Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-10-23

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

All times shown according to UTC.

Time Nick Message
00:09 dwt2 joined #puppet-openstack
00:14 ryanycoleman joined #puppet-openstack
00:20 ryanycoleman joined #puppet-openstack
00:29 e1mer joined #puppet-openstack
00:58 ari_ joined #puppet-openstack
01:05 ari__ joined #puppet-openstack
01:36 ryanycoleman joined #puppet-openstack
01:47 xarses joined #puppet-openstack
02:14 ari_ joined #puppet-openstack
02:33 ryanycoleman joined #puppet-openstack
02:50 badiane_ka joined #puppet-openstack
03:04 xingchao joined #puppet-openstack
03:17 openstackgerrit joined #puppet-openstack
04:15 comptona joined #puppet-openstack
04:24 comptona joined #puppet-openstack
04:36 e1mer joined #puppet-openstack
04:37 ryanycoleman joined #puppet-openstack
05:26 openstackgerrit A change was merged to stackforge/puppet-keystone: Fix duplicated keystone endpoints  https://review.openstack.org/52675
05:28 bodepd_ mgagne: merged your patch. I did some testing on it, and reviewed the code
05:45 odyi joined #puppet-openstack
06:19 xarses joined #puppet-openstack
06:58 Totonyus joined #puppet-openstack
07:06 ldachary joined #puppet-openstack
07:06 ldachary \o/
07:20 marun joined #puppet-openstack
07:25 ndubrunfaut joined #puppet-openstack
07:42 e1mer joined #puppet-openstack
07:44 mmagr joined #puppet-openstack
07:49 rongze joined #puppet-openstack
08:11 derekh joined #puppet-openstack
08:32 xingcha__ joined #puppet-openstack
08:32 rongze_ joined #puppet-openstack
08:38 beddari joined #puppet-openstack
08:45 francois1 joined #puppet-openstack
09:27 Totonyus joined #puppet-openstack
09:31 Retarded joined #puppet-openstack
09:42 ryanycoleman joined #puppet-openstack
09:44 marun joined #puppet-openstack
11:20 rongze joined #puppet-openstack
11:24 mmagr joined #puppet-openstack
11:44 ryanycoleman joined #puppet-openstack
12:02 bauzas joined #puppet-openstack
12:10 morazi joined #puppet-openstack
12:57 dprince joined #puppet-openstack
13:14 morazi joined #puppet-openstack
13:18 mmagr joined #puppet-openstack
13:29 otherwiseguy joined #puppet-openstack
13:45 ryanycoleman joined #puppet-openstack
13:48 fvollero fellas
13:49 fvollero i have a puppet 3.1.1 and when I run facter, i just see ipaddress value but not ipaddress_eth0 and ipaddress_eth1
13:50 dmsimard joined #puppet-openstack
13:54 dmsimard dachary: ping
13:54 dachary dmsimard: \o
13:54 dmsimard dachary: does every file really need an apache header ?
13:54 dmsimard dachary: or a copyright notice, for that matter
13:55 dachary dmsimard: did you know about https://github.com/ceph/ceph/bl​ob/v0.61.9/src/ceph-disk#L2027 ?
13:56 dachary dmsimard: legally speaking, no. But files travel from project to project and it's really painful to have to figure out the license / author when they do. The ceph_config is a good example. And you cannot (never) assume a license unless it's written down somewhere.
13:57 dachary It's simple enough to just add it always. No trouble at all.
13:57 dachary copy/paste 2 sec
13:57 dmsimard dachary: was a genuine question cause looking at other modules, I don't see headers like that.
13:58 dmsimard be right back
13:58 dmsimard left #puppet-openstack
13:58 dachary I know and we discovered that it's a issue that forces us to dig the history to make sure we're not copying over something that we should not.
13:58 dmsimard joined #puppet-openstack
13:59 dachary It's not a legal issue, it's a practical issue.
14:02 dachary inserting the header is a few seconds when the file is creating. Digging the history to check the license is a few minutes in the best case scenario. IIRC when the puppet modules for openstack were still young, we copied a file from a github repository which had no mention of license whatsoever. The author was known for his work with OpenStack so we figured he intended apache. But we knew that guessing was not good enough when it comes
14:02 dachary to licensing and I asked him for a clarification. It took two months and weekly gentle nudge to get a definitive answer ( that took a few seconds of *his* time but hours of mine ).
14:05 dmsimard dachary: Okay - I guess I haven't worked a lot with licensing...
14:05 dmsimard dachary: Saw the link, what about it ?
14:05 dachary did you not suggest to have a list of devices be excluded from auto discovery ?
14:06 dachary I may be confused ;-)
14:08 dmsimard dachary: Yes and no. In my use case, servers for ceph storage are dedicated to ceph - the OS is installed somewhere and the remainder of the disks are to be used for ceph.
14:08 mjblack joined #puppet-openstack
14:08 dachary ok :-)
14:13 dmsimard dachary: Sorry - got interrupted
14:14 dmsimard dachary: So, how I do it is that I list all of the server's block devices and I exclude the OS disk
14:21 dachary ok
14:22 piliman974 joined #puppet-openstack
14:30 dmsimard Starting to work on ceph::repo now, don't you guys be as hard on me as Xarses :P
14:34 openstackgerrit Maru Newby proposed a change to stackforge/puppet-openstack: Ensure Nova Network-compatible provisioning  https://review.openstack.org/53366
14:46 dmsimard Do we really want ceph::repo to be optional ? As in, should it have a parameter "before => Package['ceph']" for instance ?
14:47 ryanycoleman joined #puppet-openstack
14:47 mgagne dmsimard: people might not have the choice to use the version provided by the distribution
14:47 mgagne dmsimard: if enabled, the package installation should behave correctly
14:47 dmsimard mgagne: Does various openstack modules depend on openstack::repo ?
14:47 mgagne dmsimard: no
14:48 mgagne dmsimard: totally optional
14:48 dmsimard mgagne: ok
14:50 dmsimard In that respect, I see that puppetlabs/apt is not a dependency in puppet-openstack
14:50 dmsimard Okay.
14:51 marun bodepd_, mgagne: reviews welcome on the backport of my provisioning fix: https://review.openstack.org/#/c/53366/
14:51 marun bodepd_, mgagne: I need to get a dependent packstack backport in ahead of a packaging update.
14:51 ryanycoleman joined #puppet-openstack
14:52 marun bodepd_, mgagne; I have tested it to work with quantum enabled and disabled.
14:55 rongze joined #puppet-openstack
15:00 marun joined #puppet-openstack
15:04 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add module's Puppetfile  https://review.openstack.org/53382
15:05 dmsimard That's an easy one.
15:08 dmsimard bodepd_: I notice that in puppet-openstack the Modulefile, fixtures and Puppetfile are not consistent. Is there a reason for that ?
15:08 ryanycoleman joined #puppet-openstack
15:08 ryanycoleman joined #puppet-openstack
15:26 dachary bodepd_: ok / late reply to : cisco may we are in the process of moving our cluster from behind the firewall
15:26 dmsimard dachary: What version(s) would you like to pin stdlib at.. ? Looking at various modules, stdlib is not pinned
15:27 dachary I don't have an opinion as long as it's pinned to avoid regression.
15:28 dmsimard I'll pin at 4.x
15:28 dachary there as two things that make the life a puppet developer a recuring nightmare : lack of integration tests and fuzzy dependencies ;)
15:29 dachary I'm working on integration tests right now
15:37 andi- joined #puppet-openstack
15:39 marun dachary: the same is sadly true of openstack
15:40 marun though maybe s/fuzzy dependencies/distributed complexity/
15:42 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add module's Puppetfile  https://review.openstack.org/53382
15:45 ari__ joined #puppet-openstack
15:51 dachary dmsimard: I did not know you could :ref => 'origin/4.x' :-) nice
15:51 morazi joined #puppet-openstack
15:51 dmsimard dachary: Yeah, it pins to that branch - it's the same syntax used elsewhere. I don't deserve any credit :)
15:51 openstackgerrit A change was merged to stackforge/puppet-ceph: Add module's Puppetfile  https://review.openstack.org/53382
15:52 dmsimard mgagne explained to me a bit why puppet-openstack's Modulefile, Puppetfile and fixtures were different
15:53 dmsimard In the dependency they declare, at least
16:01 mgagne dachary: https://review.openstack.org/#/c/53382/
16:02 mgagne dachary: I don't know why you approved right off the bat
16:02 mgagne dachary: this puppetfile is unusable without puppet-ceph itself
16:02 dachary How do you mean ?
16:03 mgagne dachary: as mentioned in the email sent to puppet-openstack mailinglist, we (bodepd_ and myself) have to chance to review stuff.
16:03 dachary arg
16:03 dachary mgagne: my bad :-(
16:03 mgagne dachary: and give a chance to other core reviewers to review too
16:04 dachary I thought the rule was enforced to have 2 core reviewers before a change goes in.
16:04 mgagne dachary: no
16:04 mgagne dachary: don't check "approve" until you mean it
16:04 dachary mgagne: point taken
16:07 rongze joined #puppet-openstack
16:07 mgagne dachary: as core reviewers, you can +2 stuff in the "Code Review" section. +2 only mean your vote have more weight in the final decision. In the "Approved" section, this is where you can accept a change. There is no logic/check associated to that vote which means it will get merged right now. There is no rule checking if someone else has +2 or not the change. it's all based on good faith.
16:09 dachary I'll remember that, thanks for explaining. Again, I apologize for abusing the merge rights. :-(
16:09 marun mgagne, bodepd_: I would really appreciate having one more review to approve the provisioning backport: https://review.openstack.org/#/c/53366/
16:10 mgagne marun: here we go, lgtm
16:10 marun mgagne: much appreciated! :)
16:10 dmsimard dachary: I discussed with mgagne - I understand his position and it's legitimate. The Puppetfile is meant to be used to provision a puppetmaster - My objective with that commit was to add the Puppetfile so we can add dependencies as they are added in future commits
16:10 dmsimard dachary: Since puppet-ceph doesn't quite exist yet, the Puppetfile is more or less useless
16:11 mgagne dachary: what dmsimard said
16:11 dmsimard mgagne: At this point, do we remove the file or work with it ?
16:11 mgagne dachary: Puppetfile could be leveraged for integration tests
16:12 mgagne dmsimard: up to you
16:12 dachary I thought about integration tests when I saw it and it seemed great in that regard
16:12 dmsimard mgagne: I don't think it hurts to have the file there, albeit useless for the present time
16:16 hogepodge joined #puppet-openstack
16:18 mjblack joined #puppet-openstack
16:19 mjblack joined #puppet-openstack
16:19 openstackgerrit A change was merged to stackforge/puppet-openstack: Ensure Nova Network-compatible provisioning  https://review.openstack.org/53366
16:20 xingchao joined #puppet-openstack
16:28 ryanycol_ joined #puppet-openstack
16:28 mjblack_ joined #puppet-openstack
16:39 ryanycoleman joined #puppet-openstack
16:39 mjblack joined #puppet-openstack
16:40 xarses joined #puppet-openstack
16:41 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-keystone: Remove the get_service_id method and inline python  https://review.openstack.org/53399
16:42 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-keystone: Fix duplicated keystone endpoints  https://review.openstack.org/53400
16:43 badiane_ka joined #puppet-openstack
16:48 morazi joined #puppet-openstack
16:52 mjblack mgagne: hmm I reviewed that wsgi keystone and it looks like the primary issue is that rspec-puppet issue
16:54 mjblack mgagne: pupept is converting integers to strings and when it tries to evaluate it in rspec its expecting an integer and not a string
17:02 hogepodge joined #puppet-openstack
17:03 ari_ joined #puppet-openstack
17:08 mjblack_ joined #puppet-openstack
17:12 rongze joined #puppet-openstack
17:23 ryanycoleman joined #puppet-openstack
17:24 xarses dachary: re: https://review.openstack.org/#/c/53014/ There are four options to get the rspc test to test the ' = ' that i was able to find, 1) run as root 2) mimic the inifile code the way i did (except reversed) 3) create another subclass to re-override file_path  4) don't test it here, test that inifile supports it instead
17:26 mjblack joined #puppet-openstack
17:27 mjblack joined #puppet-openstack
17:29 mgagne xarses: better check how upstream is doing: https://github.com/puppetlabs/puppetlab​s-inifile/blob/master/spec/unit/puppet/​provider/ini_setting/ruby_spec.rb#L518
17:30 mgagne xarses: don't spend more time reinventing stuff people already figured out
17:30 xarses mgagne thats how i came to the tests that i wrote
17:30 mgagne xarses: cool
17:30 xarses without being able to use :path
17:30 xarses you cant change the file with out hacking the class
17:31 xarses and therefor it requires write access to /etc/ceph/ceph.conf
17:31 mgagne xarses: would there be an interest in making file_path overridable?
17:32 xarses honestly, i dont know enough ruby to do that
17:32 mgagne xarses: I guess you found that one already: https://github.com/puppetlabs/puppetlab​s-inifile/blob/master/spec/unit/puppet/​provider/ini_setting/ruby_spec.rb#L86
17:32 mgagne xarses: if you REALLY need to override the file_path
17:32 xarses didn't see that one
17:32 bodepd_ xarses: check out the rspec stub method
17:33 mgagne xarses: if there is an interest to allow override of file_path, I don't mind spending time to figure out how to do it. someone already asked us to be able to do it on mailinglist.
17:34 bodepd_ dmsimard: I did *some* scroll-back and it looks like mgagne got you sorted
17:35 dmsimard bodepd_: Yeah
17:35 xarses mgagne: doesn't allowing the path param allow the correct override of file_path (if you want to be able to use it in puppet)
17:40 ryanycoleman joined #puppet-openstack
17:41 rcrit any puppet-horizon maintainers about? I need some core acks on https://review.openstack.org/#/c/49799/
17:53 ryanycoleman joined #puppet-openstack
17:56 ryanycoleman joined #puppet-openstack
17:59 xarses still need dachary to comment
17:59 dachary bodepd_: mgagne what contact email should I provide for the gerrit service user of puppet-ceph ? https://bugs.launchpad.net​/openstack-ci/+bug/1243776
18:00 mgagne dachary: who is going to maintain and host such infra?
18:00 mgagne dachary: I don't plan on working on it or maintaining it in the foreseeable future
18:01 dachary The idea is that you would need an openstack tenant and a single script with credential running on a vm on the tenant.
18:01 mgagne dachary: I do understand the idea
18:02 mgagne dachary: are you planning on working in it?
18:02 mgagne on it*
18:02 dachary yes, I'm working on making it happen in a way that it can be trivially hosted by anyone with an openstack tenant with enough resources
18:03 dachary and I'll use a tenant on my personal infra to start until a more permanent host is found
18:03 dachary mgagne: does that make sense ?
18:03 mgagne dachary: better put an email you can access for the moment
18:03 mgagne dachary: it can be changed later if needed
18:05 ryanycoleman joined #puppet-openstack
18:10 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add ceph::repo  https://review.openstack.org/53412
18:10 ryanycoleman joined #puppet-openstack
18:13 rongze joined #puppet-openstack
18:16 xarses dachary: ping
18:21 bodepd_ dachary: when I created CI users in the past, I just made up an email
18:29 dachary mgagne: ok
18:29 dachary xarses: pong
18:30 xarses dachary: re https://review.openstack.org/#/c/53014/
18:30 xarses There are four options to get the rspc test to test the ' = ' that i was able to find, 1) run as root 2) mimic the inifile code the way i did (except reversed) 3) create another subclass to re-override file_path  4) don't test it here, test that inifile supports it instead
18:31 xarses thoughts?
18:33 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add ceph::repo  https://review.openstack.org/53412
18:34 dmsimard dalgaaf: Switched to cuttlefish as default
18:34 xarses dmsimard: no rhel options?
18:35 dmsimard xarses: I have no experience on ceph on redhat-like - the existing puppet-ceph does not support it
18:35 mgagne dmsimard: we could throw a warning or fail if platform is not supported by ceph::repo
18:36 ryanycoleman joined #puppet-openstack
18:36 dmsimard mgagne: I thought about that, yeah, but then.. where does it end ? What about other distros ? We need to handle this in all of the classes ?
18:36 mgagne dmsimard: this is usually handled by ceph::params
18:37 mgagne dmsimard: see https://github.com/puppetlabs/puppetlabs​-apache/blob/master/manifests/params.pp
18:37 xarses dmsimard: ya, I need to support RedHat os's, you can reference what we've done in fuel or create bugs for me to look over to add similar parts
18:37 xarses mgagne: except for the repo's fun
18:38 dmsimard xarses: I would be more comfortable if you could add on top of what's in this commit for redhat support
18:38 dmsimard mgagne: Yeah I saw what's done for nova - ceph::params doesn't exist yet :)
18:39 xarses https://github.com/Mirantis/fuel/blob/mast​er/deployment/puppet/ceph/manifests/yum.pp and https://github.com/Mirantis/fuel/blob/master​/deployment/puppet/ceph/manifests/params.pp
18:39 dachary xarses: what do bodepd_ or mgagne prefer ? They are much more knowledgeable than I am in this ;-)
18:40 dmsimard xarses: If you're comfortable with that, I'll open a bug
18:40 xarses dachary: mgagne said something about someone would want to support being able to override file_path anyway, maybe allowing the path param settles both.
18:40 xarses dmsimard: please do
18:42 dachary I'll let them decide on this, I'm not the right person to ask how it should be done.
18:44 dmsimard xarses: https://bugs.launchpad.net​/puppet-ceph/+bug/1243872
18:45 xarses dmsimard: ty
18:47 hogepodge joined #puppet-openstack
18:56 ryanycoleman joined #puppet-openstack
18:57 ari_ joined #puppet-openstack
19:11 xarses dmsimard: there is no copyright, author, or license in your files since we are apparently doing all that
19:13 rongze joined #puppet-openstack
19:14 dmsimard xarses: yeah ….
19:17 dmsimard xarses: I personally have a hard time with that, I don't believe it's very common to do this for puppet modules
19:19 dmsimard xarses: I get the point but looking at a lot of modules, license blocks are a rare occurence. There is a LICENSE at the root of the repo, collaborators..
19:21 mgagne dmsimard: the current opinion is that file has life of their own
19:22 badiane_ka joined #puppet-openstack
19:23 openstackgerrit Matthew J Black proposed a change to stackforge/puppet-keystone: Enable serving keystone from apache mod_wsgi  https://review.openstack.org/53430
19:28 marun joined #puppet-openstack
19:34 dmsimard EmilienM: ping
19:35 dmsimard dachary: ping
19:36 dachary dmsimard: pong
19:38 dmsimard dachary: I remember leseb saying Enovance would change puppet-ceph to Apache 2, did I miss an update ? It's obvious we'll borrow snippets from the existing puppet-ceph, what should we do regarding licensing ?
19:38 dmsimard dachary: xarses mentioned that I didn't include a License/Copyright/Author block for ceph::repo, so it's on that topic.
19:38 dachary dachary: that would be a question for EmilienM || leseb ;-)
19:39 dachary dmsimard: what snippets do you need urgently ?
19:39 EmilienM plop
19:39 dmsimard dachary: Yeah, tried asking you since you're more of a licensing guy than I am - I guess they're asleep
19:39 dmsimard EmilienM: ^
19:39 EmilienM i'm not a licensing guy, when i work on openstack, it's license apache 2.0
19:39 xarses dmsimard: if you are copy and pasting, you can just post that license for that file and we can clean it up when the contributors agree to the license change
19:40 EmilienM when i work for me, it's gpl
19:40 dmsimard EmilienM: Okay, I'll try talking to leseb tomorrow, he's not pinging :)
19:41 hogepodge joined #puppet-openstack
19:44 dmsimard xarses: like so ? http://paste.openstack.org/show/49274/
19:45 xarses yes
19:46 dmsimard Okay I'll send another patch
19:46 xarses that makes _that_ file agpl
19:47 mgagne xarses: and the whole module agpl no?
19:49 xarses mgagne: effectivly
19:50 mgagne xarses: better rewrite it from scratch no? But I guess it would be harder for people unfamiliar with puppet
19:50 dmsimard I got a hold of leseb on #ceph, didn't see he was in here for some reason
19:50 dmsimard mgagne, xarses: There are snippets used, it's not a carbon copy
19:50 leseb dmsimard: we can continue here :)
19:51 dmsimard leseb: yeah, sorry about that.
19:51 mgagne bodepd_: mind reviewing https://review.openstack.org/#/c/53399/ and https://review.openstack.org/#/c/53400/ ?
19:51 leseb dmsimard: haha np
19:52 dmsimard leseb: So, I've very obviously taken parts of ceph::apt::ceph from enovance/puppet-ceph to create "ceph::repo": https://review.openstack.org/#/c/53412/
19:53 xarses dmsimard: can you use https://github.com/Mirantis/fuel/blob/mast​er/deployment/puppet/ceph/manifests/apt.pp as the snippit source?
19:53 dmsimard leseb: xarses got me thinking puppet-ceph was still agpl - I'm not a license guy so I just want to make sure I'm not in the wrong
19:54 leseb dmsimard: same goes for me, I'm not a license guy at all
19:54 xarses dmsimard: how close are they?
19:55 dmsimard leseb: Okay, I'll give it some time and see what Enovance says
19:55 xarses ahh, very close
19:55 dmsimard xarses: close enough
19:56 dmsimard xarses: not exactly re-inventing the wheel and there's not 10 ways of adding a repository through puppet :)
19:56 xarses I'm aware, but its nearly the same, enough to consider it derivitave
19:57 dmsimard xarses: Yeah, that's why I'll allow leseb to get a response back and see what happens
19:57 dmsimard xarses: Then we can choose what to do - i'll put it on hold for the time being
19:59 ryanycoleman joined #puppet-openstack
19:59 xarses we can 1) use a a different source for the snippits. 2) license it as agpl and clean up later, but it pollutes the repo. 3) I can use the mirantis parts to refrence both debian and Redhad configs from our repo which is Apache 2.0 already
19:59 xarses redhad/redhat
20:00 dmsimard dachary: Most of your review comment can be answered by the blueprint ?
20:00 dachary yes, copy/paste would be good indeed
20:00 dachary dmsimard: ^
20:01 dmsimard ok.
20:02 xarses bodepd_: mgagne: thoughts on keeping the path param in https://review.openstack.org/#/c/53014/ to allow for easy testing and also added benefit of allowing the file_path to be changed if someone wants the config somwhere else?
20:10 dmsimard xarses: What flavor of redhat like will you be testing on ?
20:10 dmsimard RHEL, Fedora, CentOS ?
20:10 xarses centos 6 usually
20:10 dmsimard ok
20:10 dmsimard That's good because you will always be behind me to add redhat features :P
20:10 dmsimard (or me behind you to add debian)
20:12 xarses oh, i have to support ubuntu 12.04 too
20:13 xarses and if RedHat makes a decision about RHOS that too
20:14 dachary dmsimard: could you point me to the original code from enovance please ?
20:14 rongze joined #puppet-openstack
20:14 xarses https://github.com/enovance/puppet-ce​ph/blob/master/manifests/apt/ceph.pp
20:16 dmsimard dmsimard: https://github.com/enovance/puppet-ce​ph/blob/master/manifests/apt/ceph.pp and https://github.com/enovance/puppet-ceph/blo​b/master/spec/classes/ceph_apt_ceph_spec.rb
20:16 dmsimard er.
20:16 dmsimard dachary: ^ it's largely the same
20:23 ryanycoleman joined #puppet-openstack
20:23 dachary dmsimard: it's mostly variable settings and only a very few. Think about it in this way : say you did not read eNovance file, would you have done things differently ? And if so, show me how different.
20:23 dachary Only non trivial work is copyrightable and in this specific case, it's trivial.
20:24 dmsimard dachary: Like I told xarses earlier, there aren't 10 different ways to add a repo and test that it was added
20:24 dachary to add the repo I don't think so
20:25 xarses dmssimard: do you mind if i just redo this from my working branch ?
20:26 dmsimard xarses: let's give Enovance some time, there's no emergencies right now and retrieving code from their repository can surely be convenient
20:26 xarses dachary: its lovely gray area, some jurisdictions don't interpret trivial the same
20:26 dachary it's not
20:27 dachary we're way beyond the limit
20:27 xarses hopefuly
20:27 dmsimard xarses: I'm a user of a fork (dalgaaf's) of their puppet-ceph right now and it works well, I don't see the need to rewrite everything if it works the way we want it to. I see a need, for instance, to do something about the storeconfigs.
20:28 dachary it's obvious
20:29 dmsimard dachary, xarses: Just give it some time, it's late now in Paris. leseb might have an answer for us tomorrow, it's not necessary to debate something that could be solved in a timely manner
20:29 dachary https://github.com/stackforge/puppet-ope​nstack/blob/master/manifests/repo/uca.pp looks more like https://review.openstack.org​/#/c/53412/2/manifests/repo.pp and is licensed under apache ! That's a clear sign of trivial ;-)
20:31 dmsimard dachary: That's funny, I hadn't seen that, I actually did the "if" on purpose to avoid redhat-like problems
20:31 ryanycol_ joined #puppet-openstack
20:32 xarses dachary: +1
20:32 xarses lets merge it as Apache 2.0 =))
20:33 dachary +1
20:33 dmsimard Okay.
20:33 dmsimard I'll add Francois Charlier as an author, least I can do
20:33 dmsimard (Enovance)
20:37 dmsimard dachary: I looked and there's actually not much about supported operating systems in the blueprint. There is a mention of supporting cuttlefish and up only, but nothing about OS.
20:37 hogepodge joined #puppet-openstack
20:38 dmsimard dachary: Should we add specifically Ubuntu 12.04 and CentOS 6 ? If it works on Ubuntu 12.04, it would surely work on Debian 7 as well.. But then the integration tests you're doing get complicated ?
20:40 xarses dmsimard: I'd opt towards case $::osfamily
20:40 xarses and only add things if you must do them in specific releases
20:40 dmsimard xarses: Sure.
20:41 xarses we can allways set the facts a specific way if a second pattern arises for the test cases
20:42 dmsimard xarses: I'll somewhat adjust the spec and the class to be broader
20:43 xarses if you can start the case that would be nice, otherwise i'll end up adding it
20:43 dmsimard i'll do it
20:44 openstackgerrit Andrew Woodward proposed a change to stackforge/puppet-ceph: Add ceph_config ini helper  https://review.openstack.org/53014
20:44 mjblack what version of ruby does jenkins use to do rspec?
20:45 xarses probably 1.9.1
20:45 mjblack ewww
20:46 xarses its better than 1.8.7 that i have to work with
20:47 mjblack_ joined #puppet-openstack
20:48 openstackgerrit A change was merged to stackforge/puppet-keystone: Remove the get_service_id method and inline python  https://review.openstack.org/53399
20:48 mjblack_ xarses: some would debate you on that
20:48 openstackgerrit A change was merged to stackforge/puppet-keystone: Fix duplicated keystone endpoints  https://review.openstack.org/53400
20:48 mjblack_ I'm just not able to reproduce the rspec errors that it encountered on puppet 2.7 and 3.0
20:49 bodepd_ xarses: I don't mind allowing path overriddes
20:49 bodepd_ xarses: one concern is that, it you do it, purging will not work
20:50 bodepd_ xarses: it may be worth mentioning in the param inline docs
20:51 xarses bodepd_ : true about the purge; I'll drop a note in there
20:53 dachary back
20:53 openstackgerrit Andrew Woodward proposed a change to stackforge/puppet-ceph: Add ceph_config ini helper  https://review.openstack.org/53014
20:54 xarses bodepd_ added a note.
20:55 dachary dmsimard: regarding the comments, I'm not picky really. As long as it reflects the most important points I'm good with it. It should not be something that duplicates what git log --stat shows.
21:00 dmsimard dachary: Should we add something about Debian/Ubuntu/CentOS on the blueprint, though ?
21:00 dachary dmsimard: sure, good idea
21:01 mgagne dmsimard: I believe there would be some interest in running puppet-ceph on scientificlinux and other derivated redhat platforms
21:02 mgagne dmsimard: https://github.com/cernceph/puppet-ceph/
21:02 mgagne dmsimard: http://www.slideshare.net/Inktan​k_Ceph/cern-ceph-day-london-2013
21:02 mgagne dmsimard: the idea would be to not limit ourselves to CentOS specifically
21:02 dmsimard mgagne: What distribution is scientificlinux derived from? Debian ?
21:02 xarses redhat
21:02 mgagne dmsimard: redhat
21:02 mjblack_ rhel
21:03 xarses :)
21:03 dmsimard mgagne: Well, the idea is that the module should support both the osfamily debian and redhat
21:04 mgagne dmsimard: we had some cases where we (at puppet openstack) were too restrictive/specific in the platforms supported although SL would have worked just as fine
21:04 dmsimard mgagne: Valid point, i'll document it broad enough
21:09 * mjblack_ mutters something negative about jenkins
21:14 rongze joined #puppet-openstack
21:15 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add ceph::repo and it's unit tests to puppet-ceph  https://review.openstack.org/53412
21:17 dmsimard xarses: That should be good
21:18 mjblack joined #puppet-openstack
21:18 dmsimard xarses: Oh, actually. I JUST thought about something..
21:19 ryanycoleman joined #puppet-openstack
21:19 dmsimard xarses: It should probably be better to opt for a format more like puppet-openstack where there's like "openstack::repo" and then "openstack::repo::uca"
21:20 mgagne dmsimard: this can be refactored later on
21:22 dmsimard mgagne: Yeah, we can settle with this to get things started.
21:26 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: Add ceph::repo and it's unit tests to puppet-ceph  https://review.openstack.org/53412
21:28 ryanycoleman joined #puppet-openstack
21:34 dmsimard dachary: Do you plan on doing integration tests on both rhel and debian ?
21:34 dmsimard dachary: That's a lot of work :)
21:37 dachary You are correct. But is there another way ?
21:40 dmsimard Not that I know of :)
21:40 dmsimard I do my own CI tests with the current release of puppet-ceph, but with vagrant
21:40 dmsimard (vagrant, vagrant-openstack, openstack)
21:41 mgagne ask bodepd_ about how awesome (or not) it is =)
21:41 dmsimard hmm ? He doesn't like our setup ? :p
21:42 mgagne I think he is using a similar setup, don't know about his opinion on it
21:43 ryanycoleman joined #puppet-openstack
21:45 dachary dmsimard: is it published somewhere ? Your CI that is. It would be a big help.
21:47 dmsimard In it's current state it's too wrapped/tailored to our business processes to be of use to you, I can extract the idea and samples that could be useful
21:48 dmsimard The core of it is vagrant-openstack: https://github.com/cloudba​u/vagrant-openstack-plugin
21:59 mgagne dachary: all you need to know to start is vagrant and its concepts. once you know that, you can build CI tests over it and then move to openstack backed boxes.
22:01 dmsimard Signing off for today, talk to you guys later.
22:08 dachary \o
22:09 dachary mgagne: it has been useful in the context of openstack-installer indeed
22:10 mgagne dachary: this is IF you want to use vagrant which is an easy way to get you started with virtual machines, provisioned and controlled environments.
22:11 dachary oh, I did not realize that vagrant could use openstack instead of virtual box
22:11 dachary nice !
22:11 mgagne dachary: that's what we (dmsimard and myself) use to test our stuff.
22:12 mgagne dachary: openstack-infra uses nodepool to dynamically pre-provisionned one-time jenkins slaves. it isn't trivial to setup and I wouldn't recommend going that path due to the complexity of the system.
22:15 rongze joined #puppet-openstack
22:15 dachary why not use the openstack api directly and got thru vagrant ?
22:15 dachary s/got/go/
22:15 dachary mgagne: ^
22:16 dachary let me rephrase
22:16 dachary what vagrant has that the openstack api does not ?
22:18 xarses rainbows and unicorns
22:21 dachary :-)
22:22 Kupo24z left #puppet-openstack
22:26 mgagne dachary: vagrant up example.org
22:27 mgagne dachary: and your box starts and provisions itself
22:27 mgagne dachary: it's like asking what does IKEA provide that a hammer, some wood and nails can't provide.
22:28 dachary better than nova boot because you get to define the parameters in a separate file instead of the command line ?
22:28 mgagne dachary: you have to read about vagrant first
22:29 mgagne dachary: and make up your own opinion
22:29 mgagne dachary: http://www.vagrantup.com/
22:30 dachary I've used it recently to test openstack-installer and I did not see such a difference. Things are not defined at the same place ( more configuration in vagrant, more command line in nova ) but in the end I don't see that it's less work.
22:30 mgagne dachary: sorry to hear that your experience wasn't as nice as ours ;)
22:30 dachary I'm not complaining at all !
22:31 dachary I'm happy to have vagrant to test openstack-installer :-)
22:31 dachary I found it different, not inferior or cumbersome.
22:31 dachary Not superior or with a level of abstraction that makes it more convenient either.
22:32 mgagne dachary: with openstack, you would have to include a cloud-config to automate the puppet provisioning OR ssh into the instance to do it yourself (be it automated or not). with vagrant, it's all handled for you. People who wish to test puppet-ceph won't necessary have access to an openstack infra. With vagrant, you can host your boxes on whatever hypervisor you have, virtualbox being the default one.
22:33 dachary that's a good point
22:34 dachary Could you point me to where the vagrant doc explains how to automate puppet provisioning ?
22:34 mgagne dachary: come on =)
22:35 mgagne dachary: google "vagrant puppet provisioner", first link
22:35 mgagne dachary: http://docs-v1.vagrantup.com/v1​/docs/provisioners/puppet.html
22:35 dachary I think you convinced me with this single argument ( and now I remember that openstack-installer makes use of it but since it's implicit I did not even notice )
22:35 mgagne dachary: for latest vagrant: http://docs.vagrantup.com/v2/p​rovisioning/puppet_apply.html
22:35 dachary thanks
22:37 mgagne dachary: it's mainly about simplicity and allowing us to move faster right off the bat instead of (re)creating yet an other similar system from scratch.
22:38 dachary I'll prepare the puppet-ceph integration environment with vagrant then. And throw away what I started to do with nova cli and no regret.
22:38 mgagne dachary: people might outgrow vagrant one day and they shouldn't feel the need to stick to it if it doesn't answer their needs anymore
22:38 dachary it seems fit for puppet-ceph
22:38 mgagne dachary: whatever you feel comfortable with.
22:40 dachary I was looking for things that people use instead of starting my own thing ( which I generaly don't like ).
22:41 dachary I did not even consider vagrant because I did not know an OpenStack backend was possible.
22:42 dachary mgagne: have you published CI vagrant files / setup scripts that you use by any chance ?
22:43 dachary https://github.com/CiscoSystems/openst​ack-installer/blob/master/Vagrantfile is already a source of inspiration but more is better ;-)
22:44 mgagne dachary: you don't need that much complexity
22:45 mgagne dachary: hold on
22:47 mgagne dachary: basic Vagrantfile http://paste.openstack.org/show/49324/
22:48 mgagne dachary: this assumes modules and manifests are available in the same folder as your Vagrantfile
22:48 * dachary looking
22:48 dachary now I better understand what you mean by Ikea versus hammer
22:48 mgagne dachary: box_url is optional if you already have the precise64 box downloaded/installed on your computer
22:49 dachary mgagne: I owe you one for opening my eyes
22:49 dachary night !
22:49 mgagne dachary: np
22:49 mgagne dachary: a+
22:54 xarses hmm
22:55 xarses thinking about https://wiki.openstack.org/wiki/Puppet-o​penstack/bp-cinder-volume-multi-backend again
22:55 iwi hi there, the openstack::compute class seems to be missing tenant_network_type and all the VLAN specific parameters
22:56 iwi is this going to change in havana compatible version ?
22:57 ryanycoleman joined #puppet-openstack
22:58 xarses instead of having to to call each backend, and then call a type and pass it backends, is it possible to just add a type to each backend, and have some fancy way of it knowing if it's creating a new one or adding another node to it
23:03 mgagne iwi: unless someone else than you come up with this need, nothing will be done ;)
23:03 mgagne xarses: have an example? I don't fully understand the question
23:04 iwi mgagne: even if I propose the changes ? :)
23:05 mgagne iwi: if you propose then yes =)
23:09 xarses mgagne: https://wiki.openstack.org/wiki/Puppet-​openstack/bp-cinder-volume-multi-backen​d#Use_two_or_more_cinder-volume_backend​s_and_assign_them_to_different_types
23:10 mgagne xarses: as opposed to?
23:11 xarses the first as apposed to the second in that block
23:11 mgagne xarses: I'm confused about the terminology used
23:11 mgagne xarses: oh, checking
23:11 mgagne xarses: the goal is to associate backends to types right?
23:12 mgagne xarses: is this the right wiki page about multi-backends and types? https://wiki.openstack.org​/wiki/Cinder-multi-backend
23:13 xarses mgagne: afaik
23:14 xarses and correct, after the backends are defined, the types need to be created / associated
23:14 mgagne xarses: so: you have to create entries in cinder.conf for each backends (which unique names). Each backend also defines an "alias" (volume_backend_name) which can be used in volume type (volume_backend_name is extra_specs) to defined which backend handle which type. right?
23:16 mgagne xarses: multi-backends would require 2 features in puppet-cinder: multi-backends in cinder.conf and volume_type management (with extra-specs support)
23:16 xarses yep
23:17 rongze joined #puppet-openstack
23:18 mgagne xarses: and now the question is: "who" owns the association between the volume_type and the backend. right?
23:19 mgagne xarses: I prefer the first example as it offers more flexibility and would be easier to implement in puppet
23:20 xarses ya, seems easier
23:20 mgagne xarses: in the 2nd example, you still have to somehow declare your volume_type right?
23:20 xarses ya
23:20 xarses something would need to collect and munge them
23:25 bodepd_ xarses: that patch looks great
23:25 bodepd_ xarses: although, I don't see why you couldn't stub the class instance method of the provider
23:26 bodepd_ Puppet::Provider::Ini_setting.stub​s(:path).returns('some_temp_path')
23:26 bodepd_ (I forget exactly how to refer to the provider class objects, or if you should use the provide method
23:32 andi- joined #puppet-openstack
23:32 andi- joined #puppet-openstack
23:33 mgagne bodepd_: just in case you haven't see it already: https://github.com/maestrodev/puppet-blacksmith
23:34 bodepd_ bodepd_: I knew they had some release tools
23:34 xarses bodepd_ for the rspec, yes i see an example of that now.
23:34 bodepd_ bodepd_: but I haven't mucked with it
23:34 bodepd_ xarses: I'm fine with the implementation as is
23:35 mgagne mgagne: I wonder why bodepd_ talks to himself
23:35 bodepd_ xarses: but, I woudl hate to add a feature we don't want for testing
23:35 bodepd_ my hands and brain are not always properly linked
23:36 xarses bodepd_: mgagne mentioned that someone posted (somewhere) that they wanted to be able to change the path on occasion
23:37 mgagne xarses, bodepd_: https://groups.google.com/a/puppetlabs.com/​forum/#!topic/puppet-openstack/EV4UbDzFyyU
23:38 bodepd_ mgagne: xarses that is such a painful requirement :)
23:38 bodepd_ mgagne: they should use chroot ;)
23:39 mgagne bodepd_: come on, shouldn't be that hard :P
23:39 mgagne bodepd_: and now I just want to prove you wrong :D
23:39 bodepd_ mgagne: from an implemenation perspective of purging on providers, I am not sure that it is possible
23:39 bodepd_ I was thinking about this the other night
23:39 bodepd_ the questions is:
23:39 bodepd_ can the instances method access the catalog
23:40 bodepd_ b/c it would need to search it to see all usages of path for it's type
23:40 mgagne bodepd_: oh, purging... well, you can't purge 2 files, that's for sure
23:40 bodepd_ AFAIK, the provider class instances cannot access the catlaog programatically
23:40 bodepd_ QED :)
23:40 mgagne bodepd_: prefetch has access to it I think
23:40 mgagne bodepd_: with resources parameter
23:41 bodepd_ the problem is that instances is what purging uses
23:41 bodepd_ so it needs to be accessible to that
23:41 bodepd_ so that it can locate all resource across all files
23:42 bodepd_ I noticed that it looks like you may have gotten your keystone patch from some example in a book
23:42 bodepd_ can't seem to remember which one
23:42 bodepd_ mgagne: ^^^
23:42 mgagne bodepd_: not sure what you are talking about =)
23:47 mgagne bodepd_: personal notes about custom provider: https://gist.github.com/mgagne/7128846
23:51 ryanycoleman joined #puppet-openstack
23:52 bodepd_ the only thing to add is that your getter/setters over the auto-magical things from mkresource methods
23:54 mgagne bodepd_: true
23:54 mgagne bodepd_: in fact, in this example, I wouldn't need to define def version at all

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