Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-12-18

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

All times shown according to UTC.

Time Nick Message
23:03 openstackgerrit François Charlier proposed a change to stackforge/puppet-horizon: Introduce keystone_url parameter  https://review.openstack.org/61177
23:12 dtalton2 joined #puppet-openstack
00:16 rongze joined #puppet-openstack
00:37 dmsimard joined #puppet-openstack
01:47 rharrison joined #puppet-openstack
01:49 rmoe joined #puppet-openstack
02:01 bcrochet joined #puppet-openstack
02:14 rongze joined #puppet-openstack
02:27 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-cinder: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58003
02:53 openstackgerrit Michael Chapman proposed a change to stackforge/puppet_openstack_builder: Add doc building script  https://review.openstack.org/61296
03:01 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-neutron: Adding ml2 firewall_driver parameter  https://review.openstack.org/62535
03:05 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-cinder: Using contain_class instead of include_class  https://review.openstack.org/62799
03:10 rongze joined #puppet-openstack
03:32 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-nova: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58045
05:30 rongze joined #puppet-openstack
05:41 openstackgerrit A change was merged to stackforge/puppet_openstack_builder: Add doc building script  https://review.openstack.org/61296
05:41 xarses joined #puppet-openstack
06:08 rongze joined #puppet-openstack
08:34 EmilienM good morning here
09:06 openstackgerrit Xingchao Yu proposed a change to stackforge/puppet-glance: Add known_stores option in class glance::api  https://review.openstack.org/61018
09:12 rcrit joined #puppet-openstack
09:12 bauzas joined #puppet-openstack
09:17 derekh joined #puppet-openstack
09:21 rcrit joined #puppet-openstack
09:21 bauzas joined #puppet-openstack
10:03 openstackgerrit A change was merged to stackforge/puppet-neutron: Using contain_class instead of include_class  https://review.openstack.org/62431
11:24 rongze joined #puppet-openstack
12:26 badiane_ka joined #puppet-openstack
12:48 morazi joined #puppet-openstack
12:59 dprince joined #puppet-openstack
13:21 rongze joined #puppet-openstack
13:29 bcrochet joined #puppet-openstack
13:56 mattymo joined #puppet-openstack
14:32 dmsimard joined #puppet-openstack
14:39 EmilienM mgagne: ping
14:43 dmsimard joined #puppet-openstack
14:45 mattymo EmilienM, still need another hour :(
14:45 mattymo or 2
14:46 EmilienM mattymo: sorry ?
14:46 mattymo oh wait
14:48 mattymo mgagne doesn't appear usually for another 1-2 hours from now
15:01 openstackgerrit Chip Baeseman proposed a change to stackforge/puppet-swift: Allow running of proxy and storage on the same node.  https://review.openstack.org/62744
15:10 snuffop left #puppet-openstack
15:12 dmsimard1 joined #puppet-openstack
15:15 EmilienM mgagne: i think the neutron bug is invalid
15:19 openstackgerrit Martin Mágr proposed a change to stackforge/puppet-neutron: VXLAN support  https://review.openstack.org/55578
15:29 otherwiseguy joined #puppet-openstack
15:39 morazi joined #puppet-openstack
15:40 rongze joined #puppet-openstack
16:05 slowe joined #puppet-openstack
16:06 dtalton joined #puppet-openstack
16:36 EmilienM puppet-neutron / i still don't understand why some flags are going into agent/ovs.pp and some in plugin/ovs.pp
16:38 EmilienM mgagne: do you ?
16:40 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-cinder: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58003
16:46 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-nova: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58045
16:55 rongze joined #puppet-openstack
16:59 openstackgerrit Simon Pruvost proposed a change to stackforge/puppet-ceph: Implements: I want to run benchmarks on three new machines Closes-Bug: #1243332  https://review.openstack.org/62937
17:08 rwsu joined #puppet-openstack
17:08 mgagne EmilienM: IMO, neutron bug is invalid.
17:08 EmilienM i do agree
17:09 EmilienM mgagne: i made a deep review on https://review.openstack.org/55578
17:09 EmilienM i hope i did not say wrong
17:09 mgagne EmilienM: for agent vs plugin, some values were used by the plugin or agent only. might have changed with ML2 both it has its own manifest anyway
17:09 EmilienM mgagne: i talked with neutron guys today, conf is gonna change again
17:09 mgagne EmilienM: agent does not need (and shouldn't have) sql_connection for example
17:09 EmilienM yeah i agree, it's only server
17:10 mgagne EmilienM: server/plugin
17:12 EmilienM mgagne: there is a "[agent]" in the ovs plugin config file
17:12 EmilienM i think it's only for agent stuff
17:12 EmilienM and let's say the [ovs] stuff is for plugin
17:12 EmilienM it's not so obvious when you look at neutron code actually
17:13 mgagne EmilienM: yes, I know. I had to dig in code to know which service was using what.
17:14 EmilienM mgagne: If I could ask something for you today is to also review https://review.openstack.org/#/c/62535/
17:14 EmilienM my first -2, I don't like to be that guy, but I was not sure about this patch
17:16 mgagne EmilienM: IMO, security group has always been a problem in the past. people made changes to neutron/nova without making it security groups where working properly although they insisted they were.
17:17 EmilienM mgagne: I'll care of that during the release. I want to have something clean in our module.
17:17 mgagne EmilienM: sure. The problem I have is that I don't have an icehouse setup yet, even less havana...
17:17 mgagne EmilienM: I used to test every changes on my setup before approving.
17:20 EmilienM mgagne: i do, no worries. I handle the problem. I just wanted your thoughts as usual
17:21 mgagne EmilienM: sure. I depend on those changes too. I want to make sure the right ones get in too and won't break stuff.
17:27 angdraug joined #puppet-openstack
17:28 EmilienM mgagne: i see a lot of difference between [agent] flags & agent/ovs.pp :'(
17:28 mgagne EmilienM: like most manifests, they have to be updated each reason
17:29 EmilienM mgagne: for now, i use to deploy neutron::plugin::ovs on all nodes
17:29 mgagne EmilienM: I don't use ovs, I use linuxbridge.
17:29 mgagne EmilienM: you shouldn't
17:29 EmilienM mgagne: OVS is more interesting imo
17:29 mgagne EmilienM: unless you run redhat based os
17:29 EmilienM mgagne: do you mean i should not use neutron::plugin::ovs everywhere ?
17:29 mgagne EmilienM: to each his own
17:30 mgagne EmilienM: it was not designed that way
17:30 mgagne EmilienM: to be used that way
17:30 mgagne EmilienM: plugin goes on server, agent goes on compute node
17:33 EmilienM mgagne: i do agree at all. but some cleanup should have been done i think
17:33 mgagne EmilienM: ok
17:34 EmilienM mgagne: like "tunnel_types" flag is in [agent] section, but in neutron::plugin::ovs. Why ?
17:34 mgagne EmilienM: I don't use ovs, you know better. =)
17:35 EmilienM mgagne: you use the easiest way, i don't like easy stuffs :P
17:35 EmilienM mgagne: we deploy ml2 / ovs / vxlan now
17:35 EmilienM mgagne: trust me it's a challenge \o/
17:35 mgagne EmilienM: I don't have that liberty.
17:35 EmilienM damn
17:36 mgagne EmilienM: I can tell you that the easy way is to not digging in code to confirm how a value is used. =)
17:38 EmilienM mgagne: what i'm doing and i'll produce a patch as soon as possible. I can swear. I really like when all is clean
17:39 brdude joined #puppet-openstack
17:39 mgagne EmilienM: you are preaching to the converted =)
17:48 EmilienM mgagne: i know you :P
17:49 slowe In 3.0.0-rc1 of puppet/nova, I see references to a nova::controller class, but I don't see any files that define that class. Am I missing something?
17:50 mgagne slowe: there is no nova::controller in puppet-nova.
17:50 slowe mgagne: OK, so I'm not missing something. Thanks for the confirmation.
17:51 mgagne slowe: where did you see this info?
17:51 slowe mgagne: It's referenced in tests/site.pp for 3.0.0-rc1 of puppet-nova.
17:51 slowe mgagne: Possibly in other manifests in tests as well, I'd need to check.
17:52 mgagne slowe: not sure where this came from. there is an openstack::nova::controller in puppet-openstack though.
17:53 mgagne slowe: https://github.com/stackforge/puppet-opensta​ck/blob/master/manifests/nova/controller.pp
17:53 mgagne slowe: but I don't see the libvirt_type parameter in it
17:54 slowe mgagne: Looks like tests/site.pp is the only affected file in puppet-nova. I knew about the nova::controller class in puppet-openstack, but I'm using the modules for the individual services (helps me better understand what's going on right now).
17:55 mgagne slowe: it an example that never got updated. controller got removed a year ago: https://github.com/stackforge/puppet-nova/com​mit/372327a384ae2f179790af3a7f0019cd7bb6d555
17:55 slowe mgagne: Fair enough, I knew the tests/* files were examples to help understand how the modules were used, but the reference to nova::controller in puppet-nova was throwing me off. Thanks for clarifying that it's just an obsolete reference, I appreciate it.
17:56 mgagne slowe: I think we should remove the example if nobody bothered to update it =)
17:57 slowe mgagne: Probably a good idea, if for no other reason that to keep from throwing off others (like me) who are still learning :-)
18:02 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-nova: Remove outdated examples  https://review.openstack.org/62945
18:04 tnoor joined #puppet-openstack
18:04 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-keystone: Refactor keystone_user_role to be lazier  https://review.openstack.org/47891
18:05 mgagne I really NEED that one. bodepd marked it as WIP for months because it needed more tests. I would like people to test it and see if they can find any problem with it.
18:05 tnoor1 joined #puppet-openstack
18:05 openstackgerrit Daneyon Hansen proposed a change to stackforge/puppet_openstack_builder: Provides Consistent HA User Defaults  https://review.openstack.org/62946
18:11 rmoe joined #puppet-openstack
18:11 xarses joined #puppet-openstack
18:13 rcrit joined #puppet-openstack
18:13 bauzas joined #puppet-openstack
18:40 morazi joined #puppet-openstack
19:17 rmoe_ joined #puppet-openstack
19:23 dachary dmsimard1: xarses dalgaaf_ did you notice https://groups.google.com/a/puppetlabs.com/​forum/#!topic/puppet-openstack/qsu3PV6RSAo ? I'm not pressing you for an answer or anything ;-) Just making sure it was not overlooked.
20:03 EmilienM mgagne: this one ? > https://review.openstack.org/47891
20:04 mgagne EmilienM: yep
20:09 dmsimard1 dachary: I usually review, I abstained from reviewing the use cases for the reasons you know
20:10 dmsimard1 dachary: I'm really busy right now and since the puppet-ceph branch I am working with right now does the work I need it to do, there is little incentive to work on the stackforge one (that is, during my "work" hours)
20:13 derekh joined #puppet-openstack
20:22 prad joined #puppet-openstack
20:32 xarses dachary: didn't see it
20:35 xarses dachary: maybe we should have a regular meeting I wasn't reviewing the mon one because i was confused about what was going on.
20:35 xarses dachary: and probably like others, figured someone else would
20:39 openstackgerrit A change was merged to stackforge/puppet-ceph: README.md: fix typo in link to USECASES.md  https://review.openstack.org/62499
20:42 dachary dmsimard: understood :-) Is to the point where you would like to step down as a core reviewer ?
20:42 * dachary not sure what the terminology is ;-)
20:44 dachary xarses: what meeting frequency would you like to establish ?
20:44 xarses dachary: maybe ad-hoc review sessions for these large patches
20:44 xarses dachary: not sure, still milling it over
20:45 dachary xarses: I abandonned the patch because I figured it was too large.
20:46 dachary but https://review.openstack.org/#/c/58989/ is quite small and is three weeks old.
20:47 dachary there surely is something we can do to get things moving
20:47 xarses dachary: ya I know, I did +2 it =)
20:48 EmilienM mgagne: your "no score" on https://review.openstack.org/#/c/61177/ is intended ?
20:48 mgagne EmilienM: I removed my -2
20:48 mgagne EmilienM: didn't have time to review yet
20:48 EmilienM mgagne: oh i see, cool.
20:49 dachary xarses: yes you did :-) dmsimard has no time, dalgaaf_ neither. We can't have an Approved just between you and me unless we change the rules.
20:50 dachary I'm open to ideas to move forward.
20:51 dtalton2 joined #puppet-openstack
20:56 marun joined #puppet-openstack
20:57 xarses dachary: ya, I agree we need to move forward, just not sure about what you suggest.
20:58 dachary between you and me that's a permanent stall because there is just two of us. Therefore no way to get 2 +2 + 1 approved.
20:59 dmsimard I had +1'd a previous patch set of https://review.openstack.org/#/c/58989/
20:59 dachary Unless we change the rule there is no way to move forward and we're stalled.
20:59 dmsimard Didn't notice there was a new patch set
20:59 badiane_ka joined #puppet-openstack
20:59 dmsimard Let me check
20:59 dachary dmsimard: a +1 won't cut it I'm afraid.
21:00 dachary dmsimard: but I'm not stressing over this patch, really ;-)
21:00 dachary I think we have a more general problem that require some creating thinking to solve.
21:00 dmsimard dachary: Why are you adding use cases in a commit for ceph repo about emperor ?
21:00 hogepodge joined #puppet-openstack
21:01 dachary ?
21:01 dmsimard oh, wait, wait .. nevermind that
21:01 dachary np
21:04 xarses dachary: I'm starting to think that a solution might involve ad-hoc review sessions and a 72hour auto approve. I don't think a rotation is a fair expectation for some people. Maybe also we should encourage neutral reviewing, in that you have no feeling to - the review, but you aren't going to review this patch
21:05 dachary could you expand on the ad-hoc review sessions ?
21:05 xarses that way we know if we are waiting for someone else to review if they already no-comment
21:06 mgagne xarses: this problem has been addressed in Nova and other projets with "review days": https://wiki.openstack.org/wiki/Nova/ReviewDays
21:06 dachary interesting, thanks mgagne
21:07 mgagne dachary: only challenge is that people aren't full time on puppet-ceph
21:07 dtalton3 joined #puppet-openstack
21:07 dachary it's what makes the project interesting too ;-)
21:09 dachary establishing a sustainable development with low fragmentation relying on occasional developers is worth a significant bootstrap effort
21:11 xarses dachary: say we land a big patch like mon, we poke irc / mailing list to schedule a review session. In the review session we should discuss what the patch is doing we are comfortable to review and establish if there are fundamental issues in the design / implementation to oust any -2 issues. All of the (interested) reviews should then be at a point where we can review any further revisions and should help minimize dead waits on
21:11 xarses the reviews
21:12 dachary xarses: ok. Let me revive the mon patch and try that. Nice idea.
21:13 dmsimard I was almost ready to give ceph::mon a +2 for the sake of moving forward because most of my concerns had been addressed (so many patch sets)
21:13 openstackgerrit A change was merged to stackforge/puppet-ceph: Make ceph::repo ensurable + add support for emperor  https://review.openstack.org/58989
21:13 dachary dmsimard: thanks for approving this one, it will make it easier to rebase the mon patch ;-)
21:13 dmsimard You can't say I wasn't involved in reviewing ceph::mon :p
21:14 dachary dmsimard: I'm sorry if I gave you this impression. It is only natural that you're busy at times.
21:14 dmsimard I am okay with building incrementally and iterating to make things better
21:15 dmsimard Initial implementation does not need/require to do anything and everything perfectly
21:15 dmsimard this is why your review of ceph::mon was hard to deal with, it's like.. bam, here's something that can do everythinhg
21:15 dmsimard We'd probably have a better velocity if we built the module little by little
21:16 mgagne I agree with dmsimard, huge monolithic review are really hard to do. They take a look of time and discourage people from reviewing and contributing because there is so much to review.
21:16 mgagne I got burned myself by those reviews when we did major refactors in puppet-openstack.
21:18 dmsimard So, anyway, in that respect, I am okay with letting ceph::mon in that current state go through because "it works" and we can revisit it when/if necessary.. maybe even refactor it if we find better ways to do things
21:19 dmsimard Hope that makes sense.
21:19 dmsimard dachary / xarses ^
21:20 * dachary reading while local rspec:system checks the sanity of the rebase of ceph::mon
21:22 xarses dachary: looking over the current patch
21:23 dachary dmsimard: I relate to this (large patches are difficult to deal with). I'm quite confortable with baby steps too, even if it requires more steps. My concern is mainly that baby steps are not going to work either if each baby steps requires weeks to be approved, even when they are truly non controversial.
21:24 dachary I like xarses idea and if it works, that would allow us to quickly bootstrap the ceph module.
21:24 xarses ugg, why is vagrant so hard to customize vm options
21:25 dachary We're in the same refactoring state that mgagne mentions. Baby step refactoring is tough.
21:25 mgagne dachary: big changes discourage people from reviewing. There reasoning is they don't have time NOW to review it and probably won't have it anytime soon.
21:25 dachary mgagne: absolutely.
21:26 mgagne dachary: be it a refactor or complete bootstrap, it's the same result.
21:26 dachary And I was ready to slice ceph::mon into maybe 10 to 15 baby steps. But contemplating 10 x 3 weeks average review delay for ceph::mon ... I figured we should talk about it before going down this route.
21:28 xarses dachary: could you maybe add an example/site.pp?
21:30 dachary xarses: you have a few here https://review.openstack.org/#/c/5​6841/16/spec/system/ceph_mon_spec.rb
21:30 dachary line 88
21:30 dachary line 30
21:31 dachary line 138
21:31 mgagne dachary: worst place ever to look for examples =)
21:31 xarses lol
21:31 dachary haha
21:31 xarses dachary: maybe we sus them into examples and have the rspec load them?
21:31 mgagne xarses: reading my mind
21:31 dachary xarses: I thought you were looking for examples to read, not for examples to publish, sorry for the confusion
21:32 xarses dachary: both
21:32 otherwiseguy joined #puppet-openstack
21:32 dachary nice idea
21:32 mgagne xarses: we did the same for jenkins-jobs-builder. Docs had outdated examples and weren't tested. We used test fixtures as examples instead. Examples are now tested and updated whenever code/behaviour is changed.
21:33 dachary mgagne: how would you deal with variables ?  release => '#{release}', for instance
21:34 xarses dachary: fake facts?
21:35 xarses dachary: hmm not good for user examples
21:35 openstackgerrit Loic Dachary proposed a change to stackforge/puppet-ceph: ceph::mon create or destroy monitors  https://review.openstack.org/56841
21:36 xarses dachary: I like moving the manifests to examples from the test cases, then they are easy to find, and we test them so a user know's that format will work
21:36 dachary mgagne: could you please show a URL of the jenkins-jobs-builder tests ?
21:37 dachary xarses: absolutely
21:40 xarses Is 5 min sufficient dachary ?
21:41 xarses =P
21:41 dachary 5 min for what ?
21:41 xarses to review
21:41 dachary it's sufficient for a lot of things
21:41 dachary for the review meeting ?
21:42 dachary a 5 minutes review meeting ?
21:42 * dachary very confused right now ;-)
21:42 xarses I +2'd it for the sake of moving forward, and reviewing the patch about 10 times before you resumed it
21:43 dachary oh
21:43 xarses I didn't have an opinion prior
21:43 dachary ok thanks !. I will work on externalizing the examples, excellent idea :-)
21:44 dachary xarses: dmsimard dalgaaf_ : when would be the best time/day for you to meet for a review session ?
21:45 dachary I'm flexible.
21:46 xarses I'm on UTC-8 Summer, UTC-7 Fall, most times between 8am and 6pm are fine
21:50 hogepodge ping EmilienM
21:51 hogepodge or mgagne
21:52 EmilienM hogepodge: pong
21:53 hogepodge I'm assuming that if I want to use MongoDB with Ceilometer, I need to configure that manually, right?
21:53 hogepodge No autoconfig except for MySQL?
21:55 EmilienM hogepodge: let me check
21:55 hogepodge I'm not terribly familiar with Ceilometer. Why on earth would I choose mongo over MySQL? Do I lose anything with MySQL?
21:56 hogepodge Google tells me yes. :-P
21:56 dmsimard Ceilometer transaction volume is crazy, mongodb is good at that
21:56 dmsimard That is my understanding :)
21:57 EmilienM hogepodge: i use ceilometer::db with database_connection = mongodb://my_host/ceilometer
21:57 EmilienM hogepodge: and that's it
21:58 hogepodge Does mean I need to create the user and password manually, though? Or will the connector manage that for me?
21:58 EmilienM hogepodge: mongodb is really designed for ceilometer since telemetry stores too many datas
21:59 EmilienM hogepodge: nothing to create
21:59 EmilienM hogepodge: i think you can, but by default you have not
21:59 EmilienM hogepodge: fc__ is working on replica set in mongodb puppet module for your information
22:00 hogepodge Ok. Thanks. Having a schema-less database has its advantages.
22:00 EmilienM hogepodge: if you use MySQL, i highly suggest you using the patch I did some days ago with TTL support
22:00 EmilienM ceilometer::expirer
22:00 EmilienM in production, ceilometer could store many GB per day
22:00 hogepodge I just published a major refactor of the mongodb modules today. It's likely the API is stable now, but no guarantees.
22:01 xarses hogepodge: from testing that i've seen there have been some compelling arguments to use mongo if you say poll every 5 seconds
22:01 hogepodge I'm going to go with the mongodb backend
22:01 EmilienM good choice
22:01 EmilienM hogepodge: if you don't need me anymore, i have to go, quite late here :)
22:01 hogepodge Since there seems to be distinct advantages. Consolidation would have been nice, but it's not essential of rme.
22:02 hogepodge Thanks Emilien.
22:02 hogepodge Good night!
22:02 EmilienM my pleasure, good luck, and let me know (or fc__ ) about ceilometer
22:03 hogepodge s/of rme/for me
22:03 bcrochet joined #puppet-openstack
22:04 hogepodge Anyone know what's up with Smokestack? It's been MIA for quite a while.
22:05 hogepodge I've been approving without it, but every time I do it gives me a queasy feeling after our summit discussions.
22:10 dachary mgagne: do you think changing the rule for puppet-ceph to allow a core reviewer to approve after 2 working days of silence AND a +2 is sensible ?
22:22 openstackgerrit A change was merged to stackforge/puppet-ceph: ceph::mon create or destroy monitors  https://review.openstack.org/56841
22:23 xarses huzzah!
22:51 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: spec:system - Change VM IP to fix virtualbox/vagrant error  https://review.openstack.org/63008
22:52 openstackgerrit David Moreau Simard proposed a change to stackforge/puppet-ceph: spec:system - Change VM IP to fix virtualbox/vagrant error  https://review.openstack.org/63008
22:53 rongze joined #puppet-openstack

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