query?
15:47 fc__ https://review.openstack.org/101704/ passes jenkins again, if anyone would like to re-validate again. cc. EmilienM mfisch who already code-reviewed previously
15:47 EmilienM fc__: ack
15:52 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the domain resource  https://review.openstack.org/174439
15:52 richm first keystone v3 patch is up - ^^^
15:53 richm jumbo patch was abandoned
15:53 richm This is still a big patch - +660, -15 - but that's because 5 new files are added
15:54 omartsyniuk joined #puppet-openstack
15:58 mschmitt joined #puppet-openstack
15:58 openstackgerrit Merged stackforge/puppet-openstack: Fix unit-test in order to prepare module for deprecation  https://review.openstack.org/101704
16:03 richm Is https://review.openstack.org/174439 still too big?  That's the smallest I can make it without doing some surgery to keystone.rb
16:03 ferest joined #puppet-openstack
16:08 markvoelker joined #puppet-openstack
16:08 openstackgerrit Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
16:25 mgagne once again: https://twitter.com/mgagne/status/588739801773318144 :-/
16:26 fc__ mgagne: apt.puppetlans.com ?
16:29 openstackgerrit Javier Peña proposed stackforge/puppet-glance: Fix os_region_name in provider  https://review.openstack.org/174455
16:29 EmilienM mgagne: http://apt.puppetlabs.com/ works for me
16:30 EmilienM maybe it's fixed
16:31 openstackgerrit Ryan Bak proposed stackforge/puppet-monasca: Add region support to libvirt plugin  https://review.openstack.org/174458
16:31 jpena EmilienM: https://review.openstack.org/174455 is needed for RDO Kilo CI, if you have time ;)
16:32 EmilienM jpena: there is no tests for this provider?
16:33 xarses joined #puppet-openstack
16:34 jpena EmilienM: true! Give me a sec to update that
16:35 mgagne EmilienM: rsync, not http
16:35 EmilienM mgagne: my bad
16:35 dfisher EmilienM: just a quick FYI, I'm starting down the path for puppet-openstack on Solaris.  Starting with keystone as that's a service we don't have to touch in the slightest for Solaris.
16:36 dfisher expect questions :)
16:36 EmilienM dfisher: by puppet-openstack, you mean all Puppet OpenStack modules, right?
16:36 dfisher yes
16:36 EmilienM ok. Cool, anything you can ask here
16:36 dfisher starting with keystone because it's easiest for us.  cinder/neutron/nova will be harder
16:36 dfisher but shouldn't be too terrible.
16:37 openstackgerrit AlexandreNo proposed stackforge/puppet-cinder: Add nfs_mount_options variable when backend is NetApp  https://review.openstack.org/174460
16:39 openstackgerrit Javier Peña proposed stackforge/puppet-glance: Fix os_region_name in provider  https://review.openstack.org/174455
16:39 stevemar are we allowed to have a dfisher and a mfisch -- sounds too fishy...
16:39 * stevemar sees himself out
16:40 dfisher my brother's name is mike.  I could get him in here so he could be mfisher
16:40 dfisher :)
16:41 dvorak richm: I just took a quick look, but the policy.json piece in the keystone_domain part troubles me.  Is there a way that it can be changed to use the existing keystone::policy piece?
16:41 dfisher or, worse:  fisherm
16:41 mdorman hehe
16:41 dvorak I'm worried that they're going to end up stomping on each other
16:42 aimon joined #puppet-openstack
16:48 openstackgerrit AlexandreNo proposed stackforge/puppet-cinder: Add nfs_mount_options variable when backend is NetApp  https://review.openstack.org/174460
16:49 dvorak richm: I think the patch is reasonably sized, especially compared to the earlier one
16:52 mfisch lol stevemar
16:53 britthouser joined #puppet-openstack
16:56 richm clayton: re: policy.json - yes, the keystone_domain is_admin => true will need to require the keystone::policy
16:56 openstackgerrit AlexandreNo proposed stackforge/puppet-cinder: Add nfs_mount_options variable when backend is NetApp  https://review.openstack.org/174460
16:57 richm clayton: we can't use policy afaik - the problem is that we need the admin domain id to write to the policy - but there is no way to get that at the puppet level
16:57 richm clayton: alternately, is there some way to "call" or "instantiate" keystone::policy from ruby code?
16:59 clayton wouldn't surprise me if there is, but hunner or colleen would be better people to ask
17:00 Hunner What kind of ruby code?
17:01 clayton richm: the policy.json stuff should probably be a native type, using augeus for it is kind of gross
17:03 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the project resource  https://review.openstack.org/174468
17:04 britthouser joined #puppet-openstack
17:04 richm Hunner: is it possible to call or instantiate a puppet class/resource from a provider, in ruby code?
17:05 richm clayton: ok - but I will leave the implementation of the policy class/resource to another patch
17:07 clayton I realize it wasn't your design decision, but having to hard code the domain_id in the policy.json file is horrible
17:11 richm clayton: I've really struggled with the domain property - it is required in order to use the v3 api for many keystone resources - but in order to implement it correctly would require adding a domain parameter to every single class/resource parameter list in every single puppet file that eventually references a keystone resource - boiling the ocean
17:12 richm there is no way that can happen all at once
17:12 richm I have still not found a satisfactory way of setting a globally applicable default domain
17:12 markvoelker joined #puppet-openstack
17:12 clayton and the domain must be specified by id, not by name?
17:13 richm clayton: In general, for creating a resource, you can specify by name
17:13 richm in keystone, a domain is a "top level" object - the name must be unique across the entire cloud
17:13 richm with v3, users and projects are contained within domains
17:14 clayton you've just summed up pretty much everything I know about domains :)
17:14 richm you can have multiple users with the same name, in different domains
17:14 richm the id still uniquely identifies users/projects, otherwise, you have to specify name + domain
17:14 clayton does default_domain have a default? :)
17:15 richm clayton: yes - Default
17:15 richm keystone will automatically create a domain called "Default" with id "default" for use with v2 clients
17:15 richm but - that may not be the domain you want to use by default
17:15 clayton so I assume you have a future patch coming that would add a domain parameter to most keystone resources?
17:16 richm clayton: in fact it has already begun - https://review.openstack.org/174468
17:16 clayton what about assuming the domain is 'Default' unless specified, and people silly enough not to use that will have to specify it on every resource they want to create with puppet?
17:17 richm clayton: that's a possibility - will have to think about the implications for some of the use cases I'm targeting
17:19 stamak joined #puppet-openstack
17:21 clayton ah, ok, so reading through the keystone repo a bit, I see default_domain_id is deterministic
17:25 clayton richm: keystone master example policy.json has this: ""cloud_admin": "rule:admin_required and domain_id:admin_domain_id","
17:25 clayton although I'll admit I don't really understand it
17:28 clayton richm: why does ensure_admin_default need to happen in the provider, instead of having some code around where it's set via keystone_config set it via the keystone::policy type?
17:28 richm clayton: The cloud_admin rule is true if the user is an admin (has role 'admin' in the requested resource) and the user's domain is the admin domain
17:29 richm clayton: because crinkle suggested I do it that way
17:29 richm clayton: If I implemented it in keystone_config, how do I get the domain id?
17:30 clayton well, it's going to have to be set in keystone.conf somehow, right?
17:30 richm right
17:30 clayton it seems to me that whatever class it writing it to keystone.conf should also update the policy.json file
17:31 clayton I think I'm missing something that makes it make sense for it to be in the provider
17:31 richm clayton: but they may be different
17:31 richm that is, the default_domain_id is not necessarily the same as the admin_domain_id
17:32 richm clayton: The only place we have access to the domain id is in the provider
17:32 richm clayton: otherwise, how do you get the domain id?
17:33 crinkle ^ that's why i thought it would happen in the provider, i don't know how you would do it from a manifest
17:33 clayton I think I need to read more :)
17:35 richm clayton: maybe think about it this way - at the puppet level, how do I get the id of a keystone_domain {} resource?  I cannot pass in the id to use - it must be assigned by keystone
17:35 clayton richm: I'm trying to wrap my head around the keystone part right now :)
17:36 britthouser joined #puppet-openstack
17:36 clayton so this is the line you're adding to policy.json pol_val = "is_admin:1 or (rule:admin_required and domain_id:#{id})"
17:36 clayton and id in that line is default_domain_id
17:36 clayton but shouldn't it be the admin domain?
17:37 richm the id in that line is the admin domain id
17:38 richm clayton:  that code is only called if is_admin is true
17:38 clayton ok, that clears up part of my confusion
17:38 richm so if is_admin returns true, then this domain is also the admin domain
17:38 clayton how is that different than this: "cloud_admin": "rule:admin_required and domain_id:admin_domain_id",
17:39 richm clayton: another twist that makes things difficult for puppet in keystone v3 :P
17:39 richm the default v3 policy _does not allow admin token access_
17:39 clayton oh, because admin_domain_id is an example, and isn't actually implemented as a substituion
17:40 * clayton shakes his fist at morgan
17:40 richm so we have to add "is_admin:1 or " to several rules in order to make the admin token work for puppet for bootstrapping purposes
17:40 richm I suppose that doesn't have to be in this patch
17:41 richm It could come later with the v3 policy patch
17:41 clayton I'm still trying to wrap my head around it.  I appreciate you indulging my questions
17:41 richm hopefully this discussion with serve as a faq . . .
17:42 clayton hopefully the puppet module does all of this and no one every has to realize how horrible this is
17:42 crinkle lol
17:42 xarses joined #puppet-openstack
17:42 richm It's quite painful and error prone to set up everything after the fact i.e. puppet install v2 then manually change things to be v3
17:43 clayton seriously, this is really crappy design, and hard to automate
17:43 richm clayton -> #openstack-keystone
17:44 clayton how do I get a openstack client that supports v3, or is it already there and I have to set a flag?
17:44 richm clayton: The "openstack" (python-openstackclient) client supports v3
17:44 richm clayton:  You have to both set --os-identity-api-version 3 - and use a "/v3" url
17:45 clayton I just want to see the arguments
17:45 richm openstack help domain create
17:45 clayton this wouldn't be so bad if you could specify the id when you create the domain
17:46 richm clayton: that, or allow me to specify the domain name for the default domain and the admin domain
17:46 britthouser joined #puppet-openstack
17:50 clayton ok, more stupid keystone v3 questions.  What makes a domain an admin domain?  Assigning it the admin role?
17:51 richm 1) your cloud "super user" is a user in that domain 2) you specify the id of that domain in the policy 'cloud_admin' rule
17:52 richm note that you cannot assign a role to a domain or a project
17:52 richm only users and groups can have roles
17:53 richm e.g. assign user 'admin' the role 'admin' in the domain 'default'
17:53 richm assign user 'admin' the role 'admin' in the project 'services'
17:53 richm But I'm sure you will have a lot more questions when my keystone_user_role provider review is posted
17:54 clayton I'm regretting every opening the first review :)
17:55 richm clayton: someone had to do it ;-)
17:55 clayton I'm thinking we'll just stay on keystone v2 forever :)
17:55 richm the pioneers get the arrows, the settlers get the land
18:03 ferest joined #puppet-openstack
18:07 britthouser joined #puppet-openstack
18:13 Vinsh joined #puppet-openstack
18:29 crinkle lhcheng in #openstack-keystone gave me a terrible idea...what if we just had a script called with an exec, and it just gets triggered by a keystone_domain resource to set the admin domain id?
18:30 Hunner richm: What do you want to do with the class or define that you create in ruby? Add it to the catalog?
18:30 Hunner richm: Or just run it?
18:30 * Hunner hasn't read scrollback
18:31 Hunner Well, you said "from a provider" so I assume add to a catalog
18:31 Hunner There is a function called "generate" that types can have that can add resources to the catalog. It would probably work for classes too
18:33 mfisch EmilienM|afk: we need to use the ranked voting thing when we vote on names, there are too many for a straight up vote
18:33 mfisch Concordet?
18:34 EmilienM mfisch: indeed
18:34 Hunner richm: So... what about using function_include? :P
18:36 clayton seems fairly crazy that we can't use Puppet in the name
18:36 EmilienM clayton: I think we really need to move forward with that thing.
18:36 EmilienM I would like to have it fixed by next week meeting to be honest
18:37 clayton I understand why PuppetLabs wouldn't want to explicitly give permission, but every puppet module on github has puppet in the name
18:37 clayton EmilienM: yeah, I understand, just shaking my head over where we've ended up
18:37 EmilienM I don't want our project blocked to move under the big tent because of a name
18:37 clayton agreed
18:37 EmilienM clayton: I stopped shaking my head :)
18:37 clayton and I agree with mfisch about the voting
18:38 Hunner clayton: You can use puppet in the name... right?
18:38 crinkle apparently not
18:38 xingchao joined #puppet-openstack
18:38 Hunner crinkle: Really? :(
18:38 clayton Hunner: I think someone made the mistake of asking
18:38 Hunner They don't think that the language is called "puppet"?
18:38 crinkle Hunner: check richard's latest email
18:39 clayton well, not that it's richard's fault, but Anne brought up the issue, so he kind of had to ask
18:39 mfisch I'm adding Portlandia as a possible name to the voting list
18:39 clayton seconded!
18:39 mfisch what was the chicken's name? we could use that too
18:39 clayton it might as well just be something ridiculous
18:39 mfisch I think it was Frank
18:40 mgagne "OpenStack Teppup" guys
18:40 Hunner Hrm. I have a feeling that legal people don't understand technology... not that that's new
18:40 clayton Hunner: I suspect it's more that they have no incentive to say yes
18:40 mfisch no lawyer ever got fired for saying No
18:40 mgagne it's like the PHP trademark, they don't want people to publish application with the name PHP in it anymore.
18:41 crinkle incentive should be product marketing, "look our product supports openstack"
18:41 mfisch This will take more time than we have to get the lawyers to agree at this point. Let's just vote on something else.
18:41 mfisch for the record I really like my name
18:42 clayton Where does the name of the project even show up?
18:42 clayton I assume we can leave the repos as puppet-<service>
18:42 mfisch Wiki
18:42 mfisch I dont even know if thats true wrt leaving the repo
18:42 clayton someone should tell the infra team to rename all of theirs then.
18:42 markvoelker joined #puppet-openstack
18:42 mfisch Pupp-et it's French so it doesn't infringe
18:43 Hunner clayton: But "puppet" is the name of the language...
18:43 clayton https://github.com/openstack-infra?utf8=%E2%9C%93&amp;query=puppet
18:43 Hunner I guess "java" is the name of a language and oracle doesn't like people using that, though that is also a terrible example
18:44 Hunner clayton: Yes, puppet-* repo names should be fine
18:44 clayton Hunner: nod, so I assume we can still use it in the name of the repos, plus I doubt anyone will ask about the puppet-* names
18:44 Hunner I think the words were "oh yeah we don't have any problem with git repo names. Everyone does that"
18:45 Hunner mfisch: Heh
18:45 EmilienM clayton: no, don't worry, our puppet modules won't be renamed.
18:47 Hunner richm: Anyway, function_include and function_create_resources can both be used to declare classes or defines
18:47 EmilienM well, I hope so
18:48 EmilienM crinkle: I replied to the thread without reading your answer, mistake of me. But it seems like we have the same idea
18:48 EmilienM (the backport tag)
18:49 crinkle EmilienM: hi5
18:50 clayton EmilienM: I like the idea of automated cherry-picking for commits with that tag
18:50 clayton that would make things a *lot* easier and give people a bigger incentive to add the tag
18:50 clayton richm: morgan said he thinks you probably can refer to the domain by id in kilo, but he's not sure and wasn't going to have time this week to validate it
18:51 EmilienM well, seems like we have sort of solution here
18:51 crinkle ++ to automating it
18:54 mfisch mgagne: that idea is probably what we will do, stick a shim in our private module
18:54 mgagne mfisch: yep
18:54 EmilienM crinkle: I think we should keep names that are possible to use (ie: trademark issues) and then proceed at a vote
18:54 mgagne mfisch: I don't see any other solution for now
18:54 crinkle EmilienM: sure
18:54 mgagne mfisch: maybe we could ship the shim in upstream module but it's not work I'm willing to do atm
18:55 mgagne mfisch: until we come up with a clever way to manage and maintain it
18:55 mfisch agree
18:55 mgagne mfisch: like a yaml with translations for each versions so we can easily auto generate it from upstream code
18:55 clayton mgagne: yeah, that's exactly the sort of thing we did for upgrading from icehouse to juno
18:55 mfisch we've held them before
18:55 mgagne mfisch: and validate it
18:55 clayton mgagne: I like that idea
18:56 mgagne clayton: I like my ideas too =)
18:56 clayton when we upgraded to juno, we moved to master branch and worked out all the various fixups we needed to keep icehouse working, but most of it was deprecated parameters, not removed ones
18:56 mfisch which is doable
18:59 ericpete_ joined #puppet-openstack
18:59 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: LDAP: add support to configure credential driver  https://review.openstack.org/173565
18:59 ozialien joined #puppet-openstack
19:00 EmilienM nibalizer: I'm trying to debug the puppet-nova/beaker patch I did, and the logs are not very useful ( http://logs.openstack.org/82/174182/2/check/gate-puppet-nova-puppet-beaker-rspec/1b9000f/console.html ) do you know if there is something with nodepool ATM ?
19:03 nibalizer nodepool super busted atm
19:04 nibalizer EmilienM: oh yea that means beaker derped
19:04 nibalizer happens kindof a lot
19:04 EmilienM nibalizer: nothing I can do, right?
19:04 nibalizer i mean nothing easy
19:04 nibalizer recheck is pretty easy
19:05 rcallawa_ joined #puppet-openstack
19:05 nibalizer we could add a 'ssh localhost' hypervisor to beaker that better explains to beaker what is up
19:05 nibalizer we could also make the script that prepares the machine for local ssh access better
19:05 nibalizer i think what happens is beaker logs in to the host and issues a service sshd restart and sshd never comes back
19:05 nibalizer there is a beaker bug for it i think
19:07 Hunner richm: Actually, what problem are you trying to solve in the first place? Declaring classes or defines from providers would happen client side only...
19:13 openstackgerrit Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
19:14 richm Hunner: Ideally I could just call keystone::policy to set the admin_domain_id in the 'cloud_admin' policy, and call keystone_config { 'identity/default_domain_id': value => $default_domain_id}
19:14 richm Hunner: The problem is that there is apparently no way to get those domain ids at the puppet level
19:15 richm Hunner: The ids are generated by keystone, so are only "visible" inside the keystone_domain provider code
19:16 richm Hunner: I can't use a custom function either
19:16 richm Hunner: so I have to access keystone.conf and policy.json in the keystone_domain provider where I have access to the domain id
19:17 richm Hunner: Then the problem is that I cannot introduce Keystone_config['identity/default_domain_id'] into the catalog (or introduce the classes/resources used by the policy) in order to use for requires and notifies
19:19 clayton richm: I understand the issue with the admin domain, but why can't you have the default_domain be via keystone_config?
19:20 richm clayton: It's not "default_domain" it's "default_domain_id" - how do I get the domain id at the puppet level?  Or, alternately, how do I call the keystone_config { 'identity/default_domain_id' } resource from ruby?
19:23 clayton are you sure that can't be a name?  I assumed it could be since the default is 'default'
19:23 Hunner richm: So the domain id has to be queried on the client side?
19:23 richm Again, this where keystone is confusing - it uses 'Default' as the _name_ of the default domain, and uses 'default' as the id of the default domain
19:23 clayton that must be just generate one randomly
19:24 clayton richm: right, but if you want anything but 'default' then you have to specify it in the config file
19:24 Hunner richm: You could create a keystone_config with the domain_id and a fake value, then update the value on the client side
19:24 clayton so I think setting it via keystone_config should be sufficient
19:25 clayton again, I know you've spent a lot of time on this, and I'm coming to this cold, I'm probably missing something :)
19:25 clayton on the plus (negative?) side, I've learned a lot about domains today I guess
19:27 richm Hunner: The domain id can be queried anywhere, but we have to make sure we use the right authentication
19:27 richm Hunner: Can you explain what you mean by "create a keystone_config with the domain_id and a fake value, then update the value on the client side"?
19:28 clayton Hunner:  the issue is that the id is a uuid that keystone generates during the puppet run, while applying the catalog
19:28 richm Hunner: I don't think that will work - let's say I use keystone_config { 'identity/default_domain_id' } as a requirement - once it is set to the dummy value, the requirement will be fulfilled, and other resources will happily use the dummy value instead of the real default value
19:29 richm clayton: I'm not sure what you mean by "setting it via keystone_config should be sufficient" - I don't know how to do that if I don't have the domain id
19:30 clayton you do have the domain id for the default_domain
19:30 clayton since if it's anything but default, the user has to specify it in the config file
19:30 clayton and it's created by keystone-manage, right?
19:30 clayton that's what the config reference says, I may be naive in believing it :)
19:30 Vinsh_ joined #puppet-openstack
19:31 richm let's say I'm doing an install where I want the default domain to be my ldap server for my users - so the first thing I do in my puppet manifest is to create that domain, get the id, then write that id to keystone.conf to use for _all_ subsequent resources
19:34 clayton can the default domain be changed after it's created?
19:34 richm Yes
19:34 richm And yes, it is keystone-manage that creates the 'Default' 'default' domain
19:35 clayton so we have a chicken and egg problem here
19:35 gildub joined #puppet-openstack
19:35 richm Yes
19:35 clayton honestly, I'm not sure we should even support this as a single pass option
19:35 clayton I think I understand the issue now.  this is really horrible
19:37 clayton for example, how are you going to handle restarting keystone after changing the default_domain_id?
19:37 richm clayton: Isn't that already taken care of?    File['/etc/keystone/keystone.conf'] -> Keystone_config<||> ~> Service[$service_name]
19:38 richm Ah, I see what you mean
19:38 clayton richm: no, because you're reading and writing the config file directly, so Keystone_config isn't installed
19:38 richm yeah, I'll have to manually restart
19:38 clayton err, s/installed/involved/
19:38 richm yet another problem that would have been solved if I could use a real resource
19:39 clayton so nova/neutron has something similar, I'm on a call now, I can find a reference for you after
19:39 clayton not suhre if it'll help, but it's a similar problem
19:39 Hunner richm: Do you know about how the catalog is built on the master, and applied on the agent?
19:39 richm clayton: sure - any help would be appreciated
19:40 richm Hunner: I know nothing about that - my puppet is very limited
19:40 Hunner Ooo
19:40 Hunner So there are two phases to how puppet works, compile and apply.
19:40 richm . . . so any help from puppet experts would be greatly appreciated
19:40 Hunner The puppet agent sends its facts to the master and then asks the master for a catalog. The master reads all the manifests/modules and compiles the catalog
19:41 Hunner At that point all conditional logic, variables, and functions are evaluated
19:41 Hunner Then the agent gets the catalog which is basically a list of resources and the dependency graph, and traverses the graph while evaluating the resources
19:41 Hunner To modify the catalog after compile is odd and rare, and usually not easy
19:42 Hunner So when the domain_id can be acquired and where it is needed in the modules is pretty important
19:42 clayton and I imagine pretty brittle
19:42 Hunner clayton: There are more brittle and less brittle ways
19:42 Vinsh joined #puppet-openstack
19:43 clayton Hunner: is it even possible for puppet to restart a service more than one in a single catalog apply?
19:43 * richm is counting on the answer being "yes"
19:43 clayton seems like it should be, but I'm not sure I've ever seen it
19:44 Hunner You're talking about transition states... non ideal, but https://forge.puppetlabs.com/puppetlabs/transition should let you do that
19:44 Hunner Why would you want to restart twice though?
19:45 richm install with bootstrap config -> domain create default domain -> domain create admin domain -> write default domain to keystone.conf -> write admin domain to policy.json -> restart
19:45 clayton richm: wants to be able to start keystone with the default_domain_id set to 'default', then create a new domain and write the id of that domain to the config file as the new default_domain_id, then restart keystone again
19:46 Hunner Why not just set it as the default in the first place?
19:47 clayton Hunner: because keystone generates the uuid at random when it creates the new default domain, and you can only refernece it by uuid, not by name, even through names are guaranteed to be unique
19:47 richm specifically, you have to use the uuid as the [identity] default_domain_id value, and the policy.json admin_domain_id value
19:47 Hunner So the way I've handled uuids vs names in openstack types before is always refer to them by name, and look up the uuid when you need it
19:48 clayton I agree that'd be better, but keystone devs apparently didn't see the need for that :)
19:48 Hunner But iirc the current providers do that kind of thing all the time
19:48 richm Hunner: [identity] default_domain_id requires a uuid - policy.json admin_domain_id value requires a uuid
19:48 clayton they've acknowledged that it should probably be fixed
19:48 richm Hunner: can you explain "the current providers do that kind of thing all the time"?
19:51 crinkle neutron looks up ids by name like https://github.com/stackforge/puppet-neutron/blob/master/lib/puppet/provider/neutron.rb#L184-L197
19:51 Hunner richm: The neutron_subnet provider gets a list of existing neutron subnets (https://github.com/stackforge/puppet-neutron/blob/master/lib/puppet/provider/neutron_subnet/neutron.rb#L23) and correlates them based on name (https://github.com/stackforge/puppet-neutron/blob/master/lib/puppet/provider/neutron_subnet/neutron.rb#L45) even though it has the UUIDs
19:51 clayton richm: this is the one I was thinking of https://github.com/stackforge/puppet-neutron/blob/master/lib/puppet/provider/nova_admin_tenant_id_setter/ini_setting.rb
19:52 Hunner Any then anyone refers to the subnets by name, but it uses the uuids under the hood
19:52 crinkle lots of neutron examples :)
19:52 Hunner :)
19:56 clayton if you did something like nova_admin_tenant_id_setter, could you have it both require and notify the keystone service?
19:57 richm clayton: yes - and bonus - that shows how to create a puppet resource in ruby
19:57 crinkle that would cause a dependency cycle
19:57 clayton crinkle: that's what I was wondering.  I've always thought of notify => as a before => on steroids
19:57 crinkle it is
19:58 Hunner But subscribe => is require =>
19:58 Hunner So you can't have require & before
19:58 clayton nod
20:00 clayton richm: is changing the default domain during bootstrap something you have an actual requirement for?
20:00 openstackgerrit Ryan Bak proposed stackforge/puppet-monasca: Add region support to libvirt plugin.  https://review.openstack.org/174458
20:00 richm clayton: If it is not possible, I'll just have to deal with it
20:01 clayton it would be a lot easier if we could open bugs against keystone to make this stuff easier
20:01 richm i.e. boil the ocean by making sure the right domain is specified in every resource/class everywhere
20:01 clayton I got the impression that some of it (like refering to domains by name) could be pretty easily be added in liberty
20:02 clayton personally I'd argue it should be backported, but I doubt anyone would listen to me :)
20:02 stevemar clayton, propose it :)
20:02 richm One of my hopes for getting the puppet group under the "big tent" is that puppet developers would be part of design discussion for new features in keystone, nova, neutron, etc.
20:02 clayton stevemar: I'll worry about it when we're trying to run v3 :)
20:03 richm Instead of puppet developers being left to put the pieces together ex post facto
20:03 clayton stevemar: it's not that I'd mind doing it, but I have finite hours in the day :/
20:06 richm crinkle: Could I do something like https://github.com/stackforge/puppet-neutron/blob/master/lib/puppet/provider/nova_admin_tenant_id_setter/ini_setting.rb#L189
20:06 richm crinkle: In my patch to update keystone_config 'identity/default_domain_id' - rather than dealing with keystone.conf directly?
20:07 stamak joined #puppet-openstack
20:07 clayton richm: the tenant_id_setting is really doing something really similar to what you're already doing, but it models it better
20:07 richm clayton: right
20:07 clayton the problem is that you're going to require that keystone be up before the resource is applied
20:07 crinkle richm: yeah it's a little gross but better than gsubbing the file
20:08 clayton but you also need to notify keystone if it changes
20:08 richm I'm not gsubbing the file any more in my latest patch :-)
20:08 clayton and that is going to cause a circular dependency
20:08 EmilienM crinkle: re- backport/tag: I've looked at jeepyb and the problem is we have to execute the cherry-pick -x when the patch lands, not at first patchset (and jeepyb is executed at first patchset AFIK)
20:09 crinkle EmilienM: that's too bad
20:09 EmilienM richm: +1 for design thing.
20:09 EmilienM crinkle: well, we'll find another way to make it.
20:09 clayton could patch jeepyb :)
20:09 clayton installing a gerrit commit hook is pretty easy in theory
20:10 clayton I was looking at jeepyb last night, because I think we should probably start using it internally
20:10 clayton I assume jeepyb = GPB, but I can't figure out what that stands for.  Gerrit project builder?
20:12 markvoelker joined #puppet-openstack
20:13 EmilienM crinkle: forget what I said
20:13 EmilienM crinkle: gerrit can handle it.
20:13 crinkle EmilienM: cool
20:24 dprince joined #puppet-openstack
20:24 guimaluf joined #puppet-openstack
20:26 EmilienM clayton: did you start something already on jeepyb?
20:27 clayton nope, was just thinking outloud (kind of)
20:28 clayton did you find a way to do it with jeepyb?
20:29 EmilienM clayton: yes
20:29 clayton I just found the bug hook that runs on every changeset, what did you find?
20:30 EmilienM clayton: there is no code yet but we will make it
20:30 EmilienM I mean, if nobody is against this idea
20:31 clayton I think it makes sense, and I hadn't through of using jeepyb for it until you suggested it
20:44 EmilienM mdorman: I don't see your point on your last reply. Your feeling about the is mixed, right?
20:48 mdorman yeah that probably sums it up
20:48 mdorman basically agree that the idea situation is that master is fundamentally dynamic…. and that people should be using stable branches in practice… and we should be aggressively backporting things.
20:48 EmilienM well, I think we will decide it together during the next meeting. Before I want to make sure discussion happens on  ML
20:48 mdorman right.
20:49 EmilienM crinkle's point is fair though.
20:49 EmilienM Other projects have stable branches.
20:49 clayton personally I like following master, because it's easier, as long as it supports the current release + previous.  But I also have to acknowledge that it's possible that it miay not be practical to do in all cases.
20:49 mdorman i guess i’m advocating for anything we can do to make the backport/cherrypick process as simple and easy as possible…  to remove resistance points of doing that.
20:49 EmilienM clayton: this is not really how OpenStack manage the releases
20:49 clayton yeap, I understand
20:49 EmilienM mdorman: a bot would solve that issue.
20:49 clayton but I also think puppet-openstack is a bit different than most existing projects
20:49 mdorman clayton:  i agree with that statement,too.  following master is definitely easiest.
20:50 mdorman EmilienM:  yeah i like that bot idea.  i really like it, actaully.
20:50 clayton and I think most other projects probably have more resources to work on them
20:50 clayton EmilienM: I really think there are two issues here, whether or not you want to support not breaking people following master, but also part two, which is how to manage the upgrade path
20:51 clayton the latter issue is why we ended up on master
20:52 mdorman yeah for sure jumping from one stable branch to another is a more complicated/brute force upgrade
20:52 openstackgerrit Ryan Bak proposed stackforge/puppet-monasca: Allow monasca-agent install from wheels  https://review.openstack.org/174599
20:53 clayton I agree that it makes sense to do what other projects do to a large extent, but I think puppet openstack is fairly different than most existing projects under the openstack/ namespace right now.
20:54 clayton I'm sure it's possible, but I think it'll be very hard to ever have well tested complete support for a new release at the same time as the other projects.
20:55 clayton that's a lot of tooling and a lot of hours of work
20:56 EmilienM well, also I'll have to make sure this policy (auto-cherry-pick) is accepted by TC maybe?
20:58 clayton I don't know.  Is do they get down to that level of detail?  I'd have thought that would be the sort of thing a project could decide on it's own, if the infra team was ok with it
20:58 clayton I'd understand if the infra team was hesitant to allow adding more gerrit commit hooks
20:59 mdorman doesn’t seem like it would be a TC thing to me
21:01 clayton I'd guess the infra team would say so if asked.  Probably worth running by them first anyway
21:01 mdorman yeah.
21:02 openstackgerrit Merged stackforge/puppet-tripleo: Loadbalancer: Add support for Redis  https://review.openstack.org/171121
21:03 EmilienM I'm writing a PoC now
21:12 panda EmilienM: where are the gate-puppet-*-puppet-beaker-rspec gates hosted ?
21:12 EmilienM panda: git clone git@github.com:openstack-infra/project-config.git
21:13 EmilienM in jenkins/jobs/puppet-module-jobs.yaml
21:25 openstackgerrit Merged stackforge/puppet-monasca: Add region support to libvirt plugin.  https://review.openstack.org/174458
21:42 markvoelker joined #puppet-openstack
21:45 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-keystone: Adds OSC with V3 support  https://review.openstack.org/174618
21:48 gildub richm, ^
21:49 gildub richm, this one wraps endpoints completely transparently - but I haven't worked out the mock yet, not sure how to simulate that, but besides the rspec that works like a charm
21:50 fvollero howdy slackers
21:50 gildub fvollero, howdy rital!
21:51 fvollero gildub: bonjour
21:51 gildub fvollero, salut Francesco, ca va?
21:51 gildub fvollero, how's the tiger's life going?
21:53 fvollero gildub: Roar..I become an SSA a.k.a. Specialized Solution Architect
21:53 fvollero gildub: so, i will go selling cloudy things
21:53 fvollero LOL
21:54 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the project resource  https://review.openstack.org/174468
21:54 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the domain resource  https://review.openstack.org/174439
21:54 gildub fvollero, LOL -  I hope the variable component is a good incentive!
21:55 fvollero gildub: sure it is my friend :) The catch is that I still am EMEA, so I am traveling a lot
21:55 fvollero EmilienM|afk: salut!
21:55 EmilienM fvollero: you alive ???
21:56 fvollero EmilienM: Bien sur! But I am spinning around Europe mon ami
21:57 gildub fvollero, BTW, tigers don't roar (lions do), they furrr, but usually when you hear that it's already too late!
21:58 fvollero gildub: ahhahaha :) Really? Tigers furr ?
21:58 fvollero ahhaha
21:59 gildub fvollero, I don't know the exact term, but yes
21:59 * gildub get go, see ya!
22:01 fvollero gildub_bbl: cya!
22:03 EmilienM richm: I think https://review.openstack.org/#/c/173788/2/lib/puppet/provider/openstack.rb,cm will break v2. Am I wrong?
22:12 mdorman joined #puppet-openstack
22:23 EmilienM mdorman, clayton, crinkle: very WIP https://review.openstack.org/174627
22:26 mdorman cool will take a look
22:29 mdorman seems like a good start
22:34 ferest joined #puppet-openstack
22:37 BrianShang_ joined #puppet-openstack
22:43 rcallawa joined #puppet-openstack
22:48 markvoelker joined #puppet-openstack
22:49 markvoelker_ joined #puppet-openstack
22:49 rcallawa joined #puppet-openstack
22:50 markvoel_ joined #puppet-openstack
22:51 markvoe__ joined #puppet-openstack
22:56 gildub EmilienM|afk, re your 173788 comment, it won't break v2 if when passing version number
22:58 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-keystone: Adds OSC with V3 support  https://review.openstack.org/174618
23:05 openstackgerrit Athlan-Guyot sofer proposed stackforge/puppet-tempest: Bad configuration for glance/neutron setters.  https://review.openstack.org/174638
23:11 EmilienM gildub: ack
23:12 stevemar joined #puppet-openstack
23:19 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: LDAP: add support to configure credential driver  https://review.openstack.org/173565
23:20 EmilienM richm: I don't get what I'm missing here ^ (tests are failing)
23:38 panda joined #puppet-openstack

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary