Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-11-20

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

All times shown according to UTC.

Time Nick Message
00:19 mgagne anyone having issues with mongodb upstart script where status is always "stop/waiting" ?
01:16 rmoe joined #puppet-openstack
01:41 aimon joined #puppet-openstack
01:41 xarses joined #puppet-openstack
01:47 dgurtner joined #puppet-openstack
01:47 dgurtner joined #puppet-openstack
01:48 mmaglana joined #puppet-openstack
01:51 JayJ joined #puppet-openstack
02:02 xingchao joined #puppet-openstack
02:12 richm joined #puppet-openstack
02:14 vinsh joined #puppet-openstack
02:18 ericpeterson joined #puppet-openstack
02:28 richm left #puppet-openstack
02:37 mmagr joined #puppet-openstack
03:00 badiane_ka joined #puppet-openstack
03:06 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Refactorise Keystone resources management  https://review.openstack.org/134493
03:06 EmilienM If someone can have a look, I can't figure out why unit tests are failing, I may have miss something ^
03:17 mmaglana joined #puppet-openstack
03:17 rcallawa joined #puppet-openstack
03:27 EmilienM I found it, it's an error I did in the code
03:29 off_rhoden joined #puppet-openstack
03:36 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Refactorise Keystone resources management  https://review.openstack.org/134493
03:49 vinsh joined #puppet-openstack
04:05 giorgiodinapoli joined #puppet-openstack
04:17 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Configure database parameters on the right nodes  https://review.openstack.org/135840
04:17 EmilienM michchap: mgagne: I've got an idea ^
04:32 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: Refactorise Keystone resources management  https://review.openstack.org/135841
04:37 openstackgerrit Emilien Macchi proposed stackforge/puppet-neutron: Refactorise Keystone resources management  https://review.openstack.org/135842
04:49 openstackgerrit Emilien Macchi proposed stackforge/puppet-nova: Configure database parameters on the right nodes  https://review.openstack.org/135840
06:13 fc__ joined #puppet-openstack
06:14 saneax_-_ joined #puppet-openstack
06:21 rcallawa joined #puppet-openstack
06:26 xavpaice I have a tiny patch for puppet-neutron, but when I'm running rake spec I'm getting errors - even from a clean checkout of master
06:26 xavpaice is it that I'm running the command wrong (just 'rake spec') or am I more likely missing something on my workstation?
06:30 dachary EmilienM: I know you care about CI and that's a great confort :-) At which point of the roadmap is it going to happen ?
06:34 xavpaice in case anyone wants to see the failures https://gist.github.com/xavpaice/10888a2743b1db7d5e8c
06:46 sputnik13 joined #puppet-openstack
06:48 sputnik13 joined #puppet-openstack
06:49 sputnik13 joined #puppet-openstack
06:51 sputnik13 joined #puppet-openstack
07:06 ToMiles joined #puppet-openstack
07:17 stupidnic joined #puppet-openstack
07:32 rcallawa joined #puppet-openstack
07:43 rcallawa joined #puppet-openstack
07:45 rcallawa_ joined #puppet-openstack
07:47 * xavpaice rtfm's - much better now, thanks for great documentation :)
07:52 tfz joined #puppet-openstack
08:00 kbyrne joined #puppet-openstack
08:01 tdb joined #puppet-openstack
08:09 stamak joined #puppet-openstack
08:10 openstackgerrit Xav Paice proposed stackforge/puppet-neutron: Remove rubygems requirement  https://review.openstack.org/135878
08:10 igajsin joined #puppet-openstack
08:22 tfz joined #puppet-openstack
08:58 rcallawa joined #puppet-openstack
09:00 rcallawa_ joined #puppet-openstack
09:07 openstackgerrit Bogdan Dobrelya proposed stackforge/puppet-neutron: Raise puppet error, if N-API anavail  https://review.openstack.org/130156
09:11 derekh joined #puppet-openstack
09:13 dgurtner joined #puppet-openstack
09:35 aarefiev joined #puppet-openstack
09:35 dkaigarodsev joined #puppet-openstack
09:51 JayJ joined #puppet-openstack
09:58 JayJ joined #puppet-openstack
10:03 JayJ joined #puppet-openstack
10:08 JayJ joined #puppet-openstack
10:13 JayJ joined #puppet-openstack
10:18 JayJ joined #puppet-openstack
10:21 sanjayu_ joined #puppet-openstack
10:23 JayJ joined #puppet-openstack
10:28 dgurtner joined #puppet-openstack
10:28 JayJ joined #puppet-openstack
10:33 JayJ joined #puppet-openstack
10:38 JayJ joined #puppet-openstack
10:44 JayJ joined #puppet-openstack
10:46 xingchao_ joined #puppet-openstack
10:48 sanjayu_ joined #puppet-openstack
10:49 JayJ joined #puppet-openstack
10:54 JayJ joined #puppet-openstack
10:59 JayJ joined #puppet-openstack
11:04 JayJ joined #puppet-openstack
11:09 JayJ joined #puppet-openstack
11:14 JayJ joined #puppet-openstack
11:19 JayJ joined #puppet-openstack
11:24 JayJ joined #puppet-openstack
11:29 JayJ joined #puppet-openstack
11:32 ddmitriev joined #puppet-openstack
11:34 JayJ joined #puppet-openstack
11:40 JayJ joined #puppet-openstack
11:45 JayJ joined #puppet-openstack
11:50 JayJ joined #puppet-openstack
11:55 JayJ joined #puppet-openstack
12:01 tfz joined #puppet-openstack
12:05 EmilienM good morning!
12:18 mmagr joined #puppet-openstack
12:18 sputnik13 joined #puppet-openstack
12:19 ToMiles /disconnect
12:23 openstackgerrit Merged stackforge/puppet-nova: Set force_snat_range parameter  https://review.openstack.org/130488
12:39 dachary dmsimard: rspec-beaker tests can now be run in the infra, see openstack-infra/puppet-storyboard for an example
12:39 dmsimard dachary: Yeah, I saw something to that effect yesterday
12:40 dachary that's a big incentive :-)
12:40 dmsimard dachary: So much for trying to do our own stuff
12:41 dmsimard dachary: I'm happy that they got the leverage needed to have infra handle it but I *just* got the new server :p
12:41 dmsimard dachary: Perhaps we can have it run third party CI or something, still.. let's wait and see
12:42 dachary the tests now work again, that's what matter
12:42 dachary it's great to have an evolution path though. it also means there is now zero excuse to *not* do integration tests on stackforge puppet modules ;-)
12:43 dachary no more "yes but ..."
12:44 dachary to be fair the need to setup a hardware machine was a significant blocker for most puppet module developers
13:15 richm joined #puppet-openstack
13:29 dgurtner joined #puppet-openstack
13:30 morazi joined #puppet-openstack
13:32 ibravo joined #puppet-openstack
13:32 dgurtner -win 1
13:34 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support the ldap user_enabled_invert parameter  https://review.openstack.org/129951
13:36 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support Keystone with read-only LDAP identity backend  https://review.openstack.org/133601
13:55 richm joined #puppet-openstack
14:06 khappone_ joined #puppet-openstack
14:07 rlaurika joined #puppet-openstack
14:08 xingchao joined #puppet-openstack
14:15 khappone joined #puppet-openstack
14:16 rlaurika joined #puppet-openstack
14:17 openstackgerrit Nicolas Hicher proposed stackforge/puppet-cinder: Create type-key only if it doesn't exist  https://review.openstack.org/112955
14:18 dmsimard dachary: Where are the tests for storyboard ?
14:25 khappone joined #puppet-openstack
14:25 mmagr joined #puppet-openstack
14:29 khappone_ joined #puppet-openstack
14:30 rlaurika joined #puppet-openstack
14:40 khappone joined #puppet-openstack
14:41 rlaurika joined #puppet-openstack
14:46 rlaurika joined #puppet-openstack
14:51 dachary dmsimard: I don't know, just copy/pasting someone else comment ;-)
14:54 pradk joined #puppet-openstack
14:58 emccormick joined #puppet-openstack
14:59 kbyrne joined #puppet-openstack
15:01 dgurtner joined #puppet-openstack
15:01 dgurtner joined #puppet-openstack
15:02 vinsh joined #puppet-openstack
15:07 ibravo joined #puppet-openstack
15:07 ericpeterson joined #puppet-openstack
15:10 francois1 joined #puppet-openstack
15:18 stamak joined #puppet-openstack
15:20 patcable hey, https://github.com/puppetlabs/puppetlabs-openstack/blob/master/manifests/profile/rabbitmq.pp is missing an openstack::resources::firewall entry for RabbitMQ
15:23 crinkle patcable: what to make a pull request? i've run into that but i was "fixing" it by changing /etc/hosts
15:23 crinkle s/what/want
15:29 patcable crinkle: id love to, but Its Complicated™ with work
15:30 patcable i could do it when i get home though
15:30 patcable so ill do that
15:35 richm crinkle: I'm working on adding some v3 features to puppet-keystone, using your code from openstacklib and puppet-keystone - I'm trying to run spec tests - I've changed .fixtures.yml in puppet-keystone to use symlink to my ~/puppet-openstacklib - but spec fails - http://paste.openstack.org/show/135342/
15:42 vinsh joined #puppet-openstack
15:47 crinkle richm: your spec tests need to add the fixtures/openstacklib/lib directory to the $LOAD_PATH since it's talking across modules, where in a real environment they would all live in /var/lib/puppet/lib so the load path isn't a problem
15:48 crinkle for example first line of https://review.openstack.org/#/c/134844/1/spec/unit/provider/keystone_tenant/openstack_spec.rb
15:49 richm crinkle: Is there some environment variable I can set, rather than having to edit the .rb file?
15:50 crinkle richm: i'm not sure
15:50 crinkle but for the spec to pass on ci it needs to be in the file, afaik
15:50 JayJ joined #puppet-openstack
15:51 richm crinkle: ok
15:51 crinkle there might be a better way to do it, i'm not sure
15:52 mdorman joined #puppet-openstack
15:54 richm this seems to work: $ RUBYLIB=~/puppet-openstacklib/lib GEM_HOME=vendor bundle exec rake spec
15:55 crinkle oh cool
15:57 dgurtner joined #puppet-openstack
15:57 dgurtner joined #puppet-openstack
15:58 richm crinkle: should I expect a lot of spec test failures?
16:06 crinkle richm: if you look at my review i hadn't worked out all the tests yet, so kind of depends what you're doing and whether you got further than me
16:07 ibravo joined #puppet-openstack
16:08 richm crinkle: https://rmeggins.fedorapeople.org/puppet-keystone.spec.output.txt
16:09 crinkle yep i hadn't fixed keystone_spec.rb yet
16:10 crinkle gonna work on it soon
16:11 crinkle i'll bbiab, gotta go to work
16:13 dgurtner joined #puppet-openstack
16:19 mdorman i need to build a unit test for https://review.openstack.org/#/c/134667/ which tests the behavior of nova_admin_tenant_id_setter based on the contents of neutron.conf.  is there a way in rspec to stub out a dummy neutron.conf that’s used for the context of that test?
16:25 EmilienM mdorman: like using pre_condition and setup a class like neutron::server::notifications with some params?
16:26 xingchao joined #puppet-openstack
16:30 mdorman EmilienM:  what you said doesn’t really make sense to me, but yes something along the lines of a pre condition sounds right. :)
16:30 EmilienM mdorman: kind of middle of a meeting now, I'll have a look asap
16:30 mdorman np. thanks
16:37 imcsk8 joined #puppet-openstack
16:39 openstackgerrit Mike Dorman proposed stackforge/puppet-neutron: Implement better nova_admin_tenant_id_setter exists? method  https://review.openstack.org/134667
16:41 mmaglana joined #puppet-openstack
16:53 mschmitt joined #puppet-openstack
16:56 crinkle mdorman: keystone does it like https://github.com/stackforge/puppet-keystone/blob/master/spec/unit/provider/keystone_spec.rb#L56-L58
16:58 EmilienM crinkle: thanks I've learnt one thing today at least
16:58 mdorman crinkle:  thanks
16:59 dgurtner joined #puppet-openstack
16:59 dgurtner joined #puppet-openstack
17:00 dtrainor joined #puppet-openstack
17:10 dmsimard mdorman: You're the one setting up beaker tests ?
17:18 mdorman dmsimard:  nope
17:19 dmsimard mdorman: Ah, ok. np :p
17:32 tfz joined #puppet-openstack
17:39 openstackgerrit Mike Dorman proposed stackforge/puppet-neutron: Implement better nova_admin_tenant_id_setter exists? method  https://review.openstack.org/134667
17:40 mdorman EmilienM / crinkle, think i finally got the tenant id setter test working ^
17:40 ToMiles joined #puppet-openstack
18:06 RichardRaseley joined #puppet-openstack
18:06 arnaud_orange left #puppet-openstack
18:18 EmilienM mdorman: awesome!
18:19 mdorman rspec is mostly black magic to me :)
18:23 rmoe joined #puppet-openstack
18:25 sputnik13 joined #puppet-openstack
18:26 richm crinkle: what is versioned_endpoint supposed to do?  tack on a "/v2.0" or "/v3" to the given url?
18:28 richm crinkle: If so, how does it know if it should use v2.0 or v3?
18:28 crinkle richm: I think so, but I thought I got rid of that in the openstackclient iteration
18:29 crinkle richm: let me look
18:29 richm crinkle: Is this commit the latest?  "Use openstackclient for keystone_tenant"
18:29 crinkle yes
18:30 crinkle what file is it?
18:30 richm puppet-keystone/lib/puppet/provider/keystone.rb
18:34 crinkle i think I copypasted that from the aviator version of keystone_tenant but I took it out of openstacklib
18:34 crinkle that shouldn't even work
18:36 crinkle in the openstackclient version i just had an auth_url key in the auth parameter which should include the /v2.0 or /v3
18:36 crinkle in a meeting right now, will try to fix in a bit
18:37 richm crinkle: Thanks
19:09 vinsh joined #puppet-openstack
19:15 vinsh joined #puppet-openstack
19:33 rwsu joined #puppet-openstack
19:42 xarses joined #puppet-openstack
19:49 rcallawa joined #puppet-openstack
19:53 openstackgerrit Colleen Murphy proposed stackforge/puppet-keystone: Use openstackclient for keystone_tenant  https://review.openstack.org/134844
20:01 openstackgerrit Malik Bougacha proposed stackforge/puppet-nova: upporting lxc cpu mode The default cpu_mode with the lxc libvirt type doesn't support the cpu_mode configuration. If one is defined, the nova-compute service will fail.  https://review.openstack.org/136108
20:03 EmilienM there is still a patch needed for juno support: https://review.openstack.org/#/c/131576/
20:04 EmilienM mgagne: crinkle: what do you think about this one ? https://review.openstack.org/#/c/135840/
20:05 richm crinkle: looks much better - now I have only 2 spec failures related to the fact that the spec expects the endpoint to end in v2.0/
20:08 vinsh joined #puppet-openstack
20:10 crinkle EmilienM: I don't think that would work, I don't think you can just use parameters like ::nova::sql_connection without including ::nova
20:11 crinkle the parameters should be passed into the class or stuck in something like a nova::config or nova::globals
20:11 crinkle richm: \o/
20:15 richm crinkle: these are the diffs required to make spec/unit/provider/keystone_spec.rb pass at 100%
20:15 richm crinkle: http://paste.openstack.org/show/135480/
20:15 openstackgerrit Malik Bougacha proposed stackforge/puppet-nova: supporting lxc cpu mode Fixing the default cpu_mode from None to none  https://review.openstack.org/136108
20:17 crinkle richm: okay cool
20:22 richm crinkle: I don't like self.keystone_file and keystone_file - should keystone_file just call self.keystone_file?
20:22 richm I don't know enough ruby to know what to do in this situation
20:25 aimon joined #puppet-openstack
20:28 crinkle richm: i had a reason for making them different
20:29 crinkle i don't quite recall what it was
20:29 richm crinkle: so is it ok as it is?  If so, should I go ahead and submit a review?
20:29 crinkle it had to do with using instance variables in an uninstantiated class
20:30 crinkle let me work on it a bit and see
20:30 richm or, if you like, just take my patch and roll it into yours
20:31 badiane_ka joined #puppet-openstack
20:33 cwolferh joined #puppet-openstack
20:36 mschmitt joined #puppet-openstack
20:43 mschmitt joined #puppet-openstack
20:44 crinkle richm: with self.get_admin_token_from_keystone_file I would like to handle the error in self.keystone_request, so I need to change the test to reflect that rather than changing self.get_admin_token_stuff to raise an ambiguous error
20:44 richm crinkle: ok
20:45 crinkle for self.keystone_file I had the instance method using instance variables to save time and the class method rereading the file each time because i think there were issues with using an instance variable in a class method
20:45 crinkle i'll update my review
20:45 crinkle i'm hesitant to roll yours into mine since there is still indecision about whether this is the right direction
20:46 richm ok
20:47 EmilienM crinkle: I'm doing this in my composition layer and also in the unit tests and it works. How can it work?
20:47 EmilienM crinkle: isn't ::nova automatically included?
20:51 crinkle it might sometimes or even usually work if ::nova is parsed before ::nova::scheduler or maybe there is a collector dependency setup somewhere, but it shouldn't be counted on
20:51 crinkle https://docs.puppetlabs.com/puppet/latest/reference/lang_scope.html#accessing-out-of-scope-variables
20:51 crinkle mgagne: do you agree? ^
20:51 EmilienM crinkle: ok. So I should add it everywhere, right?
20:52 crinkle what is "it"?
20:53 EmilienM crinkle: ::nova
20:53 mgagne crinkle: could you summarize the conversation in one sentence? is it about keystone_file or something else?
20:54 crinkle mgagne: about https://review.openstack.org/#/c/135840 , moving database config setup out of nova init.pp
20:54 dgurtner_ joined #puppet-openstack
20:54 mgagne crinkle: I think it adds complexity for little gain
20:54 mgagne crinkle: if you wish to clear database_connection from compute node, add something in your composition layer to remove it
20:55 crinkle mgagne: this is EmilienM's patch
20:55 EmilienM mgagne: seriously?
20:55 mgagne crinkle: I would refrain from making important design changes so close to juno branching
20:55 EmilienM mgagne: we don't have to merge it today, I don't care
20:55 EmilienM mgagne: I wanted to bring-up this old blueprint
20:55 vinsh joined #puppet-openstack
20:56 crinkle EmilienM: it seems like if you have to include nova in every class then you're losing the value of moving it out of the nova class
20:56 EmilienM mgagne: and I thought about doing something like this and discuss with you. To me the composition layer should not fix this.
20:56 EmilienM crinkle: I just want to move out the database credentials from compute nodes
20:56 mgagne EmilienM: I agree but the proposed design isn't ideal due to backward compatibility crap we just kept
20:56 EmilienM which is to me a huge security issue
20:57 EmilienM mgagne: I can refactore the code
20:57 mgagne EmilienM: it is for me too and I fixed it in my composition layer
20:57 EmilienM mgagne: I want to fix the upstream
20:58 mgagne EmilienM: yes, like all the openstack contributors who wishes to "fix" upstream with their drivers and stuff. I however don't think we should propose such changes without proper (re)design
21:00 EmilienM mgagne: I can add a nova::db::connection which use the code from nova::init, and use ensure_resource('class', 'nova::db::connection' ...) in scheduler, api & conductor?
21:00 jamespd joined #puppet-openstack
21:00 mgagne EmilienM: how about include class instead
21:01 mgagne EmilienM: one thing I will strongly enforce is consistency
21:01 EmilienM mgagne: yeah, include or ensure_resource that works for me
21:01 mgagne EmilienM: see https://github.com/stackforge/puppet-ceilometer/blob/master/manifests/db.pp
21:01 mgagne EmilienM: why can't nova follow the same pattern?
21:01 EmilienM +1
21:01 mgagne EmilienM: nova::db should be used
21:01 EmilienM mgagne: let's do it
21:01 EmilienM how do we manage backward compat though?
21:01 mgagne EmilienM: or we refactor ceilometer with a better name
21:01 EmilienM thought*
21:01 mgagne EmilienM: I don't see any backward compatibility issue
21:01 EmilienM hum
21:02 mgagne EmilienM: is nova::database a batter name?
21:02 EmilienM because we are going to drop the code in nova::init
21:02 mgagne EmilienM: nova::init defines the variable, nova::database uses it, nova::scheduler includes it
21:02 mgagne EmilienM: I don't see any backward compatibility issues
21:02 EmilienM mgagne: fair enough, I'll do it.
21:02 EmilienM there is not.
21:03 EmilienM mgagne: thx for feedback
21:03 mgagne EmilienM: yw
21:03 mgagne crinkle: what do you think about the proposed solution?
21:04 crinkle I think that will face the same issue as the current setup, nova::database can't just grab variables from nova::init
21:05 crinkle er not current setup, EmilienM's proposed change
21:05 mgagne crinkle: user already has to include nova::init to use nova::scheduler
21:05 mgagne crinkle: it's not enforced but user will have bad surprises if he doesn't
21:06 mgagne EmilienM: how about we use nova::database to configure database? and "migrate" $nova::database_connection and related database configs to $nova::database::connection instead?
21:07 mgagne EmilienM: have a $connection parameter to nova::database which defaults to undef and use pick() in implementation to pick $nova::database::connection if defined and fallback to $nova::database_connection if not defined
21:08 crinkle why not have nova::init include nova::database for a deprecation cycle?
21:09 mgagne crinkle: true. people with ONLY nova::init for obscure reason (like being able to run db sync) won't have database connection defined to run db sync
21:12 * EmilienM catches up the backlog
21:12 EmilienM crinkle: I may miss something, but what is the reason of not including nova::database from nova::{api,conductor,scheduler} directly?
21:13 mgagne EmilienM: I think we are not saying to NOT include it
21:13 mgagne EmilienM: but to ALSO include it from nova::init for some time
21:14 mgagne EmilienM: due to possible use case mentioned above
21:14 EmilienM mgagne: what will it change?
21:14 EmilienM ok
21:14 EmilienM dbsync things, gotcha
21:14 EmilienM ok, let's do it... I'll do the patch later though
21:14 crinkle hrm, that might cause duplicate declaration errors though
21:14 EmilienM crinkle: ensure_resource?
21:15 crinkle oh alright
21:15 EmilienM :)
21:15 mgagne crinkle: if nova::init includes nova::database, I don,t see problem
21:15 crinkle okay
21:15 mgagne crinkle: include makes it idempotent
21:15 EmilienM mgagne: why not using ensure_resource
21:15 mgagne EmilienM: the question is: WHY using ensure_resource?
21:15 mgagne EmilienM: you don't need it
21:15 EmilienM mgagne: for the duplicate declaration thing
21:15 EmilienM but okay
21:16 crinkle if you're just including nova::database then you aren't passing the parameters from init into it
21:16 mgagne EmilienM: I don't see a valid use case where there would be a resource duplication
21:16 mgagne crinkle: you don't need to
21:16 crinkle so you're still having nova::database grab variables from nova::init, which while I guess it will work still seems ugly to me
21:16 mgagne crinkle: you should see all puppetlabs modules
21:16 crinkle heh
21:17 mgagne crinkle: if have bigger problems if it's an issue
21:17 mgagne you*
21:17 mgagne or we8
21:17 crinkle fair enough
21:17 mgagne https://gist.github.com/mgagne/6ed00808815b0af0fe0e
21:18 EmilienM mgagne: -1, you miss unit tests in your gist.
21:18 EmilienM :)
21:19 mgagne EmilienM: =)
21:20 mgagne crinkle: are class resources in the catalog or are they merely containers used during catalog compilation and then discarded?
21:26 crinkle they have properties of both in my opinion
21:27 crinkle i'm ready to drop it, the proposed solution seems fine to me
21:34 emccormick hey guys, do we have a process for submitting feature requests? Or is it pretty much an "if you want it, code it" sort of thing? ;)
21:34 emccormick would I need to do something like the openstack projects do specs or blueprints?
21:35 EmilienM emccormick: depends of the feature, if it requires design, we need a specs
21:36 EmilienM emccormick: can you tell us more about your feature?
21:36 emccormick I think this is not so complicated but you tell me. Keystone WSGI lets you specify a private and admin port and creates 2 virtual servers, but uses the same servername, bind, and ssl_ variables for both
21:37 emccormick I would love for those to be able to be defined independently per service / virtual server
21:37 emccormick sorry, public and admin port I meant to say
21:38 jgrassler joined #puppet-openstack
21:39 emccormick my usecase being that I've got all internal stuff running on one interface, public stuff on another with different hostnames and certificates. Also I really don't want 35357 open on the public interface even if it is firewalled off
21:39 emccormick do I make sense?
21:41 dgurtner joined #puppet-openstack
21:42 richm emccormick: so you want to have a $public_bind_host and $admin_bind_host?
21:42 emccormick indeed. And also public_ssl_* / admin_ssl_* and public_servername_* and admin_servername_*
21:43 emccormick though the servername is mostly ornamental here I think so not so so important
21:44 ibravo joined #puppet-openstack
21:44 richm it's almost as if you would want keystone::wsgi::apache to be a resource instead of a class - so you could do keystone::wsgi::apache { 'public.host.tld': ... public parameters } and keystone::wsgi::apache { 'private.host.tld': .... private parameters }
21:44 emccormick that would work as well
21:44 emccormick probably more elegant
21:46 emccormick for now I was just going to hack those parameters into apache.pp, but obviously that's not a long term solution ;)
21:48 emccormick also from the "probably too much damn work" department, I'd love if the wsgi stuff was in the other modules too, but one thing at a time.
21:50 richm emccormick: https://bugs.launchpad.net/puppet-keystone/+bug/1394765
21:51 richm emccormick: Does that accurately convey your requirements?
21:53 richm emccormick: https://bugs.launchpad.net/puppet-openstacklib/+bug/1394767
21:53 emccormick Yes, perfect
21:54 emccormick even better :)
21:55 emccormick +1 on both.
22:03 richm crinkle: How would you do get_keystone_object in the new openstack way?
22:15 rwsu_ joined #puppet-openstack
22:18 xarses joined #puppet-openstack
22:20 sputnik13 joined #puppet-openstack
22:22 crinkle richm: i think the call to auth_keystone(type-get,id) would be replaced with request(thing, 'show', id, resource[:auth]), and all the stripping of the tabular formatting would go away and be replaced with JSON.parse(...)
22:22 crinkle self.authenticate_request in openstacklib would have to be modified to add the --format json for the show action
22:23 crinkle i'm wondering if we need to keep that method at all or if there's a way to get that info via the list action
22:32 morazi joined #puppet-openstack
22:32 richm crinkle: looking . . .
22:34 richm crinkle: no - I think "show" is the way to go - "list" doesn't appear to have a filter or selector option - don't want to have to parse a 100000 member user list
22:36 crinkle okay
22:36 richm crinkle: the reason I'm asking is because I'm working on converting keystone_user/keystone.rb to openstack.rb
22:43 RichardRaseley joined #puppet-openstack
22:51 mdorman joined #puppet-openstack
22:52 sputnik13 joined #puppet-openstack
22:53 sputnik13 joined #puppet-openstack
23:27 aimon joined #puppet-openstack
23:33 gildub joined #puppet-openstack
23:43 crinkle richm: regarding get_keystone_object for keystone_user, it looks like it's used to get the name of the tenant it belongs to, but since that information is already part of `openstack user list --long` it should already be part of the instance, right?
23:43 crinkle it doesn't seem necessary to me to make another call for it
23:43 crinkle but maybe there's something i'm not seeing
23:44 richm crinkle: yes, looks like it has been added to the openstack user list --long output
23:59 openstackgerrit Colleen Murphy proposed stackforge/puppet-openstack-specs: Use OpenStackClient in Module REsources  https://review.openstack.org/136186
23:59 openstackgerrit Colleen Murphy proposed stackforge/puppet-openstack-specs: Use OpenStackClient in Module Resources  https://review.openstack.org/136186

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