Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-04-18

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

All times shown according to UTC.

Time Nick Message
00:05 thuc joined #puppet-openstack
00:06 thuc joined #puppet-openstack
00:08 thuc_ joined #puppet-openstack
00:09 thuc__ joined #puppet-openstack
00:14 thuc joined #puppet-openstack
00:54 rmoe joined #puppet-openstack
01:02 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
01:07 kitp joined #puppet-openstack
01:24 thuc joined #puppet-openstack
01:48 kitp joined #puppet-openstack
02:05 rmoe joined #puppet-openstack
02:14 Aimon_ joined #puppet-openstack
02:19 xingchao joined #puppet-openstack
02:37 xarses joined #puppet-openstack
02:37 rmoe joined #puppet-openstack
03:28 kitp joined #puppet-openstack
03:34 rmoe joined #puppet-openstack
03:39 lnxnut joined #puppet-openstack
03:39 kitp joined #puppet-openstack
03:41 kitp_ joined #puppet-openstack
03:54 thuc joined #puppet-openstack
03:58 thuc_ joined #puppet-openstack
04:26 thumpba_ joined #puppet-openstack
04:42 kitp joined #puppet-openstack
04:42 thuc joined #puppet-openstack
05:31 EmilienM good morning
05:43 kitp joined #puppet-openstack
05:53 thuc joined #puppet-openstack
05:55 thumpba_ joined #puppet-openstack
05:56 sanjayu joined #puppet-openstack
06:03 dgollub joined #puppet-openstack
06:38 dgollub joined #puppet-openstack
06:44 kitp joined #puppet-openstack
07:26 EmilienM very useful for us: https://wiki.openstack.org/​wiki/ReleaseNotes/Icehouse
07:39 bauzas joined #puppet-openstack
07:44 kitp joined #puppet-openstack
08:22 trefzer joined #puppet-openstack
08:45 kitp joined #puppet-openstack
09:13 admin0 joined #puppet-openstack
09:26 admin0 joined #puppet-openstack
09:37 hogepodge joined #puppet-openstack
09:46 kitp joined #puppet-openstack
10:18 admin0 joined #puppet-openstack
10:47 kitp joined #puppet-openstack
11:16 waterkinfe joined #puppet-openstack
11:47 kitp joined #puppet-openstack
11:56 dprince joined #puppet-openstack
12:08 lnxnut joined #puppet-openstack
12:34 PaxIndustria joined #puppet-openstack
12:39 PaxIndustria joined #puppet-openstack
12:48 kitp joined #puppet-openstack
12:53 larsks Good mornin EmilienM.
12:54 EmilienM larsks: hey
13:12 EmilienM larsks: i'm reviewing your patch
13:12 larsks Spiffy.
13:12 meylor1 joined #puppet-openstack
13:12 larsks If you have experience writing rspec tests, I'm probably going to need some help with that.
13:13 EmilienM larsks: ok, i'll see
13:17 meylor1 left #puppet-openstack
13:24 PaxIndustria joined #puppet-openstack
13:26 EmilienM larsks: 1st review done
13:26 EmilienM larsks: open to discussion here
13:32 kitp joined #puppet-openstack
13:37 larsks EmilienM: Thanks, finished looking through the comments.  Regarding the spec tests, right now things are failing because the module is unable to contact an actual keystone server ("Failed to connect to Keystone server at http://127.0.0.1:35357/v2.0").  Do we need to mock out the keystone_* functions in keystone_tenant_by_name.rb?
13:38 larsks Presumably that's also how we would add tests for the new failures.
13:38 prad_ joined #puppet-openstack
13:40 larsks EmilienM: Re: your comments about examples/neutron.pp, a quick test on the command line suggests that maybe we can use a single set of credentials.  I'm going to make those changes and see what happens in an actual packstack run.
13:44 thuc_ joined #puppet-openstack
13:48 kkirkpatrick joined #puppet-openstack
13:57 EmilienM larsks: ok
14:02 kkirkpatrick joined #puppet-openstack
14:03 daneyon joined #puppet-openstack
14:06 kevin joined #puppet-openstack
14:09 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
14:10 larsks EmilienM: So, it seems to work with just the nova_* credentials, so I've stripped out all the keystone_* parameters.
14:10 EmilienM larsks: awesome
14:10 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
14:10 EmilienM larsks: we have to find a way for spec tests
14:15 larsks Yeah.  I think for that to work we're going to need to mock out some of the functions in keystone_tenant_by_name.rb. I'm reading about that right now.  It's possible I will need to reorganize the code so that we can fake the right sorts of errors.
14:16 EmilienM larsks: and btw, I -2 my patch myself, because I had a +2
14:17 namredips joined #puppet-openstack
14:21 larsks This has some mocking examples: https://github.com/TomPoulto​n/rspec-puppet-unit-testing
14:21 larsks Looking through it right now.
14:22 EmilienM oh yes, the MockFunction example
14:22 EmilienM awesome, I did not know that
14:28 larsks EmilienM: Do you have any examples of how to run "rpsec" tests locally?
14:30 EmilienM larsks: of course:
14:31 EmilienM rake spec 'SPEC_OPTS=--format documentation'
14:34 larsks Okay, I thought maybe I was doing something wrong, but running "rake spec" I get NoMethodError: undefined method `lookup' for Puppet:Module
14:34 larsks ...for *all* the tests.
14:46 thuc joined #puppet-openstack
14:49 fvollero EmilienM: :)
14:50 fvollero EmilienM: pro active in packstack world today ? :)
14:50 larsks fvollero: Have you seen that error from rspec before?
14:50 EmilienM fvollero: nah, we don't use packstack, we use our puppet module (open source)
14:50 fvollero EmilienM: ol'school
14:51 fvollero larsks: reading up
14:51 fvollero larsks: nop, never saw that error before
14:52 larsks fvollero: Darn.  Even trivial no-op tests fail with that error on my system.
14:52 markvoelker larsks: running rspec tests locally?  Did you see this? https://wiki.openstack.org/wiki/Pu​ppet-openstack#Rspec_puppet_tests
14:53 fvollero larsks: but you have the right fixtures ?
14:53 larsks fvollero: I think?
14:53 larsks markvoelker: Reading...
14:53 larsks markvoelker: I had not seen that previously.
14:53 fvollero larsks: bundle install the right stuff? :)
14:55 larsks I had bundle installed all the right stuff...but it was the lack of "bundle exec" that was causing my grief.  markvoelker, thanks!
14:55 markvoelker larsks: cool, glad it helped
14:57 prad_ joined #puppet-openstack
14:59 openstackgerrit Pradeep Kilambi proposed a change to stackforge/puppet_openstack_builder: Fix to specify the endpoint url directly  https://review.openstack.org/87372
14:59 namredips joined #puppet-openstack
15:03 admin0 joined #puppet-openstack
15:04 larsks markvoelker fvollero Do either of you have experience mocking out custom functions in rspec tests?
15:08 dtalton joined #puppet-openstack
15:11 thuc_ joined #puppet-openstack
15:13 markvoelker larsks: no, unfortuantely
15:19 daneyon joined #puppet-openstack
15:23 thuc joined #puppet-openstack
15:39 thuc joined #puppet-openstack
15:40 thuc joined #puppet-openstack
15:48 kitp_ joined #puppet-openstack
16:07 thuc joined #puppet-openstack
16:15 rmoe joined #puppet-openstack
16:18 meylor joined #puppet-openstack
16:22 thuc joined #puppet-openstack
16:38 thuc_ joined #puppet-openstack
16:51 Aimon joined #puppet-openstack
16:59 xarses joined #puppet-openstack
17:02 thuc joined #puppet-openstack
17:21 angdraug joined #puppet-openstack
17:23 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
17:23 larsks EmilienM: There, mocked http requests.  Now the existing tests pass, and I added a test suite for the custom function.
17:23 larsks Let's see if Jenkins likes it.
17:24 EmilienM larsks: awesome, I'm looking with pop corn \o/
17:25 larsks Always writing the tests is more difficult than writing the code.
17:26 EmilienM larsks: oops
17:27 larsks ?
17:28 EmilienM larsks: it failed
17:28 EmilienM i may be wrong but the jenkins jobs does not run rspec using bundle
17:29 EmilienM undefined method `to_json'
17:50 EmilienM joined #puppet-openstack
17:53 larsks EmilienM: Hmmm. I'll take a closer look later this afternoon, I guess :/
17:57 namredips joined #puppet-openstack
18:01 hogepodge Anyone feel good about checking this patch? I’m running it locally, and it’s really nice for moving Tempest off of the control node. https://review.openstack.org/#/c/86751/
18:05 hogepodge (say EmilienM or fvollero)
18:05 hogepodge :-)
18:13 thuc joined #puppet-openstack
18:31 xarses joined #puppet-openstack
18:32 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
18:34 marun joined #puppet-openstack
18:46 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
18:48 fvollero hogepodge: i was about to leave, but i'll do later today or in the weekend :)
18:50 fvollero hogepodge: and could you give a look at this? https://review.openstack.org/#/c/88305/
18:56 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
19:00 larsks EmilienM: Passing tests! https://review.openstack.org/#/c/88410/
19:00 larsks fvollero: If you haven't already, can you take a look at https://review.openstack.org/#/c/88410/ and let me know it looks sane?
19:00 larsks No rush.
19:03 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
19:04 thumpba_ joined #puppet-openstack
19:07 markvoelker larsks: was just testing that and ran into the ordering issue I'd been fearing (this code tries to run before Keystone is setup).  Time to go track down the higher-order resource chaining I need to do. =)
19:08 larsks markvoelker: How/where are you running this?  I'm deploying with packstack, which applies things serially so tends to hide/not be affected by this issue as long as the global ordering is correct.
19:08 markvoelker larsks: nice job getting the gate checks working though
19:09 larsks markvoelker: Wah, tests are hard.
19:09 markvoelker I'm using stackforge/puppet_openstack_builder as the comp layer on a fresh server
19:10 larsks markvoelker: Maybe making the neutron::server::notifications class have an optional dependency on the keystone class?
19:11 markvoelker larsks: yeah, although I need to check where the nova admin tenant creation gets done.  That's probably the real resource we want to depend on
19:11 larsks markvoelker: Right, good point.
19:12 larsks markvoelker: Looks like nova::keystone::auth
19:13 markvoelker larsks: yep, just looking at that
19:16 larsks markvoelker: Is that going to look something like Nova::Keystone::Auth<||> -> Neutron::Server::Notification<||>
19:18 markvoelker larsks: something like that yeah. Or we could tie it directly to the resource rather than the class I think.
19:18 markvoelker larsks:  (sorry, in a troubleshooting w/someone on another thing while doing this so may be a bit slow to reply)
19:21 larsks Tie it directly to the resource like... <| auth_name == "nova" |>?
19:21 larsks NO worries about slow responses :)
19:30 dmsimard_ joined #puppet-openstack
19:30 thumpba_ joined #puppet-openstack
19:38 rmoe joined #puppet-openstack
19:38 marun_ joined #puppet-openstack
19:46 larsks markvoelker: Oh, right, like on the keystone_user resources.  See, the advantage of slow responses is that I actually do more than scan the header of the manifest.
20:07 larsks markvoelker: I added Keystone_user <| title == $nova_admin_username |> -> Class['neutron::server::notifications'] to the manifest and at least it didn't break anything.  Sending to gerrit.
20:07 openstackgerrit Lars Kellogg-Stedman proposed a change to stackforge/puppet-neutron: Neutron Nova Interactions Support  https://review.openstack.org/88410
20:09 thuc joined #puppet-openstack
20:13 thuc_ joined #puppet-openstack
20:18 markvoelker larsks: I'd tried that and it didn't seem to work, unfortunately. =/  Will try to investigate further in a bit when I'm not multitasking.
20:24 larsks Sure.  Just out of question did it not work as in "broke with error" or not work as in "still tried to look for nova before the principal had been created"?
20:24 markvoelker larsks: the latter
20:25 larsks Okay.  I will try to set up a test environment where I can investigate this, too.
20:26 markvoelker larsks: worst case I may have to wait until Monday to debug it...family coming into town today so I'm expecting my phone to start ringing soon =p
20:27 larsks That's okay, I've got family this weekend, too :)
20:30 * markvoelker 's phone rings right on cue as if it knew it were being talked about
20:31 markvoelker larsks: I'll see if I can find some time this weekend to poke at it. Otherwise, have a good weekend!
20:35 marun joined #puppet-openstack
20:47 daneyon joined #puppet-openstack
20:50 daneyon joined #puppet-openstack
20:50 thuc joined #puppet-openstack
21:17 PaxIndustria joined #puppet-openstack
21:23 thumpba_ joined #puppet-openstack
21:23 thumpba__ joined #puppet-openstack
21:28 larsks markvoelker: if I run http://chunk.io/f/e888b17224a94dd0838ddca91b09c324
21:29 larsks Then I get a dependency graph http://chunk.io/f/365f87db9e084b1189a81c6169873782
21:29 larsks Which seems to imply the correct relationship between the keystone_user and the neutron::server::notifications class.
21:42 larsks Arg, it's possible that puppet just runs functions too darn early.
21:46 marun_ joined #puppet-openstack
21:59 marun joined #puppet-openstack
21:59 hogepodge larsks looks like good work there.
22:00 larsks hogepodge: Thanks.  It's been a learning experience. :)
22:01 hogepodge I like how you’re using the JSON endpoint.
22:01 larsks I just hope this dependency issue doesn't prove intractable.
22:01 hogepodge Calling the clients always struck me as the wrong way to do it.
22:01 hogepodge What’s the issue exactly? How is it manifesting?
22:01 larsks The issue is that the keystone principals used by the lookup function to authenticate to keystone need to be created first.
22:02 larsks If you look at http://chunk.io/f/e888b17224a94dd0838ddca91b09c324
22:02 larsks The custom function will try to connect to keystone before the "nova" user is created.
22:02 hogepodge Yup. This might be a problem. I run into the same thing with Tempest. Hunter did some work to address it there.
22:03 hogepodge see https://review.openstack.org/#/c/86751/
22:03 hogepodge It may not be exactly the same, though. He was looking at resources.
22:03 hogepodge Yup.
22:04 hogepodge In the best case, you can figure out how to anchor the dependencies. In the worst case the providers are being pre-cached.
22:05 dmsimard hogepodge: !
22:05 dmsimard Are we getting a hangouts for the summit? :)
22:05 hogepodge Yes.
22:05 dmsimard Awesome.
22:05 dmsimard mgagne will be at the summit but i'm staying in Montreal :(
22:05 hogepodge I just closed the RFP today. I’ll be coming up with the final schedule next week.
22:06 dmsimard A RFP ? So you're going to go big ?
22:09 hogepodge I think we have 8 submissions, so not huge.
22:09 hogepodge I need to look at the list again.
22:09 hogepodge Are the any ceph developers here?
22:11 hogepodge Oh yes, I see dmsimard
22:11 hogepodge :-)
22:11 dmsimard I guess I sort of fit in that category
22:11 hogepodge I need to get more involved with that.
22:11 dmsimard dachary is probably more of a developer than I am, he goes pretty deep in the code
22:11 dmsimard Knowing him, he's probably going to the summit
22:12 hogepodge You might want to ask for your own time before the RFP closes in case you want to really spend a session focusing on ceph.
22:12 dmsimard puppet-ceph ?
22:12 hogepodge yeah.
22:12 hogepodge although I reached out to inktank and they submitted a proposal also.
22:12 dmsimard I'll check with dachary, is the RFP on the mailing list ?
22:12 dmsimard They did ?
22:12 dmsimard Who ?
22:13 dmsimard That'd be odd because no one from Inktank actively works on puppet (or chef anymore)
22:13 dmsimard AFAIK, at least
22:14 hogepodge It’s not puppet work specifically, but ceph integration with OpenStack.
22:14 dmsimard Ah, yes.
22:14 dmsimard dachary might want to talk about puppet-ceph, i'll send him an e-mail. How do we get in touch for the RFP ?
22:15 mgagne hogepodge: any links to those RFP? I think I missed it
22:15 hogepodge http://www.openstack.org/blog/2014/03/call-​for-proposals-open-source-openstack-summit/
22:16 hogepodge It was on the OpenStack blog, and I’ve tweeted about it a bit.
22:16 hogepodge I’ve taken care of it for the puppet-openstack work.
22:17 dmsimard OH
22:17 dmsimard I thought you meant if we wanted a spot on the hangout of something
22:17 dmsimard (this has been a long week..)
22:17 hogepodge Oh, no.
22:17 hogepodge Everyone is taken care of for that. :-)
22:17 dmsimard Good.
22:18 mgagne hogepodge: I'll have to find a better way to catch those announcements... :O
22:18 hogepodge Nobody reads blogs any more. It’s so aughts.
22:19 mgagne hogepodge: my RSS reader is rotting...
22:22 hogepodge I’ll keep using mine until my secret Fever sever is shut down.
22:28 mgagne hogepodge: do you think there is a way to migrate something using puppetlabs-apache<=0.6.0 with the template parameter from apache::vhost in a backward compatible way?
22:29 hogepodge You want to migrate forward?
22:29 mgagne hogepodge: yep
22:29 mgagne hogepodge: well, in a way that would support both versions
22:29 hogepodge Have you looked at the work michchap did for mysql?
22:29 hogepodge https://github.com/stackforge/puppet-neutron​/commit/2b310c22100446cc117809ae9b0e3b671bcf​7f12#diff-92f31c6843c08811f71bc76449d7542d
22:30 mgagne hogepodge: the challenge is that the template parameter got dropped in 0.7.0 in favor of custom_fragment
22:30 hogepodge (for example)
22:30 dachary hogepodge: nice idea this call for proposal.
22:30 mgagne hogepodge: if only there was a way to determine the module version at run time =)
22:30 mgagne hogepodge: or available parameters
22:31 hogepodge Yeah, at some point we want to drop backwards support. Why do we need to keep the old apache version?
22:35 mgagne hogepodge: trying to see if there is a way to update your manifests without having to run and update your apache module
22:35 hogepodge I see.
22:37 thumpba_ joined #puppet-openstack
22:48 kkirkpatrick joined #puppet-openstack
22:51 kitp joined #puppet-openstack
23:01 hogepodge mgagne “make a fact that ssh's to the master and checks the metadata.json” (kidding)
23:03 mgagne hogepodge: well, I was thinking about something similar, if only there was a way for the puppetmaster to parse the actual module version from Modulefile or metadata.json and make it available at catalog compilation time.
23:03 hogepodge It would be nice.
23:03 mgagne hogepodge: this would greatly ease migration and support for multiple versions of a core module such as apache, mysql, etc.
23:03 hogepodge It’s no fun having to pass a parameter.
23:04 mgagne hogepodge: o rly? =)
23:04 hogepodge Who am I kidding. I’m all about new parameters!
23:04 mgagne hogepodge: are puppet functions evaluated from top to bottom?
23:05 mgagne hogepodge: could those help or would it be a fragile concept depending on order and phase of the moon?
23:06 hogepodge resources will be evaluated in file order sometime soon.
23:06 mgagne hogepodge: on #openstack: imcsk8: hello, i'm having a problem installing neutron using the puppet module can somebody give me a hand? http://pastebin.com/RkgyGpgM
23:07 hogepodge For types/providers I think that classes are created as early as possible, meaning ‘static’ variables (I’m using C++ terminology there) can be loaded earlier than you might expect.
23:07 mgagne hogepodge: will try something
23:14 mgagne hogepodge: those guys on #openstack-infra are hilarious
23:15 mgagne <mordred> because if you do that too much, I'm going to replace all of our puppet infrastructure with a pile of makefiles
23:15 mgagne <mordred> and THEN where will we be?
23:15 mgagne <clarkb> the 80s
23:15 hogepodge :-)
23:16 mgagne hogepodge: they do have an interesting problem, how to orchestration puppet runs across multiple nodes
23:16 mgagne hogepodge: with complex cross node dependencies
23:17 hogepodge Yeah. It’s a sad that it’s a problem for them.
23:17 mgagne hogepodge: well, it is for us too
23:17 hogepodge swift.
23:17 hogepodge tempest
23:17 hogepodge controller vs workers
23:17 mgagne hogepodge: yep
23:18 hogepodge It only gets worse if you want to start installing things like corosync
23:18 mgagne hogepodge: when testing with vagrant, we "fixed" this problem on our side by creating a provision script waiting for dependencies to complete before provisioning the next one
23:18 hogepodge I do the same.
23:19 hogepodge Run controller.
23:19 hogepodge Then run the rest of the nodes.
23:19 hogepodge Then run tempest.
23:19 thumpba__ joined #puppet-openstack
23:19 mgagne hogepodge: so I feel like there is a layer of orchestration missing in the great scheme of things
23:19 hogepodge I try to live in an ideal world where I don’t have to do multiple puppet runs to convergence, though.
23:20 hogepodge There is much pain surrounding that, and although it may not seem like it, the global ‘we’ are listening.
23:51 thumpba_ joined #puppet-openstack
23:52 kitp joined #puppet-openstack
23:57 mgagne hogepodge: ok, almost got it. But internal puppet module utils expects metadata.json to exists...

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