Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-06-22

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

All times shown according to UTC.

Time Nick Message
00:00 ducttape_ joined #puppet-openstack
00:01 markvoelker joined #puppet-openstack
00:02 gildub joined #puppet-openstack
00:13 ducttape_ joined #puppet-openstack
00:24 ianw joined #puppet-openstack
00:52 ducttape_ joined #puppet-openstack
01:07 gildub joined #puppet-openstack
01:49 fedexo joined #puppet-openstack
01:50 markvoelker joined #puppet-openstack
01:57 stevemar joined #puppet-openstack
02:07 ducttape_ joined #puppet-openstack
03:08 ducttape_ joined #puppet-openstack
03:21 xingchao joined #puppet-openstack
03:22 ducttape_ joined #puppet-openstack
03:38 markvoelker joined #puppet-openstack
04:44 gildub richm, hi - do you know if beaker issue with user_role has been resolved, I can't reproduce it. I'm using 174999/25
04:52 chandankumar joined #puppet-openstack
04:55 gildub richm, never mind, now reading the channel logs
05:27 markvoelker joined #puppet-openstack
05:43 ducttape_ joined #puppet-openstack
06:44 arnaud_orange joined #puppet-openstack
06:47 chandankumar_ joined #puppet-openstack
07:01 jpena joined #puppet-openstack
07:03 fedexo joined #puppet-openstack
07:16 markvoelker joined #puppet-openstack
07:17 saneax joined #puppet-openstack
07:32 ducttape_ joined #puppet-openstack
07:33 xingchao joined #puppet-openstack
07:35 dgurtner joined #puppet-openstack
07:35 dgurtner joined #puppet-openstack
07:40 chandankumar joined #puppet-openstack
07:41 gfidente joined #puppet-openstack
07:41 gfidente joined #puppet-openstack
07:52 cwolferh joined #puppet-openstack
08:07 tshtilma joined #puppet-openstack
08:14 wabu joined #puppet-openstack
08:15 paramite joined #puppet-openstack
08:33 saneax joined #puppet-openstack
08:38 openstackgerrit David Gurtner proposed stackforge/puppet-ceph: Fix dependency version  https://review.openstack.org/193869
08:44 derekh joined #puppet-openstack
09:02 linkedinyou joined #puppet-openstack
09:05 markvoelker joined #puppet-openstack
09:20 ducttape_ joined #puppet-openstack
09:22 dgurtner joined #puppet-openstack
09:22 dgurtner joined #puppet-openstack
10:03 richm joined #puppet-openstack
10:11 cdent joined #puppet-openstack
10:15 chandankumar joined #puppet-openstack
10:43 dweaver joined #puppet-openstack
10:43 mgagne joined #puppet-openstack
10:43 partner joined #puppet-openstack
10:44 zigo joined #puppet-openstack
10:44 khappone joined #puppet-openstack
10:53 markvoelker joined #puppet-openstack
11:03 openstackgerrit Yanis Guenane proposed openstack/puppet-neutron: Opencontrail: Create neutron::plugins::opencontrail  https://review.openstack.org/187986
11:06 linkedinyou joined #puppet-openstack
11:09 ducttape_ joined #puppet-openstack
11:13 gildub joined #puppet-openstack
11:22 openstackgerrit Yanis Guenane proposed openstack/puppet-neutron: Opencontrail: Create neutron::plugins::opencontrail  https://review.openstack.org/187986
11:23 cdelatte joined #puppet-openstack
11:28 xingchao joined #puppet-openstack
11:32 ducttape_ joined #puppet-openstack
11:51 xingchao joined #puppet-openstack
11:54 markvoelker joined #puppet-openstack
11:55 openstackgerrit Lukas Bezdicka proposed openstack/puppet-neutron: Fix support for puppet4  https://review.openstack.org/194097
12:00 dprince joined #puppet-openstack
12:03 markvoelker joined #puppet-openstack
12:05 todin left #puppet-openstack
12:10 EmilienM good morning!
12:25 ducttape_ joined #puppet-openstack
12:31 iurygregory joined #puppet-openstack
12:35 openstackgerrit Iury Gregory Melo Ferreira proposed openstack/puppet-openstack-specs: Enabling-k2k-federation  https://review.openstack.org/190361
12:44 rcallawa joined #puppet-openstack
13:19 xingchao joined #puppet-openstack
13:24 trash It's not just you! http://review.openstack.org looks down from here.
13:24 trash :-(
13:26 openstackgerrit Merged openstack/puppet-nova: Better handling of package dependencies in nova generic_service  https://review.openstack.org/192009
13:27 morazi joined #puppet-openstack
13:57 magicboiz joined #puppet-openstack
14:00 openstackgerrit Stefan Andres proposed openstack/puppet-glance: Support identity_uri  https://review.openstack.org/194150
14:04 stevemar joined #puppet-openstack
14:05 zigo joined #puppet-openstack
14:09 dfisher joined #puppet-openstack
14:12 openstackgerrit Giulio Fidente proposed openstack/puppet-nova: Add parameter to control use of rbd for the ephemeral storage  https://review.openstack.org/119093
14:16 dfisher EmilienM:  thanks for the explanation on the revert for my cset for puppet-keystone.  Working now on puppet-cinder
14:16 EmilienM dfisher: cool
14:17 dfisher apologies for not catching that one in keystone
14:17 dfisher looks like cinder is clear after cleaning up db/sync.pp:
14:17 dfisher dfisher at rawr-3 in /export/home/dfisher/puppe​t-cinder-1458915/manifests
14:17 dfisher $ grep cinder_conf **/*(.) | grep -v cinder_config
14:17 dfisher dfisher at rawr-3 in /export/home/dfisher/puppe​t-cinder-1458915/manifests
14:18 openstackgerrit Drew Fisher proposed openstack/puppet-cinder: Remove POSIX users, groups, and file modes.  https://review.openstack.org/193189
14:28 chem joined #puppet-openstack
14:31 sergmelikyan joined #puppet-openstack
14:34 ferest joined #puppet-openstack
14:34 dtalmage joined #puppet-openstack
14:38 xingchao joined #puppet-openstack
14:41 sergmelikyan joined #puppet-openstack
14:43 sergmelikyan joined #puppet-openstack
14:49 ducttape_ joined #puppet-openstack
14:53 openstackgerrit Jason Guiditta proposed openstack/puppet-neutron: Add manage_service parameter to all agents::ml2 classes that did not yet have it.  https://review.openstack.org/194177
15:06 dmsimard gfidente: Let's add the rbd ephemeral patchset to tomorrow's meeting if it hasn't merged by then :)
15:07 gfidente dmsimard, fine for me, meeting is at same time of now, tomorrow right?
15:07 dmsimard Yeah, just about
15:08 dmsimard FWIW we're working with that patchset in our environment - works well and passes integration tests
15:12 mdorman joined #puppet-openstack
15:13 sergmelikyan joined #puppet-openstack
15:21 morazi joined #puppet-openstack
15:21 paramite joined #puppet-openstack
15:33 openstackgerrit Mike Dorman proposed openstack/puppet-nova: Add scheduler_driver parameter to nova::scheduler class  https://review.openstack.org/193609
15:35 bklei joined #puppet-openstack
15:36 mgagne joined #puppet-openstack
15:36 chandankumar joined #puppet-openstack
15:39 openstackgerrit Mike Dorman proposed openstack/puppet-nova: Add api max limit and instance naming parameters  https://review.openstack.org/193328
15:46 _ryanv_ joined #puppet-openstack
15:56 openstackgerrit Bradley Klein proposed openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
15:56 dfisher so I updated to the 5.1.0 commits and now I'm starting to see:
15:56 dfisher Notice: Puppet::Provider::Openstack: project service is unavailable.
15:56 dfisher Will retry for up to 10 seconds.
15:56 dfisher Error: /Stage[main]/Keystone_users/Keystone_user[glance]: Could not
15:56 dfisher evaluate: Execution of '/usr/bin/openstack token issue --format value'
15:56 dfisher returned 1: ERROR: openstack 'tenant_id'
15:56 dfisher I'm clearly missing a module … but I'm not sure which one
15:57 rcallawa joined #puppet-openstack
16:01 dfisher i think I see it in openstacklib .. right?
16:01 dfisher but I have the 5.1.0 commit of that module too
16:03 richm dfisher: hmm - not sure where tenant_id is coming from
16:04 richm dfisher: it should be calling token issue with OS_AUTH_URL, OS_PASSWORD, OS_USERNAME, and OS_PROJECT_NAME
16:06 richm dfisher: you could try moving /usr/bin/openstack to /usr/bin/openstack.orig, and create a shell script wrapper /usr/bin/openstack that logs the arguments, the env, and the output of calling openstack.org
16:06 mdorman do you have a kilo version of openstackclient ?
16:06 richm at least 1.0.2 or later
16:06 richm preferably 1.0.3
16:06 dfisher i grabbed the 5.1.0 tag for openstacklib
16:06 dfisher and
16:06 dfisher $ pkg list openstackclient
16:06 dfisher NAME (PUBLISHER)                                  VERSION                    IFO
16:06 dfisher library/python/openstackclient                    1.0.1-5.12.0.0.0.77.0      i-
16:06 dfisher so, that might be it
16:07 dfisher but is 1.0.2/1.0.3 suitable for Juno?
16:07 dfisher from
16:07 dfisher https://github.com/openstack/requirements​/blob/stable/juno/global-requirements.txt
16:07 dfisher python-openstackclient>=0.3.0,<=1.0.1
16:08 dfisher doesn't look like it ..
16:10 dfisher https://launchpad.net/pyth​on-openstackclient/+series doesn't mention juno or 1.0.2
16:10 mfisch joined #puppet-openstack
16:25 dfisher ok, 1.0.2 fails in the same way.
16:27 EmilienM dmsimard: I don't see any agenda topic for that, can I help with that?
16:27 dmsimard EmilienM: I'll add it to the agenda if it isn't merged by tomorrow :)
16:27 EmilienM dmsimard: very agile!
16:28 rcallawa joined #puppet-openstack
16:28 richm joined #puppet-openstack
16:31 chandankumar joined #puppet-openstack
16:35 openstackgerrit Bradley Klein proposed openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
16:36 EmilienM mfisch: would you mind to help me in adding puppet-manager-core to puppet-monasca module?
16:39 mfisch sure
16:39 mfisch whats needed?
16:40 richm dfisher: I don't see why you couldn't use > 1.0.1 with juno, unless there is some weird bug that I'm not aware of, or unless the output is subtly different, enough to cause problems
16:40 richm There's also the question of dependencies - python-openstackclient has a lot of dependencies
16:40 dfisher 1.0.2 works.  1.0.3 adds kilo deps
16:40 dfisher well
16:40 dfisher 1.0.2 works with juno.  not with puppet-keystone
16:41 dfisher i'm reverting back to 1.0.1 for now
16:41 EmilienM mfisch: permissions?
16:41 dfisher and will try to debug puppet-keystone
16:41 mfisch EmilienM: do I do it in jenkins UI or is it a code submit?
16:41 RichardRaseley joined #puppet-openstack
16:42 richm 1.0.2 doesn't work with puppet-keystone ?
16:42 EmilienM mfisch: mh, we have the same situation like in puppet-murano
16:42 EmilienM mfisch: it's not fair (I think) to switch to puppet-manager-core because we did not write the module
16:42 EmilienM mfisch: but I would like to see both groups able to vote for these modules
16:42 mfisch yes agreed
16:43 dfisher richm: for me, no
16:43 dfisher i'm trying to figure out if this is something on my end (99% likely) or something in the module itself
16:44 richm dfisher: ok - I can help - if you want me to take a look at something - paste.openstack.org
16:45 dfisher thanks
16:45 dfisher i'll let you know.
16:45 rwsu joined #puppet-openstack
16:46 mfisch EmilienM: here internally I'd just change it in jenkins but I think its a code review needed now
16:52 xingchao joined #puppet-openstack
16:53 xarses_ joined #puppet-openstack
17:00 sergmelikyan joined #puppet-openstack
17:04 openstackgerrit Denis Egorenko proposed openstack/puppet-sahara: Update Sahara module  https://review.openstack.org/192721
17:04 aimon joined #puppet-openstack
17:06 RichardRaseley richm: Are you around?
17:08 richm RichardRaseley: yes
17:13 RichardRaseley richm: This is perhaps a trivial observation on your 'How to uniquely name Keystone v3 resources in puppet?', but I was wondering about the use of the '::' as a separator for domain and user. In my mind this creates some conflation between Puppet namespaces. We already use '/' to support name-spacing in INI configs, perhaps we could use this (e.g. DOMAIN/user)?
17:14 RichardRaseley richm: (Acknowledging that this may be 100% personal preference)
17:14 richm RichardRaseley: / may used in user names - e.g. if using kerberos, a valid username might be admin/foo.example.com@EXAMPLE.COM
17:14 richm likewise \, if using AD
17:15 RichardRaseley richm: Ah, you're too correct.
17:15 RichardRaseley richm: Thank you for helping me better understand.
17:15 richm I know - it is a pita finding some sort of delimiter character that is intuitive, yet not already used . . .
17:15 morazi joined #puppet-openstack
17:15 dfisher bust out the unicode.   module<pile_of_poo>class
17:16 RichardRaseley +1 =]
17:16 richm I'm open to suggestions if "::" is too similar to ruby/puppet/perl namespacing
17:17 * richm stares at the top row of keys on the keyboard . . .
17:17 RichardRaseley richm: OK. I'll gather some feedback internally here to see if anyone has any bring ideas.
17:18 degorenko EmilienM, hi. I don't know why tests (for init.pp) are failed. Can you help me? https://review.openstack.org/#/c/192721/ For 'rabbit' rpc - ok, but for zmq and qpid - fail. I spent many time - but i really don't know why it's failed. Leave comments please.
17:20 dfisher class  openstack💩keystone
17:20 dfisher it took me WAY too long to figure out how to do that.
17:21 bklei looking for help here with a puppet-monasca build failure -- puppet 4.0 unit related, seems to pass locally for me, but jenkins no bueno for https://review.openstack.org/#/c/193175/
17:21 bklei the specific failure is var/lib/gems/1.9.1/gems/puppetlabs_s​pec_helper-0.10.3/lib/puppetlabs_spe​c_helper/module_spec_helper.rb:23:in `block in <top (required)>': undefined method `environmentpath=' for #<RSpec::Core::Configuration:0x000000018008d0> (NoMethodError)
17:21 mfisch I thought we had the 4.0 stuff working?
17:22 mfisch bklei: maybe we need to update your fixtures
17:22 mfisch I mean gemfile
17:22 bklei i think it did at one point -- but started failing at some point, i think it was non-voting so we ignored it for a bit
17:23 mfisch yeah thats it
17:23 mfisch you need rspec 2.1.0
17:23 mfisch bklei: I'll push a review
17:25 bklei i see that in the puppet-nova Gemfile, but hate flailing if i can't reproduce locally
17:25 bklei :(
17:25 mfisch you probably have rspec 2.1.0 installed on your mac
17:26 mfisch hmm gitreview is giving me troubl
17:26 mfisch trouble
17:26 bklei i had to reclone with 'openstack' instead of 'stackforge' to get gitreview working
17:26 mfisch it keeps trying to hit stackforge, wonder where that info is
17:26 mfisch yeah I did a reclone
17:27 openstackgerrit Matt Fischer proposed openstack/puppet-monasca: Sync Gemfile  https://review.openstack.org/194244
17:27 mfisch bklei: ^
17:27 mfisch I manually added the remote
17:27 bklei gracias mfisch, will see if jenkins blesses ^^ and rebase
17:28 crinkle you need https://review.openstack.org/#/c/191296/ to merge for git review to work
17:31 rcallawa joined #puppet-openstack
17:31 mfisch bklei: you have +2 on that
17:31 mfisch merge it
17:32 mfisch crinkle: what is needed to change the puppet-monasca owner group so that puppet-core can vote on the reviews?
17:32 crinkle EmilienM: ^
17:32 mfisch he's asked me to help but I dont know the process
17:33 crinkle i think buttons need to be clicked in the gerrit ui
17:33 mfisch ok
17:33 mfisch same as here
17:33 mfisch I can do that
17:33 EmilienM mfisch: thx
17:33 mfisch I thought it needed a review
17:34 mfisch puppet-manager core right?
17:35 EmilienM mfisch: yes
17:35 mfisch done
17:35 mfisch crinkle: https://review.openstack.org/#/c/194254/
17:35 Vivek joined #puppet-openstack
17:35 mfisch wth its already in merge conflict?
17:36 rwsu joined #puppet-openstack
17:36 mfisch crinkle: did you do it too or something?
17:37 crinkle mfisch: i did nothing
17:42 mfisch crinkle: I need to go ask infra
17:44 reed joined #puppet-openstack
17:49 EmilienM richm: could we add gilles's patches in the Gerrit dependencies? I think it makes sense for people who test the patches and also for the Beaker patch so we can validate all the workflow.
17:50 tshtilma joined #puppet-openstack
17:51 aimon joined #puppet-openstack
17:51 richm EmilienM: The keystone v3 patches?
17:51 EmilienM richm: yes
17:51 richm yes
17:52 iurygregory joined #puppet-openstack
17:52 EmilienM richm: I tested to deploy master, run beaker, apply all patches, run beaker and it fails now. I'm digging
17:52 richm and this one too https://review.openstack.org/189764
17:52 EmilienM richm: all the patches we need, need to be configured with dependencies
17:53 mfisch EmilienM: this is what you wanted for puppet-monasca: https://review.openstack.org/#/c/194263/
17:53 mfisch I will also ask bklei for a +1
17:53 richm EmilienM: do you mean an explicit Depends-On: Ixxxxx ?
17:53 EmilienM mfisch: yeah. Thanks. I hope it works
17:53 EmilienM richm: or classic rebase
17:54 EmilienM I prefer rebase because at least we can test them with beaker
17:54 EmilienM crinkle is working on the capability to use zuul cloner but it"s WIP afik
17:54 EmilienM richm: so if we could use git rebase that would be awesome
17:55 richm afaik all of the patches have been rebased
17:55 richm What patches need a rebase?
17:57 richm For example - https://review.openstack.org/#/c/189764/ only depends on https://review.openstack.org/187654 which has been merged
17:58 EmilienM richm: ok
17:59 EmilienM richm: maybe there is no patch to rebase then. I'm testing upgrade now
18:00 mfisch EmilienM: did you ever setup a bug tracker for puppet-monasca?
18:01 EmilienM mfisch: of course Sir (note the 'Sir') > https://bugs.launchpad.net/puppet-monasca
18:02 openstackgerrit Matt Fischer proposed openstack/puppet-monasca: Sync Gemfile  https://review.openstack.org/194244
18:03 mfisch ty
18:03 mfisch bklei: the state of lint in monasca is bad, so I will comment stuff out in that Gemfile to get it passing, I filed a bug with the issues
18:04 mfisch EmilienM: https://bugs.launchpad.net/​puppet-monasca/+bug/1467625 if you have an intern ;)
18:04 openstack Launchpad bug 1467625 in puppet-monasca "puppet-monasca has lots and lots of lint issues" [Undecided,New]
18:04 EmilienM mfisch: lol
18:08 bklei perfect, thx mfisch
18:09 mdorman mfisch/ clayton (or anybody), have you done any work to implement the oslo_messaging_rabbit/heartbeat_timeout_threshold and heartbeat_rate settings in the puppet modules?   in kilo you have to explicitly enable that stuff (default is off).   thinking about going and doing that work, but if there is stuff in flight already, i don’t want to dupe
18:10 mfisch we have not
18:10 mdorman kk
18:10 mfisch would love to have support
18:10 mdorman i assume that is something you will want
18:10 mdorman k
18:11 openstackgerrit Matt Fischer proposed openstack/puppet-monasca: Sync Gemfile  https://review.openstack.org/194244
18:22 crinkle EmilienM: https://review.openstack.org/194287
18:36 EmilienM crinkle: you can also create the governance patch if you like
18:40 crinkle EmilienM: it looks like puppet-openstack-cookiecutter and puppet-modulesync-configs aren't in governance, what is supposed to be there and what is not?
18:41 openstackgerrit Merged openstack/puppet-monasca: Sync Gemfile  https://review.openstack.org/194244
18:41 EmilienM crinkle: good point. I suggest we import the repos in governance in one shot
18:41 EmilienM spredzy, sbadia ^
18:42 EmilienM crinkle, richm: I've tested upgrade from puppet-keystone/master to v3 patches and everything looks okay when you don't change your manifests
18:54 rcallawa joined #puppet-openstack
19:06 _ody EmilienM: https://review.openstack.org/#/c/194097...reasoning behind no running ruby code through spec tests?  Since we only run rspec-puppet I completely forgot to even look at providers.
19:06 EmilienM social: ^
19:08 richm EmilienM: ok
19:10 _ody ...no...we do according to the rake task.
19:11 social _ody: this is kind of internal puppet issue, if you have idea how to test this I'm all ears
19:12 chem richm: are you working on the 'absent' user for Keystone_user_role which make the final beaker test fail. I think I could help here.
19:13 _ody social: Yeah.  Reading thorugh tests now.  Just surprised the test didn't bomb out with a similar no method found error.
19:15 richm chem: yes
19:15 richm chem: The problem is deeper than I thought - re: my recent email to openstack-dev
19:16 richm chem: The real problem is - how does self.instances create the right resource name?
19:16 richm chem: How does it know when the resource name needs to be 'name::domain' or just plain 'name'?
19:17 richm When I am writing a manifest, how do I know that the username needs to be 'glance::services' rather than just keystone_user { 'glance': domain => 'services' }?
19:17 richm I mean, yeah, we could bang out a solution to that particular problem and fix the acceptance test
19:18 chem richm: yes, I think I see.  I'm going over the mailing list to dig on the details, thanks
19:19 richm I'm working on a poc for keystone_tenant to see if I can do something that works with existing manifests, still allows you to do a simple name of 'glance' even if 'glance' is in the 'services' domain, and allows full support for multi-domain v3 installations
19:19 _ody social: Oh.  We just don't set tenant_name in the tests and the code is only triggered if it is.
19:27 rcallawa joined #puppet-openstack
19:28 rcallawa joined #puppet-openstack
19:32 spredzy EmilienM, crinkle puppet-openstack-cookiecutter is under review https://review.openstack.org/#/c/192636/ for governance
19:33 crinkle spredzy: oh I see
19:34 EmilienM spredzy: and I +1... :)
19:34 spredzy crinkle, EmilienM if you want me to squash puppet-modulesync-config in it I don't mind. If you prefer another review I don't mind either way :)
19:34 EmilienM it's okay
19:38 rwsu joined #puppet-openstack
19:40 rwsu joined #puppet-openstack
19:41 openstackgerrit Mike Dorman proposed openstack/puppet-nova: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194315
19:47 richm chem: EmilienM: something like this - http://paste.openstack.org/show/314143/
19:47 richm I have to run for an appt. - back later
19:47 mgarza joined #puppet-openstack
19:50 openstackgerrit Bradley Klein proposed openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
19:55 openstackgerrit Bradley Klein proposed openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
19:56 openstackgerrit Bradley Klein proposed openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
20:01 chem richm: looking into it ...
20:03 openstackgerrit Mike Dorman proposed openstack/puppet-ceilometer: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194336
20:08 chem richm: the first comment is weird regarding the code "# Use long name for resource if not unique and not in default domain" and you have == default_domain_id just above (L7). Am I missing something ?
20:15 chem richm: oki, I got the hang of it.  Your talking about the new resource...
20:16 openstackgerrit Mike Dorman proposed openstack/puppet-cinder: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194341
20:17 openstackgerrit Dan Prince proposed openstack/puppet-tripleo: Add a function to write package names  https://review.openstack.org/194342
20:19 openstackgerrit Mike Dorman proposed openstack/puppet-cinder: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194343
20:25 dgurtner joined #puppet-openstack
20:31 iurygregory joined #puppet-openstack
20:32 openstackgerrit Mike Dorman proposed openstack/puppet-glance: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194354
20:35 chem richm: so I get this right this is solution 1) of your post on the mailing list, isn't ?
20:39 openstackgerrit Colleen Murphy proposed openstack/puppet-openstack_extras: Fix kilo repo installation for RedHat  https://review.openstack.org/193761
20:42 chem richm: with this code you cannot predict what will be the name of the resource from one run to another, is that ok ?
20:44 iurygregory joined #puppet-openstack
20:49 openstackgerrit Mike Dorman proposed openstack/puppet-heat: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194369
20:50 openstackgerrit Mike Dorman proposed openstack/puppet-glance: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194354
20:52 openstackgerrit Mike Dorman proposed openstack/puppet-cinder: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194343
20:52 openstackgerrit Mike Dorman proposed openstack/puppet-nova: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194315
20:53 openstackgerrit Mike Dorman proposed openstack/puppet-ceilometer: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194336
21:01 rcallawa joined #puppet-openstack
21:03 rcallawa_ joined #puppet-openstack
21:04 rcallawa_ joined #puppet-openstack
21:07 openstackgerrit Mike Dorman proposed openstack/puppet-neutron: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194381
21:10 rwsu joined #puppet-openstack
21:18 richm chem: That's one way to describe the problem - I cannot predict what the resource name will be in the manifest
21:23 chem richm: that I get, but I was talking about this paste http://paste.openstack.org/show/314143/
21:24 richm chem: This paste attempts to use the short name when it is unique.  If the short name is not unique, it uses the short name if it is in the default domain.  Otherwise, it uses the long name.
21:26 chem richm: well, exactly, so if we take your example in the ML, we can end up with admin { domain => dom1} and admin::dom2 or the opposite depending on how the hash is build
21:27 richm chem: right - it is non-deterministic
21:28 richm chem: so how can we make it deterministic so that we have 'admin' and 'admin::dom2' every time?
21:28 richm if dom1 is the default domain, it will be 'admin' and 'admin::dom2' every time
21:28 chem richm: yeap
21:29 richm but if neither dom1 nor dom2 is the default domain, we don't have a way for an operator to say "ok - prefer dom1 for shortnames"
21:29 chem richm: but I think that solution 2) in the mailing list is the way to go
21:29 chem short name only for default domain
21:29 richm chem: So that means we will always have 'admin::dom1' and 'admin::dom2' and never just plain 'admin'
21:30 richm chem: I think that will cause problems in this case
21:30 richm chem: Consider the case where you want to put regular users in the default domain backed by the enterprise LDAP identity backend, and put the service accounts in the domain 'services'
21:30 chem richm: if the user is not using domain then it will work as expected.  If the user is using domain then he/she uses full domain name
21:31 richm That means you must use 'glance::services', 'cinder::services', etc. everywhere
21:31 richm which will break a lot of existing manifests
21:31 chem hum ...
21:32 * EmilienM cries
21:32 richm If there were a way for operators to tell puppet "Use short names for resources in these domains: Default, services, etc."
21:32 richm I mean, the easy way to do it would be "ok - everyone must use long names everywhere" which would of course break the world, but maybe it is worth the one time pain?
21:34 chem richm: with a default namespace without the long name.  The services domain is mostly internal openstack machinery and could be solved (I've got a  idea to test the name of the resource)
21:35 EmilienM richm: can't we do 'if just plain admin', then use default domain elseif admin::domain_admin then domain=domain_admin
21:35 EmilienM ?
21:35 chem richm: then when you don't use don't care about domain, everything work as expected and when you use them .. well you have to type more character ^^
21:35 richm EmilienM: Are you talking about converting the output of openstack user list to a puppet resource, or taking a puppet resource name and figuring out which actual user it refers to?
21:36 EmilienM richm: #2
21:36 chem EmilienM: I have the feeling that it's risky ... like unsecure
21:36 chem EmilienM: but it's just a feeling ..
21:37 richm chem: EmilienM: If we can guarantee that no other puppet module refers to a keystone_user, keystone_tenant, or keystone_user_role (or the forthcoming keystone_group, etc.) resource directly - that is, we "hide" those resources from other puppet modules, then it might work
21:37 richm We already mostly do that - thanks to the keystone::resource::service_identity and keystone::roles::admin
21:38 richm But there are a lot of places that refer to keystone_user_role
21:38 richm and if someone is using openstack directly or the REST API directly . . .
21:38 richm and heat is doing something weird with v3 and domains already
21:38 richm ditto neutron
21:40 richm If we did have some sort of global setting like "Use the short name for resources in these domains: Default, services, etc." - where/how would such a setting be done?  facter?  env. var.?
21:41 openstackgerrit Mike Dorman proposed openstack/puppet-keystone: Add support for RabbitMQ connection heartbeat  https://review.openstack.org/194399
21:44 chem richm: IMHO, the fact that we cannot predict the name of the resource to reference is the problem here.  So with solution 1)  we add to the turmoil, so I would rather go with 2) and refactore the current puppet module using keystone name directly.  For the user the rules would be simple to use and adopt.  But I may be missing the big picture here.
21:45 chem richm: ^^ facter is not very user friendly.  Not you would have to do that in the keystone class entry point, the only thing that is unique in the system, I think.
21:46 richm chem: Yeah, I agree, 2) is simple and deterministic.  But it means that if I want to put service accounts in their own separate domains, we have to make sure all existing manifests that refer to 'glance' change to use 'glance::services' instead - same with 'cinder', 'nova', 'neutron', 'swift', 'heat', etc. etc.
21:47 richm Which may cause an operator revolt . . .
21:49 zigo joined #puppet-openstack
21:49 chem richm: well I see your point ...
21:50 richm On the other hand, if we can encapsulate all references to domain scoped keystone resources, so that puppet-glance, puppet-cinder, etc., and composition layers, and orchestration layers, don't have to know or care what the resource names are, then this would be a good solution
21:50 chem richm: That definitively the way to go :)  Is that doable ?
21:51 chem richm: are you talking about the auth.pp file in all those modules ?
21:51 chem richm: or is there other cases ?
21:52 richm chem: I don't know if there are other cases, but that's the main one - the references to keystone_user_role
21:52 richm which implicitly refer to the keystone_user and keystone_tenant names
21:53 richm we could get rid of that code - have e.g. glance::keystone::auth pass in a list of service names to keystone::resource::service_identity
21:54 richm and just have keystone::resource::service_identity set up the user role ~> Service[name] notifications
21:55 xavpaice joined #puppet-openstack
21:57 chem richm: well, encapsulation, determinism, retro-compatibility, happy user. It looks like a good idea
21:57 richm swift::keystone::dispersion references keystone_user directly
21:59 richm so does neutron::server::notifications
22:00 richm that looks like it as far as the standard openstack puppet modules shipped with rdo
22:01 wabu joined #puppet-openstack
22:02 chem richm: same finding here that's the only module that uses elsewhere that in the auth.pp modules (there is the  auth_cfn.pp in heat, but that's nearly a auth.pp also)
22:02 richm ok
22:02 chem I think I can handle this part if you like
22:03 dalees joined #puppet-openstack
22:03 richm Is self.instances used for anything except to do the search for self.prefetch, and for the puppet resource command?
22:04 xarses joined #puppet-openstack
22:04 crinkle those are the two uses that i know of
22:06 richm does it matter that puppet resource may show different resource names than are in the manifest?
22:07 chem richm: ^^ I'm looking
22:07 richm that is, puppet resource could show keystone_user { 'admin::dom1': domain => 'dom1' }, but I have keystone_user { 'admin': domain => 'dom1' } in my manifest, and they both refer to the same keystone_user?
22:08 crinkle richm: it's okay that puppet resource shows an id property when the manifest doesn't, so by that rationalization i think it's fine if puppet resource shows something slightly different than the manifest
22:08 blkperl joined #puppet-openstack
22:11 richm ok
22:11 richm so then maybe it is ok if puppet resource always shows the fully qualified name
22:12 crinkle i think that makes sense
22:12 richm then it's just a little extra work for self.prefetch to match the resource from the manifest with the output of of self.instances
22:13 richm crinkle that also means that "puppet resource keystone_user admin" will show nothing
22:14 richm that may surprise a lot of operators who use puppet resource
22:14 dfisher erm.  keystone_service is really unhappy with openstackclient 1.0.1 …
22:15 richm dfisher: what do you get?
22:16 dfisher ah, crap.   that's what I get for stashing things in /tmp  ...
22:16 dfisher i rebooted
22:16 dfisher i had to do this:
22:16 dfisher def create
22:16 dfisher -    properties = ['--name']
22:16 dfisher -    properties << resource[:name]
22:16 dfisher +    properties = []
22:16 dfisher if resource[:description]
22:16 dfisher properties << '--description'
22:16 dfisher properties << resource[:description]
22:16 dfisher end
22:16 dfisher raise(Puppet::Error, 'The service type is mandatory') unless resource[:type]
22:16 dfisher +    properties << '--type'
22:16 dfisher properties << resource[:type]
22:16 dfisher +    properties << resource[:name]
22:16 dfisher self.class.request('service', 'create', properties)
22:16 dfisher @property_hash[:ensure] = :present
22:16 dfisher http://paste.openstack.org/show/314508/
22:16 dfisher there's no —name flag
22:17 dfisher and no —type flag in the cmd construciton
22:17 bklei joined #puppet-openstack
22:17 zigo joined #puppet-openstack
22:19 crinkle i thought we depended on >=1.0.3?
22:20 dfisher juno's cap is 1.0.1
22:22 richm When you are using OS_IDENTITY_API_VERSION=3, you have to do service create --name $name $type
22:23 dfisher still using v2
22:23 richm but I thought with API v2 you used --type $type $name
22:23 dfisher ^ exactly what I posted
22:24 richm I'm looking at the keystone_service/openstack.rb code
22:24 richm for v2, it uses --name $name $type
22:24 richm I think that's wrong
22:25 dfisher it is :)
22:25 richm crap
22:25 dfisher you see my paste?
22:25 richm yes
22:25 dfisher ok, good.
22:25 dfisher i got yelled at by freenode for flood :)
22:25 dfisher didn't know if you saw it.
22:25 richm lol
22:27 openstackgerrit Merged openstack/puppet-monasca: Update vertica projections for better performance  https://review.openstack.org/193175
22:27 dfisher i'm not against switching everything to v3
22:27 dfisher just … not sure yet how :)
22:27 richm dfisher: neither are we :P\
22:27 dfisher hahahhaha
22:28 dfisher ^ how openstack works.
22:28 richm dfisher: no, the problem is that the v3 code snuck into the v2 implementation
22:28 rushiagr_away joined #puppet-openstack
22:28 richm that needs to be reverted back to the v2 style
22:28 richm your patch is correct
22:29 * dfisher nods
22:31 * richm considers waiting for gildub to fix this . . .
22:31 dfisher you could if you want.  i have a local patch for it
22:32 richm dfisher: ok - then go ahead and submit a review
22:32 dfisher do I need a launchpad bug?
22:32 dfisher no problems if I do ...
22:32 richm dfisher: it falls under the auth_restructure bp
22:33 dfisher erm.
22:33 dfisher uh.
22:33 dfisher threeve.
22:33 dfisher :)
22:33 richm looks like gildub put some v3 code into the auth_restructure patch
22:38 xingchao joined #puppet-openstack
22:40 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: beaker: test to list keystone resources  https://review.openstack.org/194425
22:47 jasondotstar joined #puppet-openstack
22:49 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: beaker: test to list keystone resources  https://review.openstack.org/194425
22:49 saneax joined #puppet-openstack
22:50 openstackgerrit Drew Fisher proposed openstack/puppet-keystone: Restore Keystone v2 method for service create  https://review.openstack.org/194428
22:50 dfisher richm: posted
22:51 dfisher not sure I did the commit message right.   i think I did, but I've gotten used to being wrong :)
22:52 richm dfisher: ack
22:54 dfisher hey, so, good-ish news:  i have solaris multi-node puppet working.  one controller node, one volume node, one compute node - all deployed via puppet.
22:58 richm nice!
23:00 dfisher man, when puppet starts clicking … it's so amazing
23:00 nibalizer we're gonna put that on the box
23:00 nibalizer :)
23:01 dfisher nice.
23:06 EmilienM dfisher: congrats
23:07 dfisher thanks :)
23:08 dfisher once I get this stuff working a little bit more solidly
23:08 dfisher i'll start working on upstream
23:10 dfisher can puppet metadata.json files use optional dependencies?
23:11 dfisher for example, puppet-swift wants to depend on xinetd.  Solaris doesn't have xinetd
23:11 dfisher and neutron::puppet-vswitch.
23:11 dfisher etc.
23:11 RichardRaseley _ody: ^^ RE: Solaris
23:11 RichardRaseley dfisher: +1 =]
23:12 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack_extras: auth_file: allow to change the path  https://review.openstack.org/194431
23:13 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: beaker: test to list keystone resources  https://review.openstack.org/194425
23:15 dfisher ok, well, i'll have to dig up an answer to my dependency thing tomorrow.  gotta split....
23:17 aimon joined #puppet-openstack
23:19 aimon_ joined #puppet-openstack
23:22 linkedinyou joined #puppet-openstack
23:24 xingchao joined #puppet-openstack
23:54 jasondotstar joined #puppet-openstack

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