Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-03-05

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

All times shown according to UTC.

Time Nick Message
00:12 ryanycoleman joined #puppet-openstack
00:19 ryanycoleman joined #puppet-openstack
00:22 britthouser joined #puppet-openstack
00:44 openstackgerrit A change was merged to stackforge/puppet-ceilometer: Add ability to override notification_topics  https://review.openstack.org/69747
00:45 britthouser joined #puppet-openstack
00:47 thumpba joined #puppet-openstack
00:51 kitp joined #puppet-openstack
01:01 kkirkpatrick joined #puppet-openstack
01:08 prad joined #puppet-openstack
01:28 xingchao joined #puppet-openstack
01:43 angdraug joined #puppet-openstack
01:51 kitp joined #puppet-openstack
01:58 fandi joined #puppet-openstack
02:00 mgagne joined #puppet-openstack
02:10 mgagne1 joined #puppet-openstack
02:12 kitp joined #puppet-openstack
02:13 kitp joined #puppet-openstack
02:19 dvorak michchap: perhaps they shouldn't have stable in the name then? ;)
02:24 kitp joined #puppet-openstack
02:24 xarses joined #puppet-openstack
02:39 rmoe joined #puppet-openstack
02:40 prad joined #puppet-openstack
03:07 michchap dvorak: probably. I guess we merge slightly less stuff on them :)
03:33 mgagne joined #puppet-openstack
03:49 adreyer joined #puppet-openstack
03:56 meylor joined #puppet-openstack
04:01 fandikurnia joined #puppet-openstack
04:05 meylor left #puppet-openstack
04:11 ryanycoleman joined #puppet-openstack
04:40 kitp_ joined #puppet-openstack
05:04 odyi joined #puppet-openstack
05:10 kitp joined #puppet-openstack
05:13 ryanycoleman joined #puppet-openstack
06:11 kitp joined #puppet-openstack
07:02 saju_m joined #puppet-openstack
07:12 kitp joined #puppet-openstack
07:14 ryanycoleman joined #puppet-openstack
07:28 mkoderer joined #puppet-openstack
07:36 xingchao joined #puppet-openstack
07:54 bauzas joined #puppet-openstack
08:00 EmilienM good morning
08:04 rmoe_ joined #puppet-openstack
08:13 kitp joined #puppet-openstack
08:15 comptona joined #puppet-openstack
08:16 mkoderer joined #puppet-openstack
08:19 openstackgerrit joined #puppet-openstack
08:42 mmagr joined #puppet-openstack
09:00 ianw joined #puppet-openstack
09:00 bcrochet joined #puppet-openstack
09:00 odyi joined #puppet-openstack
09:00 mgagne1 joined #puppet-openstack
09:02 mgagne1 joined #puppet-openstack
09:07 comptona joined #puppet-openstack
09:13 kitp joined #puppet-openstack
09:15 ryanycoleman joined #puppet-openstack
09:27 bauzas joined #puppet-openstack
09:34 saju_m joined #puppet-openstack
09:43 openstackgerrit Gonéri Le Bouder proposed a change to stackforge/puppet-horizon: ensure online cache is and can be refreshed  https://review.openstack.org/75431
09:43 openstackgerrit Gonéri Le Bouder proposed a change to stackforge/puppet-horizon: improve WSGI configuration  https://review.openstack.org/67726
10:12 michchap goneri: ping
10:14 michchap goneri: in 75431, if $compress_offline is not set, is python-lesscpy needed? (I don't know too much about horizon, but I'm trying to review your change)
10:14 kitp joined #puppet-openstack
10:16 kitp joined #puppet-openstack
10:18 ryanycoleman joined #puppet-openstack
10:19 goneri michchap: Olla !
10:19 goneri michchap: it was more to be consistent
10:20 michchap goneri: is that package installed as a dep by horizon anyway?
10:20 haleema joined #puppet-openstack
10:21 goneri michchap: no because ubuntu guys pre-generate the case. so it's only a build dependency for them.
10:21 goneri michchap: on Debian, it's different, the cache is not generated during the .deb construction.
10:22 goneri So basicly, with my changes, we ignore Ubuntu pre-generated cache.
10:22 michchap goneri: ok. So on Ubuntu I don't need to set compress_offline, but on Debian I do?
10:23 goneri michchap: without the patch, yes. But that's not an option actually. Because online cache generation is very painful
10:23 michchap goneri: ok so you're saying we just ignore the packagers because they are not doing the right thing?
10:23 goneri michchap: this take about 10s on a modern server for every pages.
10:23 michchap goneri: No problem. Any idea on rhel?
10:23 comptona joined #puppet-openstack
10:23 goneri michchap: once the cache is refreshed, it's ok, but the very first user will get a poor experience.
10:24 goneri I don't know for RHEL, I remember python-lesscpy is a dependency.
10:27 goneri michchap: http://pkgs.org/fedora-20/fedora-i386/python-​django-horizon-2013.2-2.fc20.noarch.rpm.html
10:27 goneri 2013-09-09 - Matthias Runge <mrunge@redhat.com> - 2013.2-0.10b3
10:27 goneri - add runtime requirement python-lesscpy
10:28 michchap goneri: but it's not required by the package. Is that a mistake?
10:30 michchap goneri: in any case...I am happy with the change and have +2
10:31 goneri humI don't know, maybe the dependency is already pulled by Djanogo
10:32 goneri michchap: thank you for you time.
10:32 mattymo joined #puppet-openstack
10:33 michchap goneri: no problem. I did a noop install of dashboard on a centos6 vm and it didn't list lesscpy, but I may not have the right repo set, so it's not conclusive.
10:34 openstackgerrit A change was merged to stackforge/puppet-nova: Parameterize disk_cachemodes  https://review.openstack.org/77819
10:39 goneri michchap: python-lesscpy is BuildRequires on Fedora20, manage.py compress is called at the build time.
10:40 michchap goneri: Interesting! Thanks!
10:44 michchap EmilienM: are you happy with using ensure_package? I dislike it, but there are a couple of uses of it already in ceilometer and one in neutron.
10:45 EmilienM michchap: why aren't you happy ? it avoid dupp resources afik
10:45 goneri I would be glad to get some feedback for this changes https://review.openstack.org/#/c/67726/
10:46 michchap EmilienM: The same thing can be done by making a class that creates the package and using include. I know it's easier, but I also know there are some people that regard it (and ensure_resource) as a bad practice.
10:46 comptona joined #puppet-openstack
10:47 michchap EmilienM: I am OK with it, I just thought I would ask.
10:48 EmilienM michchap: using include don't solve the dupp resource, am i wrong?
10:48 michchap EmilienM: if you make a new class and include that class from two places, the class is only in the catalog once.
10:50 EmilienM michchap: ok
11:17 kitp joined #puppet-openstack
11:18 ryanycoleman joined #puppet-openstack
11:52 dachary_ joined #puppet-openstack
11:54 mkoderer_ joined #puppet-openstack
11:55 comptona_ joined #puppet-openstack
11:56 sbadia_ joined #puppet-openstack
11:58 spredzy joined #puppet-openstack
12:18 kitp joined #puppet-openstack
12:19 ryanycoleman joined #puppet-openstack
12:27 dmsimard joined #puppet-openstack
12:55 dvorak michchap: thanks so much for pointing out that error.  I was confused as to why all of those tests were failing, but I've never submitted anything to that project either.
13:14 morazi joined #puppet-openstack
13:18 kitp joined #puppet-openstack
13:20 ryanycoleman joined #puppet-openstack
13:25 Butcho joined #puppet-openstack
13:39 dmsimard dachary: Sorry :(
13:48 meylor joined #puppet-openstack
13:50 toabctl joined #puppet-openstack
13:51 openstackgerrit Chris Ricker proposed a change to stackforge/puppet_openstack_builder: Add swift mappings for full_ha scenario  https://review.openstack.org/78223
13:52 kitp joined #puppet-openstack
13:54 dachary dmsimard: what about ?
13:55 dmsimard dachary: reviewed ceph::osd
13:55 dachary I'm *glad* you take time to review ;-0
13:55 dmsimard dachary: Ok then :)
13:58 openstackgerrit Chris Ricker proposed a change to stackforge/puppet_openstack_builder: Add swift mappings for full_ha scenario  https://review.openstack.org/78223
14:06 kitp joined #puppet-openstack
14:07 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: define ceph::osd  https://review.openstack.org/74645
14:11 dmsimard dachary: didn't address my comments about $data and the fact
14:11 dmsimard also: whitespace
14:12 dmsimard oh, wait, I didn't read the comments
14:17 openstackgerrit Mark T. Voelker proposed a change to stackforge/puppet_openstack_builder: Expose libvirt disk_cachemodes setting  https://review.openstack.org/78227
14:17 openstackgerrit A change was merged to stackforge/puppet_openstack_builder: Expose VPNaaS settings  https://review.openstack.org/77615
14:23 morazi joined #puppet-openstack
14:42 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: define ceph::osd  https://review.openstack.org/74645
14:43 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: define ceph::osd  https://review.openstack.org/74645
14:46 openstackgerrit A change was merged to stackforge/puppet-openstack: Unpin puppetlabs-apache fixture to use latest version  https://review.openstack.org/77926
14:49 dmsimard1 joined #puppet-openstack
14:51 mgagne joined #puppet-openstack
14:56 dachary is there a convenient way to restart the jenkin tests ?
15:04 otherwiseguy joined #puppet-openstack
15:09 mgagne1 joined #puppet-openstack
15:10 britthouser joined #puppet-openstack
15:21 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: define ceph::osd  https://review.openstack.org/74645
15:21 ryanycoleman joined #puppet-openstack
15:34 britthouser joined #puppet-openstack
15:35 mgagne1 michchap: ping
15:35 michchap mgagne: hey :)
15:36 mgagne michchap: I'm having issues with keystone_user_role =)
15:36 michchap mgagne: oh dear
15:36 michchap mgagne: I broke it (again)?
15:36 mgagne michchap: somehow, it's not finding the user_id =)
15:37 michchap mgagne: I didn't change the non-laziness for user_id, only tenant_id
15:37 mgagne michchap: as I'm wondering if you had this issue (yet)
15:37 michchap mgagne: I haven't, but it makes sense...it will be handled the same as the tenant I guess.
15:38 mgagne michchap: cool, I'm reading the code and I guess I'm a bit confused with the name chosen for some methods...
15:39 mgagne michchap: get_user accepts tenant_id and name as parameters... right... as if you need a tenant_id to get a user =)
15:39 michchap mgagne: *sigh*
15:41 mgagne michchap: ok then, I'll check what I can do to add non-laziness for user too
15:41 michchap mgagne: I don't manage many keystone objects because I'm mostly looking at deployment, so I probably won't see the same issues as you
15:42 mgagne michchap: I manage deployment too =)
15:42 mgagne michchap: this deployment includes additional users =)
15:42 michchap mgagne: I thought you also had a cluster with many users
15:42 michchap mgagne: aha
15:48 mgagne michchap: damn, it's giving me headaches
15:50 rharrison joined #puppet-openstack
15:50 michchap mgagne: I get headaches every time I have to work on the custom types
15:50 mgagne michchap: this one is particularly bad
15:50 mgagne michchap: I don't know how it managed to work so far
15:51 mgagne michchap: it assumes the user is already a member of the tenant when prefetching stuff
15:51 britthouser joined #puppet-openstack
15:51 michchap mgagne: I was running without the lazy load patch for a long time because it never worked for me
15:52 mgagne michchap: I guess that's what bodepd was referring too
15:52 michchap mgagne: if I hadn't been able to fix it last month I probably would ahve proposed reverting it
15:56 ryanycoleman joined #puppet-openstack
15:59 xingchao joined #puppet-openstack
16:00 ryanycoleman joined #puppet-openstack
16:02 ryanycoleman joined #puppet-openstack
16:05 ryanycoleman joined #puppet-openstack
16:15 dmsimard joined #puppet-openstack
16:16 mgagne michchap: I got an hypothesis
16:17 mgagne michchap: I think it managed to work so far because when you create a user, you can provide a tenant too. keystone will automatically add the member role to the newly created user to the tenant in question.
16:18 mgagne michchap: then when keystone_user_role builds its cache, the user is found as a member of the tenant and user_id is cached properly.
16:19 dachary xarses: are you around ?
16:19 mgagne michchap: my use case is different, I want to add a role to a user for a "non-native" tenant.
16:19 xarses dachary: yes
16:20 mgagne michchap: therefore the user doesn't any role nor is he a member of the tenant.
16:20 dachary xarses: good morning ! Do you happen to know how to reschedule a jenkins test ( https://review.openstack.org/#/c/74645/ failed for no reason )
16:20 mgagne dachary: recheck no bug
16:20 xarses mgagne: =)
16:20 dachary as a review comment ?
16:24 mgagne dachary: yes
16:25 dachary mgagne: xarses thanks !
16:27 * dachary should learn how to read ( https://wiki.openstack.org/wiki​/GerritJenkinsGit#Test_Failures )
16:30 mgagne =)
16:39 mgagne whyyyyyyyyyyyyyy
16:39 mgagne -Explanation: puppetlabs
16:39 mgagne +Explanation: : puppetlabs
16:39 mgagne Apt::Pin
16:41 morazi joined #puppet-openstack
16:42 fandi joined #puppet-openstack
16:44 morazi joined #puppet-openstack
16:48 otherwiseguy joined #puppet-openstack
16:48 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Ensure Neutron DB is populated  https://review.openstack.org/76358
16:48 EmilienM mgagne: neutron db sync with the right tag name ^
16:52 mgagne michchap: ping
16:52 mgagne michchap: I think I found the problem
16:53 mgagne michchap: get_users is resetting the @users variable
16:58 openstackgerrit Chris Ricker proposed a change to stackforge/puppet-swift: swift classes need to include swift::params  https://review.openstack.org/78302
16:58 morazi joined #puppet-openstack
17:01 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: define ceph::osd  https://review.openstack.org/74645
17:03 rwsu joined #puppet-openstack
17:07 hogepodge joined #puppet-openstack
17:14 derekh joined #puppet-openstack
17:15 dprince joined #puppet-openstack
17:35 mgagne I'm soooooo confused by puppet providers now...
17:46 xarses joined #puppet-openstack
17:47 kitp joined #puppet-openstack
17:48 rmoe joined #puppet-openstack
17:59 sputnik13 joined #puppet-openstack
18:04 sputnik13 joined #puppet-openstack
18:10 kitp joined #puppet-openstack
18:24 sputnik13 joined #puppet-openstack
18:26 openstackgerrit joined #puppet-openstack
18:27 kitp joined #puppet-openstack
18:32 sputnik13 joined #puppet-openstack
18:45 sputnik13 joined #puppet-openstack
18:54 ryanycoleman joined #puppet-openstack
18:54 hogepodge mgagne ?
18:55 mgagne hogepodge: managed to make it work...
18:55 hogepodge :-)
18:55 bauzas joined #puppet-openstack
18:55 mgagne hogepodge: each time I dig into puppet providers, I have to unlearn everything as I feel everything doesn't behave the same each time I check
18:59 blkperl hogepodge: so whats the plan with puppetlabs-havana, I'm worried that you're the only dev on it and that if you get hit by a bus the project dies until PL hires a new person, is there a plan to have it enter stackforge?
19:00 hogepodge Well, it's open source, so it can't really ever die any more.
19:00 hogepodge :-)
19:01 blkperl hogepodge: well sure someone could fork it.. but thats not really optimal....
19:01 hogepodge My plan was to just keep working on it. If we still want to replace it into puppet-openstack we could do that.
19:01 hogepodge I think that's what we want to have happen here.
19:02 hogepodge For development I want a few things to happen, including ssl and ha where possible.
19:03 blkperl well I think that needs to happen if you want people to use it... "92 downloads vs 8,454 downloads" doesn't fill me with confidence
19:04 hogepodge <nod>
19:04 blkperl are other people excited about it? I like the role/profile layout
19:05 hogepodge I've had mixed responses to it. People who discover it after using puppet-openstack seem pretty happy.
19:05 hogepodge It's lacking if features listed above.
19:06 hogepodge I've had a lot of positive feedback privately, and I know that one group that I can't name is going to start using it.
19:06 blkperl well the openstack module doesn't support HA, and ssl doesn't seem that hard to implement :P
19:07 blkperl can we get a vote going on the mailing list? I think a decision needs to be made spliting the devlopment is time wasted that we could be spending to make the other modules more awesome
19:07 hogepodge HA is a few different problems. the databases, which has a set of solutions. The message queue, which should be the easiest to set up since it's HA out of the box, and the services themselves, which I'll probably stuff behind pacemaker
19:08 hogepodge We can, yes.
19:09 blkperl :)
19:14 ryanycoleman blkperl: a vote on what, moving to stackforge? Why? It's a reference deployment on top of the modules in StackForge, not a module for one of the OpenStack projects.
19:17 dtalton joined #puppet-openstack
19:17 blkperl ryanycoleman: because hogepodge doesn't scale, stackforge does. I assume you want people to use this reference deployment? If so then it needs to be in stackforge and it needs to replace the openstack module
19:20 ryanycoleman blkperl: I agree that Chris could use help managing contributions and releases. The rest of the mod-eng team could get more involved now that we're on the other side of PE 3.2.
19:21 ryanycoleman blkperl: My gut-take is that moving the havana reference module to stackforge would be a mis-use of that infrastructure.
19:22 ryanycoleman are you more worried that there's friction getting PRs reviewed?
19:25 hogepodge If puppet-openstack is suffering from bit-rot (which it is), it makes sense to have the parent project be something that is maintained and gives some indication of how to use the modules themselves.
19:25 hogepodge Most groups are still going to have their own composition layers, though.
19:26 hogepodge I don't see Cisco or RedHat or eNovance giving up their work, and it doesn't make sense to do that given they meet particular needs.
19:27 dmsimard yeah we use a custom layer here.
19:28 blkperl ryanycoleman: I'm worried about puppet-openstack bitroot
19:28 mgagne dmsimard: no! the secret sauce!
19:28 blkperl bit-rot*
19:28 Hunner Really puppet-openstack should be puppet-openstack-example-old, and puppetlabs-havana could be called puppet-havana-example (since most people probably don't understand they are reference examples, not full required deployments)
19:29 blkperl right, the implication from the README is that you can use it to setup a full deployment
19:29 mgagne Hunner: puppet-openstack was already considered an "example" or "reference" according to README, it didn't stop PL from advertising it as THE implementation on their blog and twitter...
19:29 blkperl that too ^ :P
19:30 Hunner Marketing oversight. Whoops
19:30 blkperl so since we have hiera why can't puppetlabs-havana be used for a full deployment
19:30 mgagne there will be issues as long as PL expects this "reference" to be THE reference and advertise it as such to their customers and publicly
19:31 mgagne people will expect this reference to support XYZ and all weird use cases they can think of.
19:32 hogepodge I try to explicitly state that my work is highly opinionated.
19:32 blkperl hogepodge: will it be good enought for 95% of users?
19:32 mgagne hogepodge: sure but who cares if it "works" and you can sell it? =)
19:32 hogepodge From a marketing standpoint, we try to talk about the modules a a collection, and they really do cover a lot, largely because they're being used out there in the wild.
19:33 Hunner mgagne: Hoge *does* try to tell them not to say that, for what it's worth
19:33 hogepodge Talking about the collective vs the individual "parent" is hard.
19:33 hogepodge The parent handles the dependency tree and gives some convenience functions. That's part of why I'm talking about a library module that pulls some of that out.
19:33 mgagne Hunner: yes, I know the song, nothing new here =)
19:33 ryanycoleman For those who only hear of OpenStack as a buzz-term, understanding the individual component modules is nearly impossible to ask. Marketing gravitates to the reference example modules because they're much easier to wrap your head around and get started with.
19:34 hogepodge If we (and by we I mean I) need to clear that language up that's good feedback to have.
19:36 blkperl I think we are straying from the original problem, "the openstack module is suffering from bit-rot" what are we going to do about it
19:37 blkperl I recognize that others have their own secret sauce that they sell but a user should be able to setup an HA openstack cloud with an "official" reference module
19:38 hogepodge That's what I would want.
19:38 blkperl enterprise customers are still going to pay for the secret sauce but hackers will use the free stuff and make the other component modules better
19:39 hogepodge I think that there can be a module that pushes out a basic robust cloud with a strong set of reasonable assumptions.
19:40 dtalton isn't that the goal of puppet_openstack_builder?
19:40 mgagne ^
19:41 mgagne I don't feel it's our "job" to provide a rabbitmq HA reference, it's the job of puppetlabs-rabbitmq to provide it
19:42 blkperl is it? because the README for puppet_openstack_builder doesn't explain what it does beyond testing.
19:43 mgagne then maybe we should state what we want (or not) to cover/support
19:44 mgagne so we don't end up with a new frankenstein
19:44 hogepodge that's franken-steen
19:46 mgagne :D
19:48 dtalton blkperl, http://docwiki.cisco.com/wiki/Ope​nstack:Havana-Openstack-Installer
19:48 dtalton blkperl, that has some cisco specific settings of course, but most everything else we push upstream
19:52 hogepodge So for those here, and we can bring this up on the mailing list too. What do _you_ want to have happen to puppet-openstack. And please, be as biased as you'd like.
19:56 blkperl start a thread and ill leave my bias :)
19:58 blkperl also people keep telling me that tripleo will "solve all the deployment problems" and make the the puppet module pointless, is that what we're headed for?
19:59 dmsimard tripleo ?
20:00 blkperl https://wiki.openstack.org/wiki/TripleO
20:03 dmsimard Oh god, I don't have time to read all that now.
20:03 dmsimard LOL
20:03 dmsimard bookmarked i'll check it out later
20:03 mgagne dmsimard: franchement david
20:03 mgagne :D
20:04 mgagne that was one of the main feature demoed at OpenStack Summit at the booths
20:04 mgagne ironically, backed by puppet =)
20:07 ryanycoleman joined #puppet-openstack
20:13 EmilienM mgagne: +1 for providing rabbitMQ HA with puppetlabs module only
20:14 mgagne I don't want to have to maintain nova::rabbitmq anymore or cinder::rabbitmq or whatever new class will get introduced to other modules because people expect it to be provided ^^'
20:15 EmilienM totally agree
20:15 mgagne We could however document in the README an example of a simple rabbitmq config
20:16 EmilienM or make it in puppet-openstack module
20:16 EmilienM (i don't use the module but afik it's an high level module)
20:16 mgagne it's already there no?
20:16 EmilienM oh maybe
20:27 dtalton joined #puppet-openstack
20:34 ryanycoleman joined #puppet-openstack
20:41 hogepodge triple o is deployment with bash scripts. The HP people behind it are very anti-config management.
20:42 hogepodge there's work to make the actual deployment layer be pluggable, though (meaning puppet, etc)
20:57 ryanycoleman joined #puppet-openstack
21:02 markvoelker hogepodge: EmilienM: if either of you have a few minutes, could you take a look at https://review.openstack.org/#/c/77785/ ?
21:02 markvoelker It's the keystone patch mgagne and I were discussing the other day.
21:08 rmoe_ joined #puppet-openstack
21:20 markvoelker hogepodge: thanks
21:20 hogepodge np
21:41 otherwiseguy joined #puppet-openstack
21:42 otherwiseguy joined #puppet-openstack
21:42 sputnik13 joined #puppet-openstack
21:46 EmilienM michchap: could you update your review on https://review.openstack.org/#/c/76462/ please?
21:46 EmilienM hogepodge: could you have a look at https://review.openstack.org/#/c/68940/ please?
21:59 xarses joined #puppet-openstack
22:13 hogepodge EmilienM yes, meeting now, but I'll get at it soon
22:14 otherwiseguy joined #puppet-openstack
22:35 dtalton joined #puppet-openstack
22:36 ryanycoleman joined #puppet-openstack
22:49 ryanycoleman joined #puppet-openstack
23:01 xarses joined #puppet-openstack
23:02 hogepodge emilienm done and done. Can you look over the MySQL patches when you get a chance?
23:04 ryanycoleman joined #puppet-openstack
23:23 ryanycoleman joined #puppet-openstack
23:42 unixpro1970 joined #puppet-openstack
23:59 kitp joined #puppet-openstack

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