Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-12-05

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

All times shown according to UTC.

Time Nick Message
00:19 openstackgerrit joined #puppet-openstack
00:59 topshare joined #puppet-openstack
01:34 aimon joined #puppet-openstack
01:40 badiane_ka joined #puppet-openstack
01:43 mschmitt joined #puppet-openstack
01:44 rmoe joined #puppet-openstack
02:03 xarses joined #puppet-openstack
02:05 xingchao joined #puppet-openstack
02:06 rcallawa joined #puppet-openstack
02:10 openstackgerrit Bob Callaway proposed stackforge/puppet-manila: Use CSV instead of array for share helpers  https://review.openstack.org/137536
02:14 rcallawa joined #puppet-openstack
03:02 topshare joined #puppet-openstack
03:22 cwolferh_ joined #puppet-openstack
03:37 markvoelker joined #puppet-openstack
04:12 xarses joined #puppet-openstack
04:30 topshare joined #puppet-openstack
04:42 topshare joined #puppet-openstack
05:03 markvoelker joined #puppet-openstack
05:03 sanjayu joined #puppet-openstack
05:33 fedexo joined #puppet-openstack
05:43 gazzaP joined #puppet-openstack
05:47 gazzaP quit
05:49 gazzaP joined #puppet-openstack
05:56 gazzaP hi all... trying to install openstack juno on centos 7 on two virtual machines (one controller, one compute node) and keep receiving this error message:
05:56 gazzaP ERROR : Error appeared during Puppet run: 172.25.1.10_neutron.pp
05:56 gazzaP Error: Duplicate declaration: Vs_bridge[br-eth1] is already declared in file /var/tmp/packstack/126dd607496e43adb94afac0c1​f9c846/manifests/172.25.1.10_neutron.pp:105; cannot redeclare at /var/tmp/packstack/126dd607496e43a​db94afac0c1f9c846/modules/neutron/​manifests/plugins/ovs/bridge.pp:9 on node os-controller.example.com
05:56 gazzaP You will find full trace in log /var/tmp/packstack/20141205-052358-klWJ​ln/manifests/172.25.1.10_neutron.pp.log
05:57 gazzaP any thoughts on what could be causing this?
06:00 sanjayu joined #puppet-openstack
06:23 topshare joined #puppet-openstack
06:27 mfisch joined #puppet-openstack
06:47 crinkle EmilienM: looks like openstackclient just dropped the 1.0.0 https://pypi.python.org/pypi/​python-openstackclient/1.0.0
06:49 openstackgerrit joined #puppet-openstack
06:51 openstackgerrit Ryan Hefner proposed stackforge/puppet-manila: Change share server config details  https://review.openstack.org/137534
07:08 topshare joined #puppet-openstack
07:11 dgurtner joined #puppet-openstack
07:11 dgurtner joined #puppet-openstack
07:26 topshare joined #puppet-openstack
07:34 markvoelker joined #puppet-openstack
08:20 rmoe joined #puppet-openstack
08:21 arnaud_orange joined #puppet-openstack
08:27 berendt joined #puppet-openstack
08:33 mmagr joined #puppet-openstack
08:36 markvoelker joined #puppet-openstack
08:39 gazzaP joined #puppet-openstack
08:47 dalgaaf joined #puppet-openstack
08:53 ddmitriev joined #puppet-openstack
08:56 tfz joined #puppet-openstack
08:58 derekh joined #puppet-openstack
09:15 mfisch joined #puppet-openstack
09:43 jistr joined #puppet-openstack
09:49 stamak joined #puppet-openstack
10:21 omolchanov joined #puppet-openstack
10:38 markvoelker joined #puppet-openstack
11:03 gchamoul joined #puppet-openstack
11:09 derekh joined #puppet-openstack
11:46 derekh joined #puppet-openstack
12:40 markvoelker joined #puppet-openstack
13:21 kskurtveit joined #puppet-openstack
13:23 TorLaedre joined #puppet-openstack
13:45 TorLaedre joined #puppet-openstack
13:45 kskurtveit joined #puppet-openstack
14:06 EmilienM good morning!
14:06 EmilienM crinkle: w00t
14:06 EmilienM it changes the game so
14:07 derekh joined #puppet-openstack
14:15 kskurtveit joined #puppet-openstack
14:16 TorLaedre joined #puppet-openstack
14:18 sbadia hi
14:19 sbadia crinkle: yep, sorry for the delay, it was just a question (about supported modules ;-))
14:21 kbyrne joined #puppet-openstack
14:28 richm1 joined #puppet-openstack
14:43 berendt joined #puppet-openstack
14:56 EmilienM richm1: https://pypi.python.org/pypi/​python-openstackclient/1.0.0
14:56 EmilienM \o/
15:01 mfisch joined #puppet-openstack
15:03 para joined #puppet-openstack
15:09 prad joined #puppet-openstack
15:11 rcallawa joined #puppet-openstack
15:12 markvoelker joined #puppet-openstack
15:12 EmilienM spredzy: I was wondering, by using include + data-binding, can you write rspec tests like we do in puppet-openstack-cloud ?
15:14 spredzy EmilienM, you can. But it doesn't make sense to actually do it. What I am trying to say is that you need  to test that you include the class. But now what the class will later do. This is up to the class itself to test
15:15 spredzy am I making any sense ? We had a discussion about that for the policy patch. Should we test only that the class is called or the implementation of the called class
15:15 EmilienM spredzy: what if i want to test my Hiera config?
15:16 spredzy EmilienM, IMO to test that you would rather use beaker than rspec-puppet
15:21 richm1 crinkle: ^^^ - openstackclient 1.0.0 is in pypi
15:32 mdorman joined #puppet-openstack
15:40 ericpeterson joined #puppet-openstack
15:54 prad joined #puppet-openstack
16:05 markvoelker joined #puppet-openstack
16:41 mschmitt joined #puppet-openstack
16:54 markvoelker joined #puppet-openstack
16:54 crinkle richm1: yep
16:54 crinkle but we need it in uca and rdo
16:54 crinkle and debian?
16:55 crinkle you guys might be able to poke somebody about rdo, but uca will only be for kilo
16:57 EmilienM crinkle: richm1: IMHO, we should target Aviator right now.
16:57 EmilienM because of that
16:58 crinkle what if we added keystone v3 using what we already have - the current keystone clients
16:58 crinkle then that could be backported, and we can work on updating the providers for kilo
16:58 crinkle with openstackclient
17:05 crinkle EmilienM: richm1 ^
17:13 richm1 what is uca?
17:14 crinkle ubuntu cloud archive
17:14 richm1 agreed - keep working on aviator
17:14 richm1 I've filed a request to update rdo
17:15 cwolferh_ joined #puppet-openstack
17:21 EmilienM crinkle: +1
17:22 aimon joined #puppet-openstack
17:36 ericpeterson joined #puppet-openstack
17:55 rickerc_ joined #puppet-openstack
18:00 rickerc_ joined #puppet-openstack
18:01 xarses joined #puppet-openstack
18:02 rwsu joined #puppet-openstack
18:08 mmagr joined #puppet-openstack
18:38 badiane_ka joined #puppet-openstack
18:39 aimon joined #puppet-openstack
19:01 mschmitt joined #puppet-openstack
19:31 crinkle EmilienM: richm1: is that one for proceed with aviator and one for use the current clients for now? or is that both for aviator?
19:33 EmilienM crinkle: I think until we can't do something with openstackclient, let's use Aviator for now. I'm not sure to answer to the question
19:34 crinkle okay
19:34 gchamoul joined #puppet-openstack
19:34 EmilienM crinkle: though I may be wrong, I'm not sure we should continue the wait. Aviator seems working with Keystone v3, let's consume it from our modules
19:35 EmilienM crinkle: if tomorrow we have openstackclient available in upstream packaging, let's use it. For now it's not, so we waste our time
19:35 crinkle yeah
19:35 EmilienM crinkle: as you mentioned so well, the transition is not a big deal. You almost did all the work.
19:36 EmilienM crinkle: richm1: do you have some patches we can review already?
19:36 crinkle not yet
19:37 EmilienM crinkle: so that one is going to be in stale: https://review.openstack.org/#/c/134843/
19:37 aimon joined #puppet-openstack
19:37 EmilienM crinkle: and this one is WIP by you and Rich: https://review.openstack.org/#/c/116754/
19:38 EmilienM crinkle: this one in stale too: https://review.openstack.org/#/c/134844/
19:38 crinkle yep I will fix them all up
19:39 EmilienM crinkle: so your roadmap is to continue the work on "Use aviator as a REST interface for keystone_XXXX providers"
19:39 EmilienM crinkle: and then, bring v3 support?
19:39 EmilienM crinkle: how will it work?
19:40 crinkle well aviator can take an api version parameter
19:40 crinkle will probably need richm1 to work on the specifics
19:41 EmilienM oh nice
19:41 EmilienM crinkle: and also we will add to add new providers
19:41 EmilienM to manage domains
19:41 EmilienM trusts, etc
19:42 EmilienM crinkle: and then, I'll patch https://github.com/stackforge/puppet-keystone/bl​ob/master/manifests/resource/service_identity.pp
19:43 EmilienM crinkle: let's write down this in our etherpad we already did: https://etherpad.openstack.org/p​/puppet-keystone-v3-api-support
19:47 EmilienM crinkle: so we have to change something in openstacklib for v3 ?
19:47 EmilienM the source of aviator, right?
19:48 crinkle we shouldn't have to
19:48 crinkle well maybe
19:49 crinkle i will definitely make sure it's configurable
19:49 EmilienM crinkle: AFIK, aviator from stackforge is old
19:49 EmilienM and does not support v3
19:49 crinkle EmilienM: yes, we have to use mark's github version
19:49 EmilienM yep
19:49 crinkle https://github.com/aviator/aviator
19:50 EmilienM let's use that one
19:50 crinkle and https://github.com/aimonb/puppet_aviator
19:50 crinkle yep that was established before
20:02 bodepd I remember that there was a patch that might have been rejected where someone wanted to remove the pre-fetching feature from keystone b/c they felt it was unreliable
20:03 bodepd ah. nm. I found it, it wasns't related to the issue that I'm having...
20:03 openstackgerrit Emilien Macchi proposed stackforge/puppet-keystone: service_identity: add user/role ordering  https://review.openstack.org/139724
20:03 EmilienM mgagne: the one we discussed the other day with swift use-case ^
20:06 mfisch EmilienM: for which os is openstackclient not packaged?
20:06 EmilienM mfisch: all AFIK
20:06 EmilienM the version we want
20:06 ericpeterson joined #puppet-openstack
20:06 EmilienM we need 1.0.0
20:07 mfisch I can solve that for you in Ubuntu, but not RHEL & derivatives
20:07 crinkle rdo has 0.4.0, ubuntu has 0.3.0 and plans to have something in uca for kilo
20:07 crinkle debian i think is on 0.4.0
20:08 EmilienM mfisch: awesome!!
20:08 mfisch I pushed 0.4.x to vervet
20:08 mfisch but I didnt do a backport
20:08 EmilienM mfisch: richm1 already submit the feature for Red Hat
20:08 mfisch the UCA guys said that they're going to pick it up for Juno too
20:08 mfisch not sure if it happened yet
20:08 crinkle oh that's not what they told me on ubuntu-server@
20:09 crinkle that would be awesome
20:09 EmilienM crinkle: we change our plans ! lol (kidding)
20:09 crinkle :P
20:09 mfisch James Page told me that a couple weeks ago
20:09 mfisch who did you talk to?
20:09 crinkle mfisch: Corey Bryant
20:10 mfisch let me follow back up with James page, I don't know Corey
20:10 mfisch but I do have universe rights ;)
20:10 crinkle https://lists.ubuntu.com/archives/ubu​ntu-server/2014-December/007003.html
20:10 crinkle mfisch: :D
20:10 ericpeterson joined #puppet-openstack
20:10 mfisch crinkle: you know he may have said Kilo
20:12 mfisch it was kilo, sorry for the confusion crinkle
20:12 crinkle aw
20:15 vinsh joined #puppet-openstack
20:17 bodepd I've just run into an issue with the openstack providers, and I was wondering if other people have seen it, have opinions, work-arounds etc
20:17 bodepd it seems that in Puppet, if prefetching fails, the resource proceed anyways
20:18 bodepd in my case, there is an intermittent network issue with the api services
20:18 bodepd so I can't populate the expected data for neutron_net ahead of time
20:18 bodepd this results in multiple instances of the type being created
20:19 bodepd have people run into this? is there an abvious work-around
20:19 bodepd it seems like it would apply to all providers...
20:23 vinsh joined #puppet-openstack
20:24 vinsh joined #puppet-openstack
20:33 berendt joined #puppet-openstack
20:33 prad joined #puppet-openstack
20:46 mdorman bodepd:  i’ve seen that issue, too, but for us it’s keystone services/endpoints.  so we end up with dupes of those
20:46 mdorman don’t have a good solution though
21:00 markvoelker joined #puppet-openstack
21:15 mdorman has anyone been following the cells v2 stuff at all?
21:19 mdorman curious which will be less painful… going to cells now under juno, then having to migrate to v2 on kilo…. or just waiting until kilo to do cells
21:21 bodepd mdorman: I can't really think of a solution that doesn't involve rewriting all of the providers...
21:22 mdorman (last two messages were on wrong channel, sorry)
21:22 mdorman bodepd:  yeah me, either.  i haven’t spent much time on it, but there’s never been an obvious easy solution
21:23 dmsimard dachary: ping ?
21:23 dachary pong
21:24 dmsimard dachary: I looked at CI where I had left off
21:24 dmsimard I was really boggled.. put the CI tests in a RAM tmpfs and they weren't running that fast :(
21:24 dachary hum, interesting
21:24 dmsimard bumping the specs of the VMs and running them in a tmpfs wasn't doing anything crazy
21:24 dachary dmsimard: did you figure out why ?
21:25 dmsimard dachary: no, that's kind of where I got sidetracked elsewhere
21:25 dachary it's not a big deal if they run slow as long as they run reliably, right ? ;-)
21:26 dmsimard Yeah
21:26 dmsimard Just a shame though
21:26 dmsimard Running these tests shouldn't take 30+ minutes
21:27 dmsimard I'll check real quick while I have some headroom, if I can't figure it out i'll resume gerritexec anyway
21:27 dachary yeah
21:27 dachary cool
21:27 dachary off_rhoden: will be gratefull ;-)
21:27 dmsimard dachary: Oh yeah, didn't I ask you about setting up new images in the rspec-system fork too ?
21:28 dachary I don't remember that you did
21:29 ericpete_ joined #puppet-openstack
21:31 vinsh_ joined #puppet-openstack
21:33 dmsimard dachary: Adding new prefabs looks fairly simple, looking at what's in https://gitorious.org/rspec-system/rspe​c-system/source/f9faf4dadb6ddd3a9adf804​2d00800324ee65277:resources/prefabs.yml
21:33 dmsimard I'd like to test against Ubuntu trusty and we can't right now.
21:33 dmsimard Testing against updated precise and centos images wouldn't hurt either
21:33 dachary I never learnt and by the time I decided to rspec-system was deprectated. That somehow undermined my motivation ;-)
21:33 dmsimard https://vagrantcloud.com/puppetlabs (https://vagrantcloud.com/puppetlabs/b​oxes/ubuntu-12.04-64-puppet/versions and https://vagrantcloud.com/puppetla​bs/boxes/ubuntu-14.04-64-puppet)
21:35 dmsimard Are you okay with me forking it in github and giving it a jab at it myself ?
21:44 richm1 crinkle: How do you create the vcr fixture files?
21:58 crinkle richm1: you have to run it against a real system so it records it (like a vcr tape)
21:58 crinkle richm1: but mark was saying that we could just stub it the way I was doing with openstack()
21:58 crinkle so I will try to simplify that down
22:03 richm1 crinkle: ok
22:12 rcallawa_ joined #puppet-openstack
22:15 bodepd fyi: the issue is PUP-3656 . This potentially makes all of the exising providers unreliable.
22:16 bodepd crinkle: is this something you are handling with the aviator providers?
22:16 crinkle bodepd: I hadn't run into it yet
22:17 crinkle we can try to find a workaround
22:17 crinkle it seems like we could just override prefetch?
22:19 bodepd it looks like it has to be solved in each provider
22:19 bodepd the best fix would be in in Puppet ;)
22:19 bodepd I was thinking the best fix would be to just track the existance of a class instance variable
22:20 bodepd that is set at the end of prefetch, and have the instances recall prefetch or something
22:20 bodepd I'm not sure that would work. the best fix might be to have everything implement its own caching instead of using prefetch
22:20 bodepd (which is kind of funny b/c that's how everything was originally written)
22:21 crinkle okay so I think the way I wrote it they don't end up using prefetch
22:21 crinkle they use self.instances
22:21 crinkle i'm not entirely sure what the difference is
22:22 bodepd prefetch is called during catalog execution
22:22 bodepd self.instances is called from puppet resource
22:22 bodepd in general self.prefetch calls self.instances to reuse code
22:22 bodepd (or at least that's the pervassive pattern in providers)
22:22 crinkle I have an instance method 'instances' that does that work during catalog execution
22:23 crinkle https://review.openstack.org/#/c/134844/7/li​b/puppet/provider/keystone_tenant/openstack.rb,cm
22:23 bodepd can you send me a link? I was trying to find the aviator code in puppet-keystone
22:23 bodepd ah. you read my mind :)
22:24 bodepd yep, that totally works around the issue by not using puppet's built in caching mechanism ;)
22:25 bodepd that's probably pretty close to how all of the old providers would need to be refactored
22:25 bodepd basically, not to rely on prefetch, but instead to build an on-the-fly cache
22:26 bodepd I wonder if that would resolve the issue Mirantis was having where they totally wanted to kill off caching...
22:26 crinkle haha my ignorance solves problems :)
22:27 bodepd what was the openstackclient vs. avaiator decision by the way?
22:27 bodepd I might be motivated to just implement the neutorn types using whatever new provider I need
22:28 bodepd as opposed to fixing the old ones
22:28 bodepd (for some reason, I'm only seeing this in neutron, probably some build dependency)
22:28 crinkle so the decision today was to go to aviator because we don't have uca/rdo packages for openstackclient and we would have to wait
22:28 crinkle but aviator doesn't support neutron yet
22:28 bodepd ah. :(
22:28 crinkle so
22:28 bodepd I was kind of hoping it would be consolidated on openstackclient
22:28 crinkle we'll have to add that support
22:28 crinkle yeah me too :/
22:29 crinkle but i guess the group wants this to get done sooner rather than later
22:29 bodepd not to be rude. I know you've spent lots of time on the aviator stuff :(
22:29 crinkle no i spent a bunch of time on both
22:29 crinkle and it's actually not that much work to switch between them
22:29 bodepd we're not afraid to build our own packages, but I guess that doens't help the general community
22:29 crinkle the actual api calls are a very minimal difference
22:30 bodepd ah, it's all potentially just a single backend that could be configurable?
22:30 richm1 is aviator in the distro repos?
22:31 crinkle aviator is a gem that aimon has copied out into a puppet module
22:31 crinkle so we don't have to rely on the distros
22:32 crinkle bodepd: not exactly configurable, but it's only a few lines that are really different
22:32 crinkle from the providers perspectives it's nearly the same
22:32 bodepd cool
22:33 bodepd we really rarely have to patch any of the modules. Just some temptest patches coming soon
22:41 mmaglana joined #puppet-openstack
22:46 badiane_ka joined #puppet-openstack
22:51 openstackgerrit David Moreau Simard proposed stackforge/puppet-ceph: Add new CI VM specifications in an attempt to speed up tests  https://review.openstack.org/139747
23:05 mschmitt joined #puppet-openstack
23:18 richm1 looks like I'll need to get aviator into rdo . . . https://bugzilla.redhat.co​m/show_bug.cgi?id=1171352
23:22 crinkle o.0
23:38 richm1 hmm - aviator has a few dependencies of its own
23:38 aimon no it has 0 dependencies
23:38 richm1 are all of faraday, multipart-post, terminal-table, thor required at run time, or only for testing?
23:39 aimon only for testing..
23:39 richm1 that is, I did gem install aviator, and it pulled in these other gems
23:39 richm1 ok
23:39 richm1 whew
23:39 aimon Mark Maglana, the creator of aviator spent a lot of time to make sure it had no external dependencies so we could make a puppet module out of it
23:40 aimon .. make a module with no deps that is
23:42 richm1 ok

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