Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-06-05

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

All times shown according to UTC.

Time Nick Message
00:01 dvorak bodepd: the problem is the requirement here: http://rubygems.org/gems/puppetlabs_spec_helper
00:01 dvorak the released gem just requires >= 2.9.0
00:02 bodepd yeah, I sent you the link of the source code where that is specified
00:02 dvorak so I think we need to put a ~> 2.10.0 in at least the stable Gemfiles
00:02 bodepd didn't I? at least I meant to.
00:02 dvorak you did, but that's on master
00:02 dvorak not the released version
00:02 djm68 joined #puppet-openstack
00:03 dvorak so puppetlabs needs to do a release that fixes this, or we need to specify the rspec version
00:03 bodepd oh, it was the second link, not the first one.
00:03 dvorak or we have to fix all the stable tests
00:03 bodepd that line needs to be fixed: https://github.com/puppetlabs/puppetlab​s_spec_helper/blob/master/Rakefile#L32
00:03 dvorak oh, I see
00:03 dvorak yes, I'm with you
00:03 bodepd yeah, it's stupid how they do it in the Rakefile. I ran into an issue with that
00:04 bodepd b/c you can't install it from source with bundler
00:04 dvorak so they either need to release a bugfix version of the gem, or we need to change at least all the stable GEmfiles
00:04 bodepd I think it's way better to just patch it upstream
00:04 bodepd and beg someone to cut a release
00:04 bodepd I should know the people who manage the gem, so I can reach out
00:04 bodepd and ask them to release
00:05 bodepd dvorak: do you mind writing the patch?
00:05 dvorak for the gem?
00:05 bodepd well, you need to update that Rakefile
00:05 dvorak sure, that's easy, the harder part is to get someone to cut a release
00:06 bodepd I can ask
00:06 * blkperl has been asking for releases cuts for a while now
00:06 blkperl but I think they finally decided the Module team owns spec_helper
00:06 blkperl so once Hunner or ashp gets around to it, we can get a new version :)
00:07 bodepd well, there still needs to be a patch
00:08 dvorak well, it's a 1 line change and it's not clear that it should go against mater
00:08 dvorak master
00:08 bodepd it shoudl
00:08 bodepd dvorak: wait, why not?
00:08 aimon joined #puppet-openstack
00:08 dvorak well, they may not be prepared to release master, it has changes since the last release.
00:08 dvorak significant changes
00:08 bodepd dvorak: that seems like the shortest path
00:09 dvorak like 68 vs 111 commits
00:09 dvorak if I were them I'd create a 0.4.x branch and from the 0.4.1 tag and just fix this one issue
00:09 bodepd ah. gotcha.
00:09 bodepd dvorak: maybe give it a day and see if there is action?
00:09 dvorak sure, I'll ping hunner and/or hoge if I see them
00:10 bodepd dvorak: befre you start thinking about the massive amounts of updates we have to make otherwise
00:10 dvorak I don't think changing the Gemfiles is really that much work, just tedious.  I'll leave it up to the puppet guys which way to go
00:12 dvorak I only brought it up because I have a patch against stable/icehouse that's failing
00:12 dvorak wow, really? https://github.com/stackforge/puppet-neutron/c​ommit/40f98eee3f53b6ecb878d27ffcecabd776ddb713
00:12 dvorak be_falsey?
00:12 dvorak that's a great breaking api change
00:13 djm68 joined #puppet-openstack
00:15 ryanycoleman joined #puppet-openstack
00:18 bodepd dvorak: it is a major version
00:18 dvorak yeah, but that's arguably not even an improvement.
00:37 gildub joined #puppet-openstack
00:40 gildub hodepodge, hi - yeah, only tested on RHEL, just saw your comment (94504), thanks
00:43 bodepd how are people installing Puppet on redhet/centos? What ruby version are people using?
00:47 rmoe joined #puppet-openstack
00:50 djm68 joined #puppet-openstack
00:57 blkperl dvorak: I know right, such a silly breaking change
00:58 starmer joined #puppet-openstack
00:59 gildub bodepd, on RHEL7, RHOS5 Beta is using Puppet 3.4.3
00:59 gildub bodepd, from RHOS5 repos
01:01 gildub bodepd, ruby is 2.0.0.353 from RHEL7 standard repo
01:02 gildub bodepd, CentOS to follow through once GA
01:03 aldavud joined #puppet-openstack
01:08 gildub bodepd, but RHSCL (software collection) will probably be consolidated to contain a puppet version for RDO too
01:08 gildub bodepd, although I suppose one can always use puppetlabs repos. Do you have a specific concern?
01:12 gildub joined #puppet-openstack
01:13 * gildub will be back in 20 minutes
01:20 rmoe joined #puppet-openstack
01:24 dvorak bodepd: we have the same problem with mocha also, it's breaking tests due to puppet_spec_helper deps not being right
01:26 dvorak hrm, or maybe I'm doing it wrong
01:27 dvorak probably the latter
01:31 xarses joined #puppet-openstack
01:38 kitp joined #puppet-openstack
01:40 dvorak nope, I was wrong, mocha is broken too
01:44 djm68 joined #puppet-openstack
01:44 gildub joined #puppet-openstack
01:57 xingchao joined #puppet-openstack
02:02 kitp joined #puppet-openstack
02:12 djm68 joined #puppet-openstack
02:12 aldavud joined #puppet-openstack
02:51 badiane_ka joined #puppet-openstack
02:55 djm68 joined #puppet-openstack
03:11 michchap gildub: on the puppet versions bodepd was talking about, I am currently packaging my own ruby 2.0.0 and installing puppet from gem, as I get segfaults during catalog compilation on ruby 187. bodepd and I are investigating options with respect to how we bootstrap nodes to prepare them for puppet runs as we want the environment to be make sense but also diverge as little as possible from what is already packaged.
03:12 michchap gildub: I am leaning towards maintaining my own omnibus packages, as there is already a repo for doing that here: https://github.com/andytinycat/puppet-omnibus
03:13 michchap gildub: having 187 for rhel 6, 193 for ubuntu and 200 for rhel 7 is something I'd like to avoid.
03:13 gildub michchap, rhel7 has ruby 2.0.0.353, is this one faulty?
03:14 michchap gildub: I'm not suing rhel7, I'm on Centos6
03:14 gildub michchap, oh, ok
03:14 gildub michchap, but they're on par
03:14 michchap gildub: It's probably identical to what I'm packaging, in fact.
03:15 gildub michchap, but 187 is deprecated, and 2.0 should be the way to go, isn't?
03:15 * gildub is ready about omnibus
03:16 michchap gildub: I don't know enough about what I should be standardising on - I agree with you, which is why I packaged 200 for C6, but I think if I do that I'll also make 200 for ubuntu so I can keep them consistent
03:16 * gildub ooops, meant reading
03:17 gildub michchap, what do you mean by "I packaged for..." ?
03:17 michchap gildub: I grabbed a spec and built an RPM, which I deploy to all nodes when bootstrapping.
03:18 gildub michchap, ah ok, so that's for you own systems/projects?
03:19 michchap gildub: Yep
03:20 michchap gildub: we have a distro built off puppet_openstack_builder based on Centos6
03:20 gildub michchap, I got confused for a second. I think making 2.0.0 for both sounds good. Omnibus is interesting
03:22 gildub michchap, but ultimately let the distros handling those would be better, latest of CentOS/RHEL, Debian/Unbuntu should come with ruby 2.0.x
03:22 mgagne I like how upgrading ceilometer on my compute nodes results in a database migration being executed somehow...
03:23 michchap gildub: but we're not always targetting the latest distro - we have to target centos 6 until customers are ready to accept the risk on a new OS.
03:23 djm68 joined #puppet-openstack
03:25 gildub michchap, well each distro comes with it's puppet version and ruby deps. Which is fine, but the issue I see is the puppet client not aligned with the puppetmaster.
03:25 mgagne Found a similar bug: https://bugs.launchpad.net/o​penstack-cisco/+bug/1321507
03:26 michchap gildub: I get segfaults when compiling large catalogs on 187, which is why I am packaging ruby 200 in the first place. I figure if I'm going to do that I may as well move all of the distros to the same version via omnibus to reduce the differences between the environments
03:28 michchap gildub: So I can't use 187 for c6, and I also don't really want to have different versions of puppet to support across the OS'es - it's easier to just pick one.
03:28 michchap gildub: but this is my problem, probably not yours :)
03:28 gildub michchap, have you tried ruby 1.9 on C6?
03:28 michchap gildub: I haven't, I went straight to 200. I assume it will work.
03:29 michchap mgagne: that's strange
03:29 gildub gildub, well I hope it does :)
03:30 gildub michchap, ^
03:34 mgagne michchap: I'm not using mysql nor sqlite...
03:35 mgagne michchap: people asked how to upgrade openstack with puppet
03:35 mgagne michchap: so far I'm trying this technic: update repository, puppet apply and pray
03:35 michchap mgagne: puppet apply + hope
03:35 mgagne michchap: hahahaha
03:35 michchap mgagne: hahaha we think alike
03:36 michchap maybe a collector to make all package ensure => latest :)
03:36 mgagne michchap: you with your crazy ideas
03:37 mgagne michchap: I suggest you try it on your compute nodes with the iscsi package =)
03:37 michchap mgagne: I will pass. I prefer the 'hope' method.
03:37 mgagne I'm working on it =)
03:40 mgagne michchap: alright, I stopped mysql on my db server and ceilometer still complains about tables existing
03:40 mgagne michchap: I suspect sqlite being used instead although I use mongodb
03:41 djm68 joined #puppet-openstack
03:41 mgagne michchap: in ceilometer-common.postinst
03:41 mgagne if grep -qE "^(sql_)?connection.*sqlite.*" /etc/ceilometer/ceilometer.conf
03:41 mgagne then
03:41 mgagne su -s /bin/sh -c '/usr/bin/ceilometer-dbsync' ceilometer
03:41 mgagne fi
03:41 kitp joined #puppet-openstack
03:42 michchap that's crude
03:42 mgagne michchap: that's silly
03:43 michchap mgagne: that's another way of putting it
03:43 mgagne michchap: now I need to use your puppet fu you used in ::compat to ensure => absent the config on the compute node
03:43 mgagne michchap: awesome: http://paste.openstack.org/show/82872/
03:43 michchap mgagne: should add purging
03:43 michchap mgagne: I added it to glance, but maybe not to other things.
03:44 mgagne michchap: someone complained about comments being stripped when enabled
03:44 mgagne michchap: someone being a coworker
03:44 michchap mgagne: when purging is enabled?
03:45 mgagne michchap: yep
03:45 michchap mgagne: Now that the .conf files aren't in github it's more trouble to find the docstrings for config options.
03:46 mgagne michchap: yep and removing it was not the best idea, especially for operators relying on it
03:46 michchap mgagne: but other than that...who cares if there are comments in your prod conf files.
03:46 mgagne michchap: the cowboy next to me cares
03:47 michchap mgagne: if the file is managed by puppet you shouldn't have to log in and mess with it anyway
03:47 mgagne michchap: =)
03:48 michchap mgagne: this is a documentation issue, not a purging conf issue
03:48 michchap mgagne: (I think)
03:49 mgagne michchap: one of the goal of the ini provider is to be as non-intrusive as possible =)
03:49 mgagne michchap: until it truncates your file and writes its own content
03:49 michchap mgagne: unless you want it to intrude :D
03:49 mgagne michchap: afaik, I didn't use ensure => intrusive :D
03:50 mgagne michchap: alright, ceilometer database migration with sqlite is bad
03:50 mgagne michchap: I can't migrate from nothing to HEAD without error
03:51 michchap mgagne: if it's a post script error, that's something to file against ubuntu?
03:51 mgagne michchap: +/- running it manually results in the same error
03:51 mgagne michchap: afaik, db sync should be idempotent
03:52 michchap mgagne: That's a fair statement about db sync
03:54 mgagne michchap: now I do I make puppet purge the config on the compute node.
03:54 mgagne michchap: this supposes the config exists right? https://github.com/michaeltchapman/puppet-open​stacklib/blob/test/manifests/compat.pp#L20-L23
03:54 mgagne michchap: if it's not defined, it won't be changed or defined right?
03:55 michchap mgagne: yep collectors do nothing if they don't match anything
04:03 mgagne michchap: help, I'm in a catch-22
04:04 mgagne michchap: ceilometer_config { 'database/connection': ensure => absent, before => Package['ceilometer-common'] }
04:04 mgagne Found 1 dependency cycle: (Ceilometer_config[database/connection] => Package[ceilometer-common] => Ceilometer_config[database/connection])
04:05 mgagne michchap: and not removing database/connection prior to ceilometer-common install results in dbsync being triggered and failing the puppet run
04:18 michchap mgagne: hmmmmm
04:20 michchap mgagne: so you want it to be unset, then handle the package, then set?
04:21 mgagne michchap: there is no raise for it to be set on a compute node
04:21 mgagne michchap: default config just happens to have a database/connection set
04:22 michchap mgagne: this is similar to the nova case? Where it's set on compute nodes just because of how we set out the classes?
04:22 mgagne michchap: I don't manage database/connection on my compute node, it's only the default value which happens to be sqlite.
04:23 mgagne michchap: with that migration issue in mind, I sure want to make sure there is no connection set to avoid trigerring the problem
04:23 michchap mgagne: oh. I see
04:23 mgagne michchap: but now I have this circular dependency
04:23 mgagne michchap: so in fact, I'm not sure how people managed to make ceilometer work on a multi-node setup
04:24 michchap Ask EmilienM when he logs in - I'm certain they are including ceilometer.
04:24 michchap mgagne: I'm not installing it so I wouldn't know. I am trying to swap over to eNovance's stack at the moment
04:26 ryanycoleman joined #puppet-openstack
04:28 mgagne michchap: afaik, they use Debian, not Ubuntu
04:28 mgagne michchap: but ceilometer should be the same code/version on both
04:29 michchap different debs though - maybe it doesn't run db sync
04:30 michchap mgagne: the post install scripts would be part of the downstream wouldn't they?
04:30 mgagne michchap: yep, there is no postinst provided by ceilometer
04:30 mgagne michchap: but the database migration doesn't work
04:34 djm68 joined #puppet-openstack
04:36 comptona joined #puppet-openstack
04:42 kitp joined #puppet-openstack
04:46 sanjayu joined #puppet-openstack
04:46 comptona joined #puppet-openstack
04:56 aldavud joined #puppet-openstack
05:03 gildub joined #puppet-openstack
05:22 hunner joined #puppet-openstack
05:31 aldavud joined #puppet-openstack
05:41 xingchao_ joined #puppet-openstack
05:43 kitp joined #puppet-openstack
06:05 ashp_ joined #puppet-openstack
06:29 sputnik13 joined #puppet-openstack
06:41 beddari lots of great discussions lately, good to see this going on
06:44 kitp joined #puppet-openstack
06:45 beddari ini file provider, cowboys vs ninjas I guess ;-)
06:46 beddari "people asked how to upgrade openstack with puppet" ;-)
06:48 michchap beddari: I notice your name on the puppet-omnibus repo - are you using it?
06:48 beddari I'm the author these days
06:48 beddari I rewrote it a year ago or so
06:48 beddari :P
06:48 michchap I'll take that as a glowing endorsement :)
06:48 beddari yes, using it, and omnibus-ruby/fpm-cookery extensively
06:49 beddari we're thinking about omnibus-openstack naturally, not gotten to that yet though
06:49 beddari craig tracey has :)
06:50 michchap I have looked at his omnibus openstack and also at anvil. Neither seems to contain init scripts, so I'm not really sold on using them since I'd have to reinvent a lot of the packaging work that goes on from both the RH and Debian groups
06:51 beddari yes and no, in my own omnibus approach I specifically do not want init scripts included, but well, you could easily pull them from where they are at build time if you do
06:52 arnaud_orange left #puppet-openstack
06:53 beddari I presented this http://deploy.no/omnibus-to-the-future/ at cfgmgmtcamp earlier this year
06:56 michchap beddari: I'm curious what you would use in place of init scripts. I would guess that puppet doesn't like it when things don't have init scripts, but does it just work?
06:56 artem_panchenko joined #puppet-openstack
06:57 beddari I do use init scripts, but I provide them with puppet templates instead of packages, just by choice. We consider them configuration
06:57 michchap beddari: honestly I do like your approach, but I also don't quite see how it works with a CM tool.
06:57 michchap beddari: ah, so you've had to start owning them. That's a bit unfortunate.
06:57 michchap beddari: I'd prefer to be consuming sane ones, but perhaps that's not realistic.
06:57 beddari but I think unavoidable if I want to keep my promise, same package everywhere
06:58 beddari including containers ..
06:58 michchap beddari: it also gets rid of the insanity around different service names in neutron I guess.
06:58 beddari hehe yes
06:59 beddari so next we want to move even further into abstraction, have been playing with jordan sissels pleaserun
06:59 michchap I haven't seen pleaserun. In general I am a fan of jordan's work.
07:00 beddari have a look :P
07:00 michchap I have also been thinking about tying the packaging DSL to the CM DSL
07:01 michchap like...puppet manages a graph of resources...so does each individual deb. There is significant overlap
07:02 michchap rewriting apt to use puppet would be insane and slow, but perhaps a faster implementation would provide some benefits.
07:02 michchap making package calls and CM calls the same thing.
07:02 beddari yeah I've been thinking along the same lines
07:02 beddari decided it is not worth pursuing though
07:03 beddari from what I see I think the role of distros is changing
07:03 michchap it would involve rewriting a CM system in something fast, and then also building an entire distro around it. Far too much work.
07:04 beddari cfengine hehe
07:04 beddari but no, better approach is to just cut down the base os as much as possible, rely on distro mechanisms for maintaining that toolchain, control the rest
07:05 michchap something something static vs dynamic linking something something.
07:06 beddari ;-)
07:06 beddari so what is missing from having a community that drives the metapackaging world
07:06 beddari a common packaging DSL :)
07:06 beddari that's what I want to get going hehe
07:07 michchap right. I can see that.
07:07 michchap I think the SUSE guys have something somewhat agnostic since they sit between deb and rpm
07:08 beddari yeah kind of but the unity there is just on the builds, you still need to have a full toolset for each supported packaging target, no?
07:08 michchap yep. So you would define some new language that fpm can consume and spit out rpm/deb/xyz?
07:09 beddari which is what omnibus can be, and fpm-cookery is closer to be, but each has its own strengths and problems at this stage
07:10 beddari stop the world for 30 days and let me dive in to fix that please :P
07:10 michchap I wave my wand and say the words.
07:10 beddari I wrote this https://github.com/bernd/fpm-cookery/wiki/wishlist
07:10 beddari which kind of defines my dream heh
07:11 triplecheesesina joined #puppet-openstack
07:11 beddari also I've helped organize a couple of packagingsucks open spaces at devopsdays
07:12 beddari but mostly what has happened has been me talking about what exists today, not moving forward
07:13 AaronGr joined #puppet-openstack
07:14 beddari thanks for listening, always nice to grow the small group I know of that I can say actually get what I'm talking about
07:15 michchap No worries. I'm pretty interested in this. I think I agree with your approach (containerisation instead of excessive dep management?) but I need to consdier it a bit more
07:16 beddari the containerization is part of what keeps it simple and understandable, it becomes an artifact that everyone accepts
07:17 berendt joined #puppet-openstack
07:18 beddari but you can't do this in the blind, you need to have a CI toolchain to properly maintain each part of the container (security, upstream changes .. long list)
07:18 michchap Right now I feel when one of the distros makes a strange packaging decision I just have to wear it and deal with it, but I guess with the omnibus approach I'm taking control of that
07:18 beddari yes that is why we feel we've "won"
07:18 michchap yes, there is definitely a maintenance overhead, but that's the price of taking ownership of part of the stack
07:19 beddari I think at one of my employers we have five apps or so (not openstack) in this workflow, and everyone is like .. it works!
07:20 beddari but of course, the more of this that could be shared maintenance wise the better
07:20 triplecheesesina if im taking ownership of part of the stack, why am i not just making sane packages and packaging decisions?
07:20 triplecheesesina i.e. if youre gonna containerise why not just build sane packages again?
07:20 triplecheesesina i probably missed something
07:20 beddari yup :)
07:20 michchap that's what you do, but the overhead of testing and maintaining CI comes with that.
07:21 beddari packages aren't sane due to distrro policies mostly
07:21 beddari or lack of maintenance
07:21 beddari or speed of upstream projects ..
07:21 michchap some of it is just differences that exist because two people made two difference choices, like the redhat service names and the debian services names are different.
07:22 beddari but the 'containerization' we talked about here was not the popular variant, rather a 'fat package'
07:22 triplecheesesina ah right
07:22 triplecheesesina i heard not long ago, that google distributes a lot of their apps as compiled python
07:22 triplecheesesina a fat binary
07:22 beddari same reasoning, I've talked to people at Google about how they deploy
07:23 beddari .. any code really, they do that
07:23 michchap an openstack omnibus would be extremely fat though.
07:23 beddari I don't think it really relates to the outside world, but on a technical level it does
07:24 michchap is that something people care about? I think enovance are putting every package on every node. Maybe it's not even an issue anymore
07:24 beddari michchap: so yeah, craig traceys is extremely fat, but there is where my thinking differs a bit, I think the line between "core / in distro" and "owned by me" is a bit blurry
07:25 beddari ah for the size, I don't think that matters no :)
07:25 beddari but for the cost of maintenance of the container
07:25 beddari so e.g libvirt, in the pure Chef omnibus-ruby approach ALL deps are vendored
07:25 beddari I don't see that being sane
07:25 tfz joined #puppet-openstack
07:26 michchap right, but if someone wants docker they need a different omnibus
07:26 beddari see, yeah, I'm not gotten to this yet hehe
07:26 michchap so here's a question: do people make their own omnibus to satisfy their plugin requirements, or are plugin requirements in their own packages
07:27 beddari the idea is to bulid in everything
07:27 beddari artifact == deliverable, what you prod
07:28 michchap Right, so if someone wants to run *crazy Neutron plugin ABC* and that needs its own stuff, the idea is they build their own omnibus and that process is very simple?
07:28 beddari yes
07:28 michchap I agree
07:29 michchap how simple is it today?
07:29 beddari vagrant
07:30 beddari pretty simple, but because of shortcomings in the DSL model/approach of the tools sharing code doesn't really work well enough
07:30 beddari e.g what I wrote in that wishlist
07:30 beddari but the build side is easy enough yes, with vagrant
07:30 beddari at one point I got a build lab for Puppet packages that built fedora, centos, ubuntu and osx
07:31 beddari and I did some early simple stuff to try and help that over at https://github.com/norcams/fpm-refinery
07:32 beddari hopefully I'll make that usable over the summer :) and point puppet-omnibus to it
07:33 beddari someone suggested making it easy to generate a bintray account+api key for uploading your own stuff, might look at that
07:34 beddari bintray hosts yum and apt repos
07:34 michchap I like the refinery.
07:34 beddari consider it an idea so far :P it is not usable
07:34 beddari but I'll grab some code from Chef to make it
07:35 michchap I think I have a lot of reading to do
07:35 michchap thanks for the info.
07:35 beddari happy to share
07:36 EmilienM good morning
07:37 michchap EmilienM: morning
07:43 michchap EmilienM: I have a quick edeploy question: how do you manage the ssh keys and root logins to the box?
07:45 kitp joined #puppet-openstack
07:45 EmilienM michchap: we don't, infra team use puppet for now
07:45 michchap EmilienM: so a node is deployed with a role, and if puppet agent fails to run, you can't log into the box?
07:46 arnaud_orange joined #puppet-openstack
07:46 arnaud_orange morning
07:47 EmilienM michchap: i have a jenkins user created by a metadata server (cloud init) with a ssh key
07:47 michchap EmilienM: that was my other question - I have cloud init installed on these nodes but there is no metadata server anywhere afaik.
07:49 michchap EmilienM: do you have a 'pretend' nova-metadata-api somewhere
07:49 EmilienM michchap: no, we use to install the metadata server on the install role
07:50 michchap EmilienM: what is the metadata server?
07:50 EmilienM michchap: let me 2 min
07:52 mmagr joined #puppet-openstack
07:53 EmilienM michchap: so
07:55 EmilienM michchap: you need to install a metadata server where you want, we do it on the edeploy server
07:55 EmilienM put the metadata like user-data, etc in /var/whereyouwant
07:55 EmilienM and set the path in /etc/edeploy.conf
07:56 michchap EmilienM: I mean...what *is* the metadata server. Is it a wsgi script in edeploy somewhere?
07:56 michchap EmilienM: Like...how do I run it?
07:57 EmilienM michchap: there is nothing to do afik
07:58 EmilienM the wizard is client side
07:58 EmilienM with the cloud-init agent
07:58 EmilienM does that make sense?
07:59 michchap ok...so when I build the role the metadata will be injected somehow into the chroot?
07:59 EmilienM michchap: no, when you bootstrap the physical nodes, edeploy will add the metadata URL in the grub config
07:59 EmilienM it's not in the role
07:59 EmilienM it's in edeploy itself
08:00 michchap ok, so it will pull down a user_data.json or .yaml and then....it makes a fake /dev/sr0 or something?
08:01 EmilienM michchap: yes, should be like this. We only use metadata to inject a jenkins user / group and a ssh key.
08:01 michchap EmilienM: yeah...I figured - I can't login to my nodes :D
08:01 EmilienM lol
08:01 michchap so the config on the edeploy server is just to add an apache vhost sharing the metadata dir under edeploy
08:05 EmilienM michchap: no need to share the metadata dir under edeploy
08:05 EmilienM i just have /var/www/metadata with two files
08:05 EmilienM user-data and meta-data
08:05 aldavud joined #puppet-openstack
08:05 michchap EmilienM: it doesn't pull from different path depending on role and version?
08:08 EmilienM michchap: metadata in edeploy itself is for upgrade informations
08:09 openstackgerrit Zhenguo Niu proposed a change to stackforge/puppet-horizon: Fix default port of MS SQL in security group template  https://review.openstack.org/98052
08:09 michchap EmilienM: Ah I get it :)
08:11 EmilienM michchap: i know it's a bit tricky to get it, but it works like this
08:45 kitp joined #puppet-openstack
08:53 lsamoila hello
08:54 lsamoila I am trying to deploy openstack icehouse using puppet on ubuntu 12.04 and on the controller node in doesn't install the glance-registry service. Any idea what could be the problem?
08:55 EmilienM lsamoila: without manifests or logs, how could we help ? :-)
08:55 lsamoila just a moment :D
08:59 lsamoila this is the output from the install process: http://pastebin.com/2hv8AJi9
09:00 lsamoila this is part of the hiera file, the rest is the same as in the modules/openstack/examples/common.yaml  http://pastebin.com/PFbctjFX
09:03 aldavud joined #puppet-openstack
09:04 lsamoila and here is the glance log http://pastebin.com/0UbESpPh
09:05 lsamoila the site.pp manifest is again, the same as in the examples, but with the hostnames of my nodes and without swift storage
09:37 derekh_ joined #puppet-openstack
09:46 kitp joined #puppet-openstack
09:53 mkoderer joined #puppet-openstack
09:53 lsamoila nevermind, i think i fixed it. I added in the controller.pp: class { '::openstack::profile::glance::api': }
09:59 starmer joined #puppet-openstack
10:22 aldavud joined #puppet-openstack
10:31 ryanycoleman joined #puppet-openstack
10:47 kitp joined #puppet-openstack
10:50 gildub joined #puppet-openstack
11:04 dprince joined #puppet-openstack
11:26 mmaglana joined #puppet-openstack
11:30 starmer joined #puppet-openstack
11:32 ryanycoleman joined #puppet-openstack
11:44 starmer joined #puppet-openstack
11:48 kitp joined #puppet-openstack
11:58 morazi joined #puppet-openstack
12:01 dprince joined #puppet-openstack
12:18 openstackgerrit Lukas Bezdicka proposed a change to stackforge/puppet-swift: Add swift::proxy::crossdomain class  https://review.openstack.org/98091
12:33 ryanycoleman joined #puppet-openstack
12:38 abramley joined #puppet-openstack
12:48 kitp joined #puppet-openstack
12:56 dgollub joined #puppet-openstack
12:59 djm68 joined #puppet-openstack
13:06 fritzroy15 joined #puppet-openstack
13:07 fritzroy15 hello, I'm trying to do a full deploy with puppet-openstack module 3.0 (havana). I see that in the example controller node, the neutron network is disabled, and trying to enable it I'm getting a bunch of values undeclared errors (such as neutron_user_password etc) is there an example document for setting these? ty
13:11 fritzroy15 Hunner? are you around?
13:18 mjblack joined #puppet-openstack
13:19 mjblack joined #puppet-openstack
13:19 meylor joined #puppet-openstack
13:24 higgins joined #puppet-openstack
13:33 djm68 joined #puppet-openstack
13:34 ryanycoleman joined #puppet-openstack
13:42 dgollub joined #puppet-openstack
13:44 starmer_ joined #puppet-openstack
13:45 kitp joined #puppet-openstack
13:50 prad joined #puppet-openstack
13:53 meylor joined #puppet-openstack
14:01 fritzroy15 anyone?
14:19 mmagr joined #puppet-openstack
14:21 Hunner fritzroy15: sup?
14:22 Hunner hogepodge: ^
14:22 fritzroy15 hey Hunner
14:22 fritzroy15 still trying to get the puppet-openstack deploy to work :D
14:22 fritzroy15 here's where I'm stuck:
14:22 fritzroy15 hello, I'm trying to do a full deploy with puppet-openstack module 3.0 (havana). I see that in the example controller node, the neutron network is disabled, and trying to enable it I'm getting a bunch of values undeclared errors (such as neutron_user_password etc) is there an example document for setting these? ty
14:23 Hunner Are you using https://github.com/puppetlabs/puppetlabs-ope​nstack/blob/master/manifests/role/network.pp ?
14:24 Hunner well, the stable/havana branch version
14:24 Hunner (and thus using https://github.com/puppetlabs/puppe​tlabs-openstack/blob/stable/havana/​manifests/setup/sharednetwork.pp)
14:25 fritzroy15 Hunner hm no, i just picked up the example control node but also tried with all.pp
14:25 fritzroy15 i'll have a look
14:25 Hunner It's also in https://github.com/puppetlabs/puppetlabs-opensta​ck/blob/stable/havana/manifests/role/allinone.pp . So I'm not sure what you mean by "disabled"
14:25 dvorak Hunner: you know what's going on with puppetlabs_rspec_helper and rspec 3.0.0?
14:26 Hunner dvorak: All sorts of bad things? :(
14:26 Hunner dvorak: I actually haven't dug into it yet. Is it causing gates to fail?
14:26 dvorak yes
14:26 dvorak even on stable branches
14:27 dvorak there seem to be 3 options
14:27 dvorak 1) cut a new gem with fixed deps, 2) change the Gemfile in all the projects to use old rspec and old mocha or 3) upgrade all the tests to use new rspec and new mocha
14:27 dvorak 1) is probably the best options, especially since this affects stable branches
14:28 Hunner Yeah... I got permission to own psh on Monday, so I'll see if I can prioritize releasing it today
14:28 dvorak but master on puppetlabs_rspec_helper is diverged significantly from the last release, so someone will probably have to create a 0.4.x branch to do that release off of
14:29 dvorak the Gemfile for the project has the correct dependencies, but the gemspec does not
14:32 fritzroy15 Hunner: I'm getting  Error 400 on SERVER: bridge_interface must be set when configuring neutron
14:32 fritzroy15 i've copied the manifest from the sharednetwork.pp in my controller node branch
14:35 arnaud_orange joined #puppet-openstack
14:36 Hunner Huh. What version of puppet-neutron is installed?
14:37 fritzroy15 Hunner: 3.3.0
14:40 fritzroy15 have to split
14:41 fritzroy15 i'll give it some more wacks when I get home
14:41 fritzroy15 thanks you!
14:45 rmoe joined #puppet-openstack
14:52 mmagr joined #puppet-openstack
14:52 ryanycoleman joined #puppet-openstack
15:09 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-keystone: fix admin_token with secret parameter  https://review.openstack.org/98138
15:09 EmilienM mgagne: sbadia ^
15:09 EmilienM backport quite important imo
15:18 mdorman joined #puppet-openstack
15:23 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-ceilometer: Fixes agent_notification_service_name  https://review.openstack.org/98150
15:23 xarses joined #puppet-openstack
15:24 aldavud_ joined #puppet-openstack
15:29 openstackgerrit A change was merged to stackforge/puppet-neutron: Added '' as a valid value for gateway_ip  https://review.openstack.org/94120
15:40 openstackgerrit A change was merged to stackforge/puppet-swift: Avoid duplication error when ringserver and storage are on the same node  https://review.openstack.org/97763
15:42 openstackgerrit A change was merged to stackforge/puppet-ceilometer: Fixes agent_notification_service_name  https://review.openstack.org/95519
15:45 rmoe joined #puppet-openstack
15:54 djm68 joined #puppet-openstack
15:57 xarses joined #puppet-openstack
15:59 badiane_ka joined #puppet-openstack
16:05 starmer joined #puppet-openstack
16:06 anode joined #puppet-openstack
16:06 openstackgerrit A change was merged to stackforge/puppet-cinder: Storage availability zone support  https://review.openstack.org/92993
16:10 sputnik13 joined #puppet-openstack
16:11 mmaglana joined #puppet-openstack
16:12 openstackgerrit A change was merged to stackforge/puppet-keystone: enabling tokens flushing using cron  https://review.openstack.org/97865
16:13 openstackgerrit A change was merged to stackforge/puppet-nova: Fix typo in README.md  https://review.openstack.org/97607
16:13 openstackgerrit A change was merged to stackforge/puppet-swift: Add swift::proxy::crossdomain class  https://review.openstack.org/97259
16:14 djm68 joined #puppet-openstack
16:15 openstackgerrit A change was merged to stackforge/puppet-swift: Add swift::proxy::crossdomain class  https://review.openstack.org/98091
16:16 openstackgerrit A change was merged to stackforge/puppet-horizon: Fix default port of MS SQL in security group template  https://review.openstack.org/98052
16:18 sputnik13 joined #puppet-openstack
16:20 starmer joined #puppet-openstack
16:21 openstackgerrit joined #puppet-openstack
16:22 aimon__ joined #puppet-openstack
16:23 openstackgerrit A change was merged to stackforge/puppet-keystone: fix admin_token with secret parameter  https://review.openstack.org/98138
16:24 arnaud_orange left #puppet-openstack
16:34 mmaglana joined #puppet-openstack
16:34 djm68 joined #puppet-openstack
16:35 starmer joined #puppet-openstack
16:43 crobertsrh joined #puppet-openstack
16:43 sputnik13 joined #puppet-openstack
16:43 crobertsrh Any chance that anyone here knows about the https://github.com/redhat-open​stack/openstack-puppet-modules repo?
16:43 crobertsrh I added a new submodule to openstack-puppet-modules for sahara via a pull request to change .gitmodules.  That change has been merged, but when I do a git clone --recursive of that repo, I still do not see the sahara submodule.  What else is required to make that happen?  Thanks.
16:48 djm68 joined #puppet-openstack
16:50 starmer joined #puppet-openstack
16:54 dprince joined #puppet-openstack
16:59 djm68 joined #puppet-openstack
17:01 hogepodge fritzroy15 sorry I missed the conversation. Feel free to ping me when you get back.
17:02 anode joined #puppet-openstack
17:03 dvorak hogepodge: you see the stuff in scrollback about rspec 3?
17:03 hogepodge dvorak no, I’ll go look
17:03 dvorak hunner and I talked about it this morning and bodepd and I talked about it late last night
17:04 dvorak it's breaking gate tests at least for neutron on stable/icehouse
17:04 dvorak probably others also
17:04 hogepodge rspec 3 and rspec puppet are a whole bag a pain right now.
17:05 hogepodge There may be an easier way to pin. Let me look something up
17:05 dvorak seems like if there isn't going to be a new release of puppetlabs_rspec_helper soonish then we should update all the Gemfiles to work around the issue
17:06 starmer joined #puppet-openstack
17:06 hogepodge yes
17:06 dvorak for master, it might be viable to just move to rspec 3, there is are apparently decent automatic upgrade tools
17:06 hogepodge I have to work up some new releases, I’ll do that today and pin rspec 3 at the same time.
17:06 dvorak seems like a bad idea for stable though
17:09 hogepodge i mean pin < rspec 3
17:10 dvorak hogepodge: there is also a potential issue with mocha I've seen on other modules.  I'd recommend stealing the dep lines from the puppetlabs_rspec_helper Gemfile.  those appear to be correct
17:14 hogepodge dvorak be_falsey, utterly stupid. Pedantic nonsense without a deprecation period.
17:15 Hunner hogepodge: releasing a fix in puppetlabs_spec_helper could be an option... I need to know how to get failures to know when it's fixed though :P
17:15 hogepodge hunner, my opinion is if stable requires an older version right now, that dependency should be expressed.
17:19 dvorak hogepodge: there is the risk that if we move master to rspec 3 and puppetlabs_rspec_helper is released later that depends on rspec ~> 2.10.0, then it'll break all those tests again
17:20 starmer joined #puppet-openstack
17:21 openstackgerrit David Gurtner proposed a change to stackforge/puppet-ceph: ceph::osd support for scenario_node_terminus  https://review.openstack.org/98182
17:32 djm68 joined #puppet-openstack
17:34 aldavud dachary: ping
17:36 starmer joined #puppet-openstack
17:37 comptona joined #puppet-openstack
17:45 hogepodge we need to go back to havana too, right dvorak?
17:46 dvorak probably.  that's why it'd be better if this was changed upstream
17:50 starmer joined #puppet-openstack
17:55 sputnik13 joined #puppet-openstack
17:55 g_306396236 joined #puppet-openstack
17:56 marun joined #puppet-openstack
17:58 anode joined #puppet-openstack
17:59 ryanycoleman joined #puppet-openstack
18:00 ryanycoleman joined #puppet-openstack
18:02 ryanycoleman joined #puppet-openstack
18:03 djm68 joined #puppet-openstack
18:04 djm68 joined #puppet-openstack
18:04 g_306396236 joined #puppet-openstack
18:05 g_306396236 left #puppet-openstack
18:05 jehb joined #puppet-openstack
18:05 starmer joined #puppet-openstack
18:08 hogepodge @here pin master to rspec < 3?
18:08 hogepodge then backport?
18:10 hogepodge dvorak problem is we’ve already started coding against the 3 api
18:10 hogepodge beu_truthy, etc
18:10 Hunner What would be the impact of not backporting?
18:10 Hunner Err, not pinning
18:11 Hunner How much would have to be changed
18:11 hogepodge icehouse and havana definitely need to be pinned. We shouldn’t be making api changes to them.
18:12 hogepodge We’ve sent patches through to fix master to kind of work with 3, and if we pin we have to undo those.
18:16 mgagne hogepodge: are RSpec 3.0.0 and RSpec 2.x.x compatible? Can you write tests compatible with both versions?
18:17 hogepodge mgagne there are api breaking changes
18:17 hogepodge Apparently there’s a tool to bring code that uses 2 up to 3
18:18 hogepodge The one we were dealing with immediately was the change of be_truth to be_truthy
18:18 mgagne hogepodge: I saw some options you can enable to support legacy stuff. Are all changes covered by those options or are some just not backward compatible?
18:18 hogepodge I don’t know
18:20 starmer joined #puppet-openstack
18:21 djm68 joined #puppet-openstack
18:26 dvorak this is what I found on migrating: http://myronmars.to/n/dev-blog/2013/11/rs​pec-2-99-and-3-0-betas-have-been-released
18:27 jehb joined #puppet-openstack
18:29 ryanycoleman joined #puppet-openstack
18:34 aldavud_ joined #puppet-openstack
18:37 bodepd who is Gurtner is here?
18:41 mgagne EmilienM: ping
18:42 dmsimard bodepd: it's aldavud
18:44 hogepodge dvorak I’m not goind to back port rspec 3 to the stable projects
18:44 bodepd dmsimard: thanks.
18:44 bodepd aldavud: I'd love to catch up and chat about your thoughts about data model
18:44 dvorak hogepodge: wasn't suggesting you would :)
18:45 hogepodge We’ve already moved two projects forward to be rspec 3is
18:45 hogepodge 3ish
18:46 bodepd hogepodge: my main concern with the 3 move is rspec-puppet. I got lots of warnings
18:46 hogepodge yes it does.
18:46 mgagne hogepodge: is rspec-puppet and puppetlabs_spec_helper fully supporting RSpec 3.0.0 yet?
18:46 hogepodge no, it’s not.
18:46 dvorak hogepodge: my concern is actually the opposite direction.  if puppetlabs_spec_helper decides not to support rspec 3.0.0
18:46 bodepd hogepodge: It seems that we shoudl not be so hasty to move until those tools works
18:47 hogepodge Which is why we’re probably better off undoing those changes in master and pinning to 2.x until the toolchain moves forward
18:47 mgagne bodepd: lets pin to RSpec 2.x.x then?
18:47 dvorak hogepodge: sounds good
18:47 hogepodge bodepd there was a rush to fix the gate.
18:47 bodepd mgagne: that feels like the wrong fix
18:47 bodepd mgagne: the right fix is to pin spec_helper b/c that is what everyone uses to ensure consistent versions
18:47 hogepodge I think it was the wrong fix, but it moved some people forward, so I understand it.
18:47 dvorak bodepd: I think it's the right fix until puppetlabs_spec_helper is clearly supporting rspec 3.x
18:48 mgagne bodepd: pinning spec_helper?
18:48 bodepd mgagne: I think pinning spec_helper is the right fix
18:48 mgagne bodepd: plz explain
18:48 dvorak mgagne: he means getting puppet to release a new 0.4.x version of the puppetlabs_spec_helper gem with more specific requires
18:48 hogepodge does pinning spec_helper fix spec?
18:48 mgagne bodepd: pinning in puppetlabs_spec_helper instead of our project?
18:48 dvorak hogepodge: yes, that's the ideal fix imo
18:48 bodepd pinning spec_helper is the ideal fix for everyone
18:49 bodepd clearly people are using it to figure out how to install test dependencies for consistency
18:49 mgagne dvorak: I agree it looks like a better idea
18:49 dvorak right, these can't be the only modules with this problem.  I ran into it with one of my unrelated ones last night
18:49 bodepd I quite like the idea of spec_helper being what people rely on to set test deps
18:50 bodepd kylo from pupppet-dev responded to my inquiry and asked what pull request he should look at
18:50 starmer joined #puppet-openstack
18:50 bodepd I'm pretty concerned about rspec-puppet b/c AFAICT, no one actually maintains it
18:50 mgagne bodepd: hilarious, it's already pinned: https://github.com/puppetlabs/puppetl​abs_spec_helper/blob/dd928baaf7a0ef9a​2030151581cb91e4e3c90d51/Gemfile#L6
18:51 dvorak well, a pull request can't create a new branch, can it?
18:51 dvorak mgagne: it's not in the gemspec, which is the problem
18:51 bodepd in the Gemfile, you have to look at the Rakefile
18:51 bodepd b/c that is what creates the gemspec.
18:51 mgagne dvorak: omg
18:51 bodepd mgagne: or you could scroll back to where dvorak and I chatted about this yesterday ;)
18:51 dvorak someone will have to create a new 0.4.x branch in that repo and patch 2 lines of the rakefile and release a 0.4.2 gem to fix this
18:51 bodepd also spec_hlper doesn't define it's gemspec correctly
18:52 mgagne dvorak: what is this twisted dependency system :O
18:52 dvorak heh :)
18:52 bodepd mgagne: https://github.com/puppetlabs/​puppetlabs_spec_helper/pull/53
18:52 dvorak a bit part of the problem is that master is like 50 commits ahead of the 0.4.1 release
18:52 dvorak bit=big
18:53 bodepd dvorak: I wouldl submit a patch to pin (that is step one)
18:53 dvorak I'm guessing master is moving towards a 0.5.0 or 1.0.0 release
18:53 bodepd dvorak: yep. although I would say 0.5.0
18:53 bodepd dvorak: b/c it is a bug fix
18:54 bodepd mgagne: would you say this breaks backwards compat: https://github.com/bodepd/lib​rarian-puppet-simple/pull/28
18:56 bodepd mgagne: dvorak what version should I pin it to? The latest 2?
18:56 bodepd 2.14 ?
18:56 hogepodge bodepd so wait for spec_helper to be fixed then pin to spec_helper? That seems like an indirect solution, and it looks like the PR causes breaking changes anyway.
18:56 dvorak bodepd: I'd pin rspec and mocha to the same as what's in the gemfile
18:57 mgagne bodepd: yep, the one already in Gemlist
18:57 mgagne bodepd: lets not complicate things
18:57 dvorak mocha got a new release about 2 weeks ago and it breaks some tests also
18:57 mgagne bodepd: ~> 2.14.0 should be ok I guess
18:57 dvorak and that's another area where the Rakefile and the Gemfile disagree
18:58 dvorak I believe the Gemfile says ~> 2.10.0
18:58 hogepodge All of this makes maven and its dependency handling look so nice and refreshing.
18:58 mgagne hogepodge: .....
18:58 dvorak I assume that's beacuse that's what puppetlabs tests it against, so I think that's what it should be pinned to
19:01 bodepd https://github.com/puppetlabs/​puppetlabs_spec_helper/pull/56
19:01 bodepd dvorak: you make a lot of assumptions about how well maintained this is :)
19:02 dvorak bodepd: the fact that I've talked to 3-4 PL employees about it today has given me a good idea of that ;)
19:02 bodepd well, at least we have a patch to talk around now
19:03 hogepodge bodepd I’m poking the appropriate people about that pr
19:03 dvorak bodepd: patch looks perfect
19:03 mgagne bodepd: where can I +1 a PR? =)
19:04 dvorak mgagne: I think the closest you'll get is a '+1' comment in the PR
19:04 dvorak but you can try a +2 comment and see where that gets you :)
19:04 bodepd the next question is what version it should be released as
19:04 aldavud_ bodepd: you were looking for me?
19:05 starmer joined #puppet-openstack
19:09 csschwe joined #puppet-openstack
19:09 mjblack joined #puppet-openstack
19:10 mgagne I created a wiki with dependencies and related stuff used by our modules for reference: https://wiki.openstack.org/wik​i/Puppet-openstack/Development
19:12 djm68 joined #puppet-openstack
19:24 bodepd aldavud: I was just curious about your thoughts, etc. around data model. Maybe we could chat as a message so we don't spam the channel
19:29 djm68 joined #puppet-openstack
19:34 djm68 joined #puppet-openstack
19:35 starmer joined #puppet-openstack
19:46 imcsk8 joined #puppet-openstack
19:50 starmer joined #puppet-openstack
20:02 aldavud joined #puppet-openstack
20:05 starmer joined #puppet-openstack
20:17 djm68 joined #puppet-openstack
20:20 starmer joined #puppet-openstack
20:23 sputnik13 joined #puppet-openstack
20:25 anode joined #puppet-openstack
20:32 djm68 joined #puppet-openstack
20:36 starmer joined #puppet-openstack
20:37 djm68 joined #puppet-openstack
20:50 starmer joined #puppet-openstack
21:00 ryanycoleman joined #puppet-openstack
21:05 starmer joined #puppet-openstack
21:06 djm68 joined #puppet-openstack
21:17 xarses joined #puppet-openstack
21:20 starmer joined #puppet-openstack
21:25 djm68 joined #puppet-openstack
21:29 djm68 joined #puppet-openstack
21:30 djm68 joined #puppet-openstack
21:33 prad joined #puppet-openstack
21:35 starmer joined #puppet-openstack
21:40 anode joined #puppet-openstack
21:50 starmer joined #puppet-openstack
21:57 hogepodge puppet_spec_helper 0.4.2 is on the way as a gem
21:57 hogepodge we’re making a stable branch, applying the patch back to it to handle the dependency, and publishing
21:58 hogepodge bodepd dvorak mgagne ^^
21:58 mgagne cool
21:59 dvorak hogepodge: awesome, thanks!
22:05 ryanycoleman joined #puppet-openstack
22:05 openstackgerrit David Medberry proposed a change to stackforge/puppet-nova: Add live_migration_flag to libvirt  https://review.openstack.org/98261
22:05 starmer joined #puppet-openstack
22:06 med_ hmmm, didn't mean to create a new review
22:07 hogepodge bodepd the version operator is too tight.
22:08 anode joined #puppet-openstack
22:09 openstackgerrit David Medberry proposed a change to stackforge/puppet-nova: Add live_migration_flag to libvirt  https://review.openstack.org/94457
22:10 kitp joined #puppet-openstack
22:12 hogepodge twiddle wakka is the technical term for that.
22:14 dvorak hogepodge: then the Gemfile should probably be changed also.  bodepd copied that from the Gemfile at my suggestion
22:14 djm68 joined #puppet-openstack
22:15 hogepodge Yuck
22:21 starmer joined #puppet-openstack
22:21 aimon__ hogepodge : https://forge.puppetlabs.com/aimonb/aviator
22:22 hogepodge aimon++
22:22 aimon__ mmaglana made aviator 1.8.7 compatible for us
22:22 dvorak hogepodge: isn't rspec ~> 2.10 going to pull in 2.99 ?
22:22 aimon__ as well as made it read a hash for config rather then yaml
22:22 hogepodge Probably. I’m asking the original author what we should do
22:24 dvorak seems like ~> 2.10, <2.99 might be the only reasonable path.  this is pretty gross :)
22:25 mmaglana dvorak: that's correct "~> 2.10" means "the latest 2.x" whereas "~> 2.10.0" means "the latest 2.10.x"
22:25 djm68 joined #puppet-openstack
22:25 mmaglana (and i realized i just jumped into that conversation without scrolling up first for context)
22:26 hogepodge Speaking to the original author, he thinks ~> 2.10.0 is correct. We’re trying to sort it out.
22:28 JayJ joined #puppet-openstack
22:28 badiane_ka joined #puppet-openstack
22:29 hogepodge He thinks that using it to manage the rspec dependency is a bad idea
22:29 hogepodge We’re going to talk later about it.
22:30 hogepodge aimon so the module is using rubygems to pull in dependencies?
22:30 dvorak well, that's perhaps reasonable if he's sure it works on rspec 3.  if it's not going to work on both rspec 2 and 3 in the same release, then you'll have to bump the version to 1.0.0 for rspec 3
22:34 djm68 joined #puppet-openstack
22:34 aimon__ hogepodge I tested on RedHat .. it had a rpm for active_support but not faraday
22:35 aimon__ it only needs rubygems if there is no native package for faraday
22:35 aimon__ not sure if ubuntu has one…
22:36 starmer joined #puppet-openstack
22:36 aimon__ If we desire I can extend the module to handle dependency install
22:38 hogepodge I think that there’s a general requirement to not use gems (michchap?)
22:39 hogepodge aimon__ https://etherpad.openstack.org​/p/puppet-openstack-2014-06-02
22:41 hogepodge looks like ubuntu has activesupport packaging
22:42 hogepodge not farady in 12, but in 13/14
22:44 aimon__ I can put faraday right into the module and isntall active_support vis native packge
22:44 aimon__ this way no non native dependencies
22:49 djm68 joined #puppet-openstack
22:49 ryanycoleman joined #puppet-openstack
22:49 aimon__ farday only has one deb multipart-post .. checking if has native pkg
22:49 aimon__ s/deb/dep/
22:50 starmer joined #puppet-openstack
22:58 aimon__ hogepodge I will release a new ver later tosay or tomorrow that is fully self contained . I t will only install active_support via native package.. rest will be included locally in module lib
23:00 ryanycoleman joined #puppet-openstack
23:00 sputnik13 joined #puppet-openstack
23:05 starmer joined #puppet-openstack
23:07 hogepodge Turns out the tighter pinning is probably correct. https://github.com/puppetlabs/​puppetlabs_spec_helper/pull/56
23:08 hogepodge especially with mocha, which isn’t semver
23:10 kitp joined #puppet-openstack
23:11 gildub joined #puppet-openstack
23:14 kitp joined #puppet-openstack
23:20 starmer joined #puppet-openstack
23:24 djm68 joined #puppet-openstack
23:26 ryanycoleman joined #puppet-openstack
23:27 aimon__ ugh right.. libs are loaded before manifests
23:27 aimon__ mark is cehcking on removing he activesupport dependency
23:32 ryanycoleman joined #puppet-openstack
23:34 sputnik13 joined #puppet-openstack
23:47 michchap hogepodge: I don't mind gems.
23:47 hogepodge Was it someone from mirantis? Or enovance (EmilienM?)
23:49 michchap I don't recall. I don't think gems are a good idea, but I would use them if I needed to.
23:50 starmer joined #puppet-openstack
23:57 snk joined #puppet-openstack
23:58 djm68 joined #puppet-openstack
23:59 hogepodge https://github.com/puppetlabs/​puppetlabs_spec_helper/pull/57

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