Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-02-13

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

All times shown according to UTC.

Time Nick Message
00:25 lnxnut joined #puppet-openstack
00:29 aimon joined #puppet-openstack
00:42 ericpeterson joined #puppet-openstack
00:52 xingchao_ joined #puppet-openstack
01:08 rcallawa joined #puppet-openstack
01:09 rcallawa joined #puppet-openstack
01:22 openstackgerrit Dulanjalie Ganegedara proposed stackforge/puppet-neutron: Adding portdb and fastpath_flood to n1kv.conf  https://review.openstack.org/155560
01:24 rmoe joined #puppet-openstack
01:44 trad511 joined #puppet-openstack
02:20 xingchao_ joined #puppet-openstack
02:42 badiane_ka joined #puppet-openstack
02:49 ilbot3 joined #puppet-openstack
02:49 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
03:02 markvoelker joined #puppet-openstack
03:03 aimon joined #puppet-openstack
03:33 openstackgerrit Dulanjalie Ganegedara proposed stackforge/puppet-neutron: Adding portdb and fastpath_flood to n1kv.conf  https://review.openstack.org/155560
03:43 aimon joined #puppet-openstack
03:52 xingchao_ joined #puppet-openstack
04:04 markvoelker joined #puppet-openstack
04:05 openstackgerrit Dulanjalie Ganegedara proposed stackforge/puppet-neutron: Adding portdb and fastpath_flood to n1kv.conf  https://review.openstack.org/155560
04:05 xarses joined #puppet-openstack
04:30 xingchao joined #puppet-openstack
05:31 badiane_ka joined #puppet-openstack
05:33 badiane_ka joined #puppet-openstack
05:50 xingchao joined #puppet-openstack
05:52 holser joined #puppet-openstack
05:54 openstackgerrit Clayton O'Neill proposed stackforge/puppet-designate: Ensure designate user/group exist  https://review.openstack.org/155434
05:54 Hunner joined #puppet-openstack
05:54 Hunner joined #puppet-openstack
06:17 xingchao_ joined #puppet-openstack
06:33 openstackgerrit Spencer Krum proposed stackforge/puppet-nova: Fix beaker tests  https://review.openstack.org/155186
06:59 tfz joined #puppet-openstack
08:06 dgurtner joined #puppet-openstack
08:06 dgurtner joined #puppet-openstack
08:09 markvoelker joined #puppet-openstack
08:18 aimon joined #puppet-openstack
08:28 jpena joined #puppet-openstack
08:43 openstackgerrit joined #puppet-openstack
08:45 gildub joined #puppet-openstack
08:48 openstackgerrit Spencer Krum proposed stackforge/puppet-nova: Fix beaker tests  https://review.openstack.org/155186
09:10 markvoelker joined #puppet-openstack
09:15 IvanBerezovskiy joined #puppet-openstack
09:16 jistr joined #puppet-openstack
09:19 derekh joined #puppet-openstack
09:46 andriikolesnikov joined #puppet-openstack
09:47 xingchao joined #puppet-openstack
09:48 gildub joined #puppet-openstack
10:07 aimon joined #puppet-openstack
10:11 markvoelker joined #puppet-openstack
10:12 morazi joined #puppet-openstack
10:28 stamak joined #puppet-openstack
10:40 sbadia morning
10:42 sbadia dvorak: hi! (sorry for the delay…) nice for pdns \o indeed, designate module is still very bind9 centric… patch 155442 is perfect!
10:57 openstackgerrit Sebastien Badia proposed stackforge/puppet-ceilometer: spec: updates for puppet-rspec 2.x and rspec 3.x  https://review.openstack.org/155704
11:08 gildub joined #puppet-openstack
11:12 markvoelker joined #puppet-openstack
11:20 openstackgerrit Sebastien Badia proposed stackforge/puppet-ceilometer: spec: updates for rspec-puppet 2.x and rspec 3.x  https://review.openstack.org/155704
11:49 andriikolesnikov joined #puppet-openstack
11:56 aimon joined #puppet-openstack
12:13 markvoelker joined #puppet-openstack
12:25 EmilienM|afk good morning
12:44 xingchao joined #puppet-openstack
12:46 stamak joined #puppet-openstack
12:54 dgurtner joined #puppet-openstack
13:10 markvoelker joined #puppet-openstack
13:11 dprince joined #puppet-openstack
13:15 ddmitriev1 joined #puppet-openstack
13:16 openstackgerrit Merged stackforge/puppet-manila: Fix manila::service_instance type  https://review.openstack.org/149874
13:28 social sbadia: thanks
13:28 sbadia social: :-)
13:30 social sbadia: btw https://review.openstack.org/#/c/154545/
13:31 sbadia oh I didn't see
13:31 sbadia done!
13:32 social thank you
13:37 xingchao joined #puppet-openstack
13:39 mfischer joined #puppet-openstack
13:39 dvorak sbadia: I've also got a virtualenv patch in progress, but I'm not sure how that's going to turn out yet :)
13:40 mfischer What's going on with the Jenkins failures? Is ruby gems.org having issues?
13:49 spredzy dvorak, on your comment (modulesync), how would you make it work for a Github backend, where a user might explcilty want to '--amend' the last change ?
13:50 dvorak yeah, I was thinking about it after I made the comment, I'm not sure my comment actually makes sense, I just woke up and had a late night :)
13:51 spredzy dvorak, hehe np, just trying to match everyone expectations
13:57 dvorak spredzy: let me take another look
13:58 dvorak ok, I think I misunderstood why you were doing --amend.  The issue isn't just adding the change id, but that someone might make a change to the config file and then resubmit the changeset?
14:00 dvorak is branch_prefix supposed to be the magic gerrit submit prefix?
14:00 spredzy dvorak, correct. Let's say I update the .travis.yml file on every modue, someone -1, so I reupdate a new patchset everywhere using `msync update --amend`
14:00 spredzy new patchset is submited
14:00 dvorak ah, ok, and the pre-commit will put the gerrit hook in place so you don't have to worry about a missing change id at all
14:01 spredzy dvorak, yup, if you look at https://github.com/puppetlabs/modulesync/pull/38 (at the top) I put an example of the modulesync.yml file
14:01 spredzy see the branch_prefix param
14:02 dvorak I think it'd be good to put that in the commit itself, like as an update to the README
14:03 spredzy dvorak, +1
14:23 richm joined #puppet-openstack
14:26 openstackgerrit Merged stackforge/puppet-designate: Ensure designate user/group exist  https://review.openstack.org/155434
14:45 aimon joined #puppet-openstack
14:53 rcallawa joined #puppet-openstack
15:05 xingchao joined #puppet-openstack
15:15 openstackgerrit Merged stackforge/puppet-manila: Write share_helpers to config as CSV  https://review.openstack.org/137536
15:18 prad joined #puppet-openstack
15:19 andriikolesnikov joined #puppet-openstack
15:36 bitblt joined #puppet-openstack
15:36 ericpeterson joined #puppet-openstack
15:49 aimon joined #puppet-openstack
16:03 mschmitt joined #puppet-openstack
16:04 rcallawa_ joined #puppet-openstack
16:04 mdorman joined #puppet-openstack
16:09 xingchao joined #puppet-openstack
16:30 jpena joined #puppet-openstack
16:35 tfz joined #puppet-openstack
16:35 pradk joined #puppet-openstack
16:36 jpena joined #puppet-openstack
16:44 igajsin joined #puppet-openstack
16:58 vinsh joined #puppet-openstack
17:00 aimon joined #puppet-openstack
17:01 IvanBerezovskiy left #puppet-openstack
17:01 aimon joined #puppet-openstack
17:04 openstackgerrit Ryan Hefner proposed stackforge/puppet-manila: Change share server config details  https://review.openstack.org/137534
17:09 alop joined #puppet-openstack
17:28 rmoe joined #puppet-openstack
17:46 aimon joined #puppet-openstack
17:56 xarses joined #puppet-openstack
18:01 ericpete_ joined #puppet-openstack
18:20 dmsimard joined #puppet-openstack
18:35 EmilienM dachary: a first output of ceph module in tripleO with basic usecase: http://logs.openstack.org/60/151960/4/check-tripleo/check-tripleo-ironic-overcloud-f20puppet-nonha/ec3edbe/logs/ov-snjbym5mrmr-0-6mqbkeikgeju-Controller_logs/os-collect-config.txt.gz
18:43 dgurtner EmilienM: looks good! but what's the planning on varying the usecases? also how do you evaluate the output in an automated way?
18:43 EmilienM dgurtner: fyi, the patch is here: https://review.openstack.org/#/c/151960/
18:43 EmilienM and for now, there no output evaluation.
18:48 dachary EmilienM: good :-)
18:48 EmilienM dachary: OSD does not work yet
18:49 EmilienM but I'll figure it out
18:49 dachary I realize that but it's a great feeling that it's making progress.
18:57 mfisch When do I get my invite to the top secret core reviewers IRC channel?
19:01 EmilienM lol
19:01 EmilienM mfisch: what project was it finally?
19:01 mfisch I dont know who, I'm catching up in the ML
19:01 mfisch I'd guess neutron or nova
19:02 EmilienM I don't think so, their channel are open
19:02 mfisch the ones you can see are
19:02 dprince joined #puppet-openstack
19:03 mfisch otherwise it wouldnt be a good secret
19:03 britthouser joined #puppet-openstack
19:03 EmilienM mfisch: anyway, nothing in puppet is secrete
19:03 mfisch we can rebuild this channel with secret => true;
19:03 EmilienM ... lol
19:04 mfisch this thread is reminding me why I dont subscribe, it would be a full time job
19:04 britthouser joined #puppet-openstack
19:07 nibalizer we need someone to summarize 'this week on openstack-dev'
19:07 nibalizer news program style
19:07 nibalizer also, check it https://review.openstack.org/#/c/155186/
19:07 nibalizer don't mind the rechecks, that was hpcloud having bad networkign
19:07 EmilienM nibalizer: looking
19:08 EmilienM nibalizer: +2
19:08 EmilienM nibalizer: where do you track your work, can I contribute to this in other modules?
19:08 mfisch nibalizer: good idea
19:09 mfisch nibalizer: that thread about secret channels devolved into arguments about whether having a conference meant that openstack is not open source
19:09 EmilienM dprince: fyi, beaker works on puppet-nova
19:09 EmilienM crinkle: nibalizer do you think we should start to set this job as voting from now?
19:09 EmilienM I would vote for yes
19:10 nibalizer EmilienM: im doing keystone now
19:10 nibalizer i was messaging with dprince earlier today
19:11 nibalizer we need to move this onto kindof a delorean or delorean-thing i think
19:11 nibalizer because right now i think we're using the juno packages with the keystone puppet modules
19:11 nibalizer EmilienM: passing today doesn't mean it will be passing forever
19:11 nibalizer lets see how it evloves over the next little while
19:12 nibalizer also all thats being tested is if the puppet applied cleanly
19:12 nibalizer we should probably add 'post steps' to check that nova daemon is actually running etc etc
19:18 EmilienM nibalizer: yeah, like smoking tests?
19:18 EmilienM or service validation I did in some modules
19:19 nibalizer well yea just kinda poke it
19:19 nibalizer the examples directory is basically exactly what it has
19:19 nibalizer what we need
19:19 nibalizer in keystone it even has stuff thats like 'run this to make sure it all worked'
19:20 nibalizer so beaker is running right now :D
19:21 EmilienM nice!
19:32 xingchao joined #puppet-openstack
19:39 imcsk8 joined #puppet-openstack
20:01 openstackgerrit Spencer Krum proposed stackforge/puppet-keystone: Beaker tests  https://review.openstack.org/155873
20:07 openstackgerrit Dulanjalie Ganegedara proposed stackforge/puppet-neutron: Adding portdb and fastpath_flood to n1kv.conf  https://review.openstack.org/155560
20:20 richm question about keystone v3 and keystone_user_role
20:21 richm with v3, a user may have a role on a project, and on a domain
20:21 richm there is a command which very directly returns all role assignments - openstack role assignment list
20:22 richm however - it returns only IDs, no names - so for each ID returned, the name would need to be looked up
20:22 richm the alternative is looking through every user in every domain and doing role list --user $user --domain $domain
20:24 richm would it be better to call role assignment list, then do user show, project show, and domain show for each record returned?  Note that many of the users and projects, and all of the domains returned, will be returned multiple times, so they can be cached
20:25 richm I would think that in most deployments, there will be many, many users, and very few projects and domains
20:32 ericpeterson joined #puppet-openstack
20:33 lnxnut joined #puppet-openstack
20:37 richm I guess the question is - is it better to make a small number of requests with very large result sets, or a large number of requests with very small result sets?
20:38 openstackgerrit Dulanjalie Ganegedara proposed stackforge/puppet-neutron: Adding portdb and fastpath_flood to n1kv.conf  https://review.openstack.org/155560
20:41 lnxnut joined #puppet-openstack
20:43 lnxnut_ joined #puppet-openstack
20:43 xingchao joined #puppet-openstack
20:54 mgagne get out your calculator and compute the Big O :D
20:55 ericpeterson joined #puppet-openstack
21:00 richm mgagne: sure
21:00 mgagne well, we only have to consider both kind of installations
21:00 mgagne and maybe there will be one obvious answer
21:01 richm Another related question - when doing a keystone_user_role { "user@project" } - there is an autorequire on the user and on the project - which means puppet is having to lookup and/or create the keystone_user and keystone_tenant resource
21:01 dvorak richm: also, take into account that there are plenty of other places where no one has ever asked this question, which is why our control node puppet runs take an extra 2-3 minutes :)
21:01 dvorak so you're ahead by even wondering!
21:02 richm well - I can't just "do it the old way" because v3 is so different in this regard
21:02 mgagne dvorak: I do wonder but we went from 15m to 5m by fixing this caching logic once =)
21:02 dvorak I know, I was trying to be funny, apparnetly failing :)
21:02 richm so I can't just "do it the old way and we'll optimize it later" because the odl way won't work
21:02 richm old
21:02 crinkle richm: so right now with https://review.openstack.org/#/c/150200 it only has to do one call to user role list for each keystone_user_role resource
21:02 dvorak mgagne: right now we spend 2-3 minutes enumeratine all neutron_networks and neutron_subnets via neutron net-show an neutron subnet-show :)
21:02 crinkle would the domains thing mean adding iteration to that?
21:03 crinkle re: autorequire, it doesn't mean extra work, it just means there's an implicit require metaparam in the resource
21:03 mgagne crinkle: I think we would have to list projects/roles/users per domain too?
21:04 crinkle so there's not a --domain for user role list? since the new resource will only have one domain per resource, right?
21:04 crinkle or am i confused
21:05 mgagne AFAIK, (some correct me), users, projects and roles are scoped to domains
21:05 mgagne not sure you can list them for all domains at the same time
21:05 mgagne richm: ^ ?
21:05 richm crinkle: there is no user role list
21:05 richm there is role list and role assignment list
21:05 mgagne so it got renamed but does the same?
21:06 richm mgagne: in v3, all users and projects are scoped to (i.e. contained within) domains
21:06 richm sort of
21:06 mgagne =)
21:06 richm roles are _not_ scoped
21:06 mgagne we just started looking at domains / keystone v3
21:06 richm roles exist "outside" of domains
21:06 mgagne they do?
21:06 mgagne damn
21:06 richm yes
21:06 crinkle so roles = request('user role', 'list', nil, resource[:auth], ['--project', get_project, get_user]) isn't going to work with domains, basically?
21:07 richm crinkle: request('role', 'list', nil, resource[:auth], ['--project', get_project_id, '--user', get_user_id]) will work
21:07 richm but there is an additional query
21:08 richm because a user may have a role in a domain but not a project within that domain
21:09 richm also note that you have to use the _id_, not the name
21:09 richm because you may have multiple projects/users with the same name in multiple domains
21:10 richm and there is no way to tell role list that you mean "richm" in domain "foo" and project "admin" in domain "foo"
21:10 richm using ids everywhere disambiguates since the id is unique across all domains
21:11 alop left #puppet-openstack
21:14 crinkle I'm just confused by the statement "there is no user role list"
21:14 crinkle there is a command openstack user role list
21:14 richm not with --os-identity-api-version 3
21:15 crinkle okay, that was the gap
21:15 richm yeah, sorry, I should have explained that first
21:15 richm there are a couple of commands that are missing or new in v3
21:15 richm so instead of user role list $user --project $project
21:16 richm you use role list --user $user --project $project
21:17 mgagne richm: since projects/users are not long unique across domains, should we qualify them (with domains) with managing them with puppet?
21:18 richm yes
21:18 mgagne richm: are domain names unique?
21:18 richm that's my proposal
21:18 richm keystone_user { "admin::somedomain" } keystone_tenant { "services::somedomain" }
21:18 mgagne richm: for backward compatiblity purpose, can we assume the guy meant "default" if no domain is provided?
21:18 richm mgagne: yes
21:18 richm that's how I'm proposing it will work
21:18 mgagne richm: what about the @ notation? is is a valid character within username?
21:19 richm if you'd rather see some ruby code, I can put up another wip
21:19 richm mgagne: yes - @ is valid - however, for backwards compatibility, keystone_user_role must still accept "user@project"
21:19 richm afaik, no one is doing "user@domain.com@project"
21:19 mgagne good lord
21:20 mgagne user@project@domain@theworld? :D
21:20 richm ack
21:20 EmilienM omg
21:20 richm which is why it was decided to use "::" as the domain delimiter
21:20 mgagne what about nested projects? will domains just disappear after that? :O
21:20 richm I don't know
21:21 richm but with :: you could easily have nested domains
21:21 richm user::smalldomain::bigdomain@project::smalldomain::bigdomain
21:21 mgagne richm: I have more questions sir! answer! :O http://www.davidmullenasc.com/iwus1.jpg ;)
21:21 richm I'll never confess
21:21 mgagne haha
21:22 dgurtner joined #puppet-openstack
21:25 mgagne how does someone logging to horizon with domains?
21:25 mgagne user@domain ?
21:25 richm re: my query about internal autorequire evaluation - if keystone_user_role { "user@project" } has an autorequire on Keystone_user['user'] and Keystone_tenant['project'] - does that mean there will be a Keystone_user['user'] resource and a Keystone_tenant['project'] resource in the puppet catalog?
21:25 richm mgagne: ask ayoung
21:25 mgagne richm: no
21:25 mgagne richm: it will only create a relation if it exists
21:25 mgagne richm: otherwise nothing happens
21:26 mgagne richm: it's the same as implicit dependencies: http://ai.stanford.edu/~llao/puppet/puppet-trainning.html
21:28 richm mgagne: but I can't do keystone_user_role { "userdoesnotexist@project" } ?
21:28 mgagne richm: userdoesnotexist => does not exist in Puppet or Keystone? ;)
21:29 richm keystone
21:29 mgagne puppet won't create the user for you as of today anyway
21:29 mgagne it won't invent an email or password for you
21:29 ericpeterson joined #puppet-openstack
21:30 richm right - what I'm getting at is that there must be a keystone_user { "user" } that gets evaluated before keystone_user_role { "user@project" } is evaluated - where the keystone_user { "user" } will either create the user or ensure that it exists with the given properties
21:31 mgagne richm: if the user exists already in keystone (created outside puppet), puppet "should" be fine with that
21:31 richm right
21:31 mgagne unless there is some weird caching system with our providers
21:31 aimon joined #puppet-openstack
21:31 richm because what I would like to do is to get the resource for Keystone_user['user'] so I can get the id parameter
21:32 mgagne I see where you are getting at
21:32 richm rather than having to do a separate lookup to get the id for the user in keystone_user_role { 'user@domain' }
21:32 richm same with domain and Keystone_tenant['domain']
21:32 richm I mean, it's all there in the puppet ruby code lurking somewhere
21:33 mgagne crinkle: can't you create a "ghost" resource for the purpose explained by richm ?
21:33 crinkle mgagne: if you can I'm not sure how
21:33 mgagne richm: so you don't have to explicitly create/manage one if none is provided
21:34 crinkle I don't think a resource instance has the ability to look up other resources
21:34 crinkle maybe bodepd_ would know better
21:34 mgagne crinkle: oh there is surely a way to do it :D
21:34 mgagne and he will say: yes but don't :P
21:34 crinkle haha
21:34 ericpeterson joined #puppet-openstack
21:34 richm . . . which is what I would expect - don't mess with the puppet internals
21:35 mgagne crinkle: you can
21:35 mgagne https://github.com/stackforge/puppet-tempest/blob/master/lib/puppet/provider/tempest_glance_id_setter/ruby.rb
21:35 mgagne but the resource has to exist
21:35 mgagne https://github.com/stackforge/puppet-tempest/blob/master/lib/puppet/provider/tempest_glance_id_setter/ruby.rb#L16
21:35 richm yes
21:35 crinkle aha!
21:35 richm so they're doing the same thing - looking up the id of another resource
21:36 mgagne yes
21:36 mgagne because of Puppet's limitations =)
21:36 mgagne can't use Glance_image['whatever']['id'] in your manifest
21:37 mgagne and can't use function because (AFAIK), they are evaluated on the puppetmaster, not the node. and the puppetmaster might not have the same access (to the API) as the node.
21:37 aimon joined #puppet-openstack
21:37 mgagne but I could be mistaken
21:37 crinkle that's my understanding
21:39 mgagne what are you guys using to package your puppetmaster? lets say one of the restriction is "no git clone/pull"
21:44 lnxnut joined #puppet-openstack
21:55 openstackgerrit Spencer Krum proposed stackforge/puppet-keystone: Beaker tests  https://review.openstack.org/155873
21:55 xingchao joined #puppet-openstack
21:58 rcallawa joined #puppet-openstack
22:00 lnxnut joined #puppet-openstack
22:30 lnxnut joined #puppet-openstack
22:43 rcallawa joined #puppet-openstack
23:03 xingchao joined #puppet-openstack
23:07 ericpeterson joined #puppet-openstack
23:51 xarses joined #puppet-openstack
23:55 markvoelker joined #puppet-openstack

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