Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-10-09

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

All times shown according to UTC.

Time Nick Message
00:01 otherwiseguy joined #puppet-openstack
00:10 harsha joined #puppet-openstack
00:11 harsha mjblack_ so how do you delete duplicate endpoints?
00:11 harsha is it manual process or you can do this via puupet?
00:14 mgagne harsha: I do it manually :-/
00:14 mgagne harsha: when updating multiple urls, endpoints are duplicated
00:15 mgagne harsha: https://bugs.launchpad.net/p​uppet-keystone/+bug/1234480
00:16 harsha mgagne: can it be automated via puppet
00:17 mgagne harsha: the idea would be to fix the problem and not put duct tape on it =)
00:17 mgagne harsha: I don't think puppet-keystone can handle that situation well
00:20 dmsimard joined #puppet-openstack
00:20 harsha hmmm.. let's see when it is getting fixed :)
00:22 harsha mgagne: how many controller nodes you are running? is there any VIP involved for load balancing?
00:30 mgagne harsha: we have multiple controllers
00:30 mgagne harsha: a VIP is used on a set of load-balancers
00:31 mgagne harsha: the endpoints point to the VIP
00:33 harsha great!!! I have similar setup, i was trying to modify controller class and install only packages on other nodes ( except VIP node ), do you think i should just apply existing controller class on other nodes, then delete duplicate endpoints manually?
00:34 mgagne harsha: if you wish all controllers to "manage" the keystone endpoints, make sure they all have the same URLs (pointing to the VIP), otherwise you will have duplicates.
00:36 harsha yessi.. but each one of them will have separate id that duplicates right?
00:37 harsha something similar to this 143627d64c5e4d9eb406b0dc6a70f433
00:37 mgagne harsha: they shouldn't if keystone shares the same database (which you should do)
00:38 mgagne harsha: unless you provision them at the exact same same time: puppet agents will fight to create the endpoint
00:41 harsha hmmm.. i do see this when i run puppet agent on a newly added ccn node
00:41 harsha Notice: /Stage[main]/Glance::Keystone::Auth/Keys​tone_endpoint[testdc/glance]/public_url: public_url changed 'http://xxx.yyy.zzz.121:9292' to 'http://xxx.yyy.zzz.122:9292' Notice: /Stage[main]/Glance::Keystone::Auth/Keyst​one_endpoint[testdc/glance]/internal_url: internal_url changed 'http://xxx.yyy.zzz.121:9292' to 'http://xxx.yyy.zzz.122:9292' Notice: /Stage[main]/Glance::Keystone::Auth/​Keystone_endpoint[testdc/glance]/adm
00:41 xingchao joined #puppet-openstack
00:47 harsha mgagne any idea what's going on?
00:47 mgagne harsha: why has the ip changed?
00:47 mgagne harsha: it should be your VIP, not the node/controller IP
00:48 ari joined #puppet-openstack
00:48 harsha you are right.. let me check my hiera :)
00:48 mgagne harsha: otherwise your controllers will fight to set the endpoints with THEIR ips
00:55 harsha mgagne: openstack::keystone::public_address:  '%{vip_server}' openstack::keystone::admin_address:  '%{vip_server}' openstack::keystone::internal_address:  '%{vip_server}'
00:55 harsha i am pointing all 3 to vip
00:55 mgagne harsha: can you add a resource which can valid the value of vip_server?
00:56 mgagne harsha: make sure those values are not superseded by something else or the values is passed correctly to *::keystone::auth.
00:56 harsha vip_server is defined in site.pp
00:56 ryanycoleman joined #puppet-openstack
00:58 mgagne harsha: what if you provide openstack::keystone::glance_public_address too?
00:59 mgagne harsha: how is the value of vip_server determined in site.pp?
00:59 mgagne harsha: is it a fact?
00:59 harsha thanks.. trying
00:59 mgagne harsha: $::ipaddress ?
00:59 harsha yes
00:59 bodepd mgagne: you are so helpful :)
00:59 mgagne harsha: *facepalm*
01:00 mgagne harsha: do you have this line in site.pp? $vip_server = $::ipaddress
01:00 bodepd mgagne: when you have time, I would like to try to sell you on what I've been up to :)
01:01 mgagne bodepd: =)
01:01 bodepd mgagne: it's starting to come together
01:01 bodepd mgagne: I can now precompile all of the hiera stuff together with classes for debugging
01:02 mgagne bodepd: precompile? like a template engine over hiera?
01:02 bodepd mgagne: this is the result of that for the controller: https://gist.github.com/bodepd/6894492
01:02 bodepd mgagne: hiera is too slow and hard to debug
01:02 fandikurnia01 joined #puppet-openstack
01:02 bodepd mgagne: I wanted to just pre-compute every thing
01:03 bodepd mgagne: right now, this is a debugging utility
01:03 mgagne bodepd: this output, is it for debugging or is it fed to hiera?
01:03 bodepd mgagne: right now debugging.
01:03 mgagne bodepd: hmmm
01:03 bodepd mgagne: the selected keys will everntually be fed into a instance variable that hiera can use
01:03 bodepd mgagne: and I'll have a simple get_it_from_memory custom backend
01:03 bodepd mgagne: I have not completed that part yet
01:03 mgagne bodepd: so this is what "puppet" sees?
01:03 bodepd mgagne: no
01:04 bodepd this is just a command line tool that I wrote
01:04 bodepd I am thinking about it like this:
01:04 mgagne bodepd: I mean, this is what puppet would "see" if it had to lookup a parameter value?
01:04 openstackgerrit A change was merged to stackforge/puppet-horizon: Updated user and group in debian params.  https://review.openstack.org/50173
01:05 bodepd basically (when it is finished)
01:05 bodepd if the data-bindings looked for: nova::rabbit_userid
01:05 bodepd via hiera
01:05 bodepd I would have a memory_backend that looks like this:
01:06 bodepd mgagne: https://gist.github.com/bodepd/6894546
01:06 bodepd the node terminus would have already set up that hash, and the value would be: openstack_rabbit_user
01:07 bodepd so, the output I showed you is using the same methods that would be used to build that per_node_hash
01:07 bodepd except in this example, it says:
01:07 bodepd 1. get all classes
01:07 bodepd 2. compile all hiera data
01:07 bodepd 3. for each class, find the mathing hiera data, and add it as a parameter to the class
01:07 bodepd 4. print
01:08 bodepd the node_terminus (when I finish it)
01:08 bodepd will do somethign like:
01:08 bodepd 1. get all classes
01:08 bodepd 2. compile all data
01:08 bodepd 3. for those classes that are used for this node, add its data to a hash that hiera uses to lookup
01:08 bodepd 4. just return the classes and top scope parameters back to Puppet
01:09 bodepd step #3 is probably bad though, b/c it assumes that all classes used are provided in the node_terminus
01:09 * bodepd takes a deep breath and waits for questions :)
01:10 mgagne bodepd: node terminus would be a hiera data precompiler then?
01:10 bodepd right now, I'm making sure that it can use the facts from puppetdb
01:10 bodepd mgagne: yes!
01:11 bodepd mgagne: and command line tools can also do the pre-compile
01:11 bodepd (which is what I am starting with)
01:11 mgagne bodepd: innocence question: can't puppetdb do it already or does it lack back reference support?
01:11 mgagne innocent*
01:11 bodepd mgagne: PuppetDB cannot do it
01:11 mgagne bodepd: I mean
01:11 mgagne bodepd: db backend for hiera
01:11 bodepd mgagne: right now, you have to wait for the data-bindings to resolve individual lookups
01:12 bodepd mgagne: in theroy
01:12 mgagne bodepd: oh, the hiera lookup is done JIT?
01:12 bodepd mgagne: you could compile all hera data down to something ahead of time
01:12 bodepd mgagne: and have a custom backend where you retrieve that pre-compiled data
01:12 bodepd mgagne: but that would be seperate from puppdb
01:12 bodepd puppetdb
01:13 bodepd I heard this is how hiera-2 is going to work with the pre-compilles
01:13 mgagne bodepd: hmmm
01:13 bodepd but hiera-2 is a bit tricky to use
01:13 bodepd 3.3 only
01:13 bodepd no guarentees they won't change it before 4.0
01:13 mgagne bodepd: we use regex in site.pp (yes, evil site.pp), how would it knows which data to precompiled?
01:13 bodepd do you know how node_termini work?
01:14 mgagne bodepd: please disregard verb tense mistakes =)
01:14 bodepd ah, and hiera-2 is absoltely not backwards compat to hiera 1
01:14 mgagne bodepd: http://docs.puppetlabs.com/​guides/external_nodes.html ?
01:14 mgagne bodepd: backward compatibility is a banned word =) /jk
01:14 bodepd the only reason it would matter is for variables in scope
01:15 bodepd (if you use top scope or node scope variables to set the hierarchy)
01:15 bodepd you cannot do this. So I have a place in YAML to define globals
01:15 bodepd it is almost an external node
01:15 bodepd so, external nodes are implemented using what is called a node_terminus
01:15 bodepd in fact, configuring and external node is:
01:16 bodepd node_termius=exec
01:16 bodepd s/and/an/
01:16 bodepd this is a node terminus which is functionally the same as an ENC, except in process
01:16 mgagne bodepd: node terminus is the api/manager which knows how to get your node definition?
01:16 bodepd and you return a Puppet::Node
01:16 bodepd mgagne: yes
01:16 bodepd mgagne: it still uses site.pp
01:16 bodepd mgagne: but this is before
01:17 mgagne bodepd: but with an external node classifier, I guess I cannot expect regex to work
01:17 mgagne bodepd: unless it is implemented by the backend used
01:17 bodepd mgagne: https://github.com/bodepd/scenario_node_terminus/b​lob/master/lib/puppet/indirector/node/scenario.rb
01:18 bodepd mgagne: I could be convined to implement regex.
01:18 bodepd mgagne: right now the mapping of role -> certname is in yaml
01:18 bodepd (this will likely be handled by foreman or anotehr front end for our use case eventually)
01:18 mgagne bodepd: it is not that much a requirement if you can teach me an elegant way to support N nodes with the same purpose/role
01:19 bodepd mgagne: I could reasonably expand that to work with regex
01:19 bodepd mgagne: I'm looking for users/use-cases atm
01:20 bodepd mgagne: since it's just a YAML hash, you could easily programatically set the role_mappings
01:20 bodepd mgagne: but yeah, I can understand the motivation for supporting regex. I think we have the same requirement
01:21 bodepd mgagne: the docs for the node terminus need to be updated for sure :)
01:22 mgagne bodepd: until I find an external tool to manage puppet and node provisioning for me ;)
01:22 mgagne bodepd: now I won't have to worry about updating puppet when spawning a new node since it would handled by the same interface ;)
01:22 nibalizer dangit, i just want to hang out with you guys and do puppet and openstack, but like zero time for that now
01:22 nibalizer <3 bodepd
01:22 mgagne nibalizer: dude, I want to do cooler stuffs than puppet and I don't have time for that :P
01:23 nibalizer haha, its all relative
01:23 nibalizer what cooler things do you want to do?
01:23 bodepd mgagne: nibalizer me too :) I want Puppet to be a piece of my puzzle ;)
01:23 mgagne nibalizer: this ^
01:24 mgagne nibalizer: more architecture works in our openstack infra, automating other tasks related to openstack, new features in openstack, etc.
01:25 nibalizer mgagne: maybe this is because im not managing my stress levels well, but i dont feel like i have time to read whats above
01:25 mgagne nibalizer: we (our team) have more internal/infra backlog than our product owner ;)
01:25 nibalizer also we have bodepd up there talking about terminuses(termini?) which is like my signal that im gonna have to do research to figure out whats up
01:25 nibalizer are you guys on the same team?
01:26 mgagne nibalizer: we are only on the same channel ;)
01:26 bodepd nibalizer: if you can end a plural word with i, you gotta take advantage
01:26 nibalizer lulz yea
01:26 badiane_ka joined #puppet-openstack
01:26 bodepd nibalizer: I'm a few weeks away from being unemployed :)
01:27 nibalizer bodepd: how unfortunate for you :)
01:27 bodepd nibalizer: looking forward to it. Been way to long since my last vacation
01:27 bodepd goal to self: don't write code for 2 weeks :)
01:28 bodepd mgagne: the more interesting part of the work is that we are storing multiple openstack reference architectures in it for CD
01:29 bodepd mgagne: I need someone to add the default hiera stuff for linuxbridges ;)
01:30 bodepd mgagne: and not to go as far as saying you were right, but I am actively reviewing my decision to separate out the data mappings for recursive hiera lookups
01:30 bodepd mgagne: that was your LHS=RHS comment about the format
01:30 mgagne bodepd: ^^'
01:30 mgagne bodepd: what would you need regarding linuxbridge?
01:30 bodepd the hiera defaults for the hierarchy
01:30 bodepd I only have ovs
01:31 bodepd mgagne: https://github.com/CiscoSystems/ope​nstack-installer/blob/master/data/h​iera_data/network_plugin/ovs.yaml
01:31 bodepd stuff like that
01:31 bodepd and...
01:31 bodepd oh, maybe that is it
01:32 bodepd it is the file that sets network plugin specific defaults in the hierarchy
01:32 mgagne bodepd: shouldn't be hard ^^'
01:32 bodepd so the user can just select the plugin and not have to adjust the class param
01:32 bodepd the validation and testing is the hard part
01:34 bodepd my dream is that everyone can use this, and we can all encode our use cases in a reusable/sharable data layer
01:34 bodepd I'm going to talk to folks at the summit to see if it is a dream or a pipe dream :)
01:35 bodepd kind of wished I had known it was going to come together before the summit
01:35 bodepd should have submitted a talk about it :(
01:35 xingchao joined #puppet-openstack
01:47 ari joined #puppet-openstack
02:10 ryanycoleman joined #puppet-openstack
02:20 xingchao joined #puppet-openstack
03:18 digiclos1up joined #puppet-openstack
03:33 ari joined #puppet-openstack
03:52 dguitarbite joined #puppet-openstack
03:59 dguitarbite left #puppet-openstack
04:38 harsha joined #puppet-openstack
05:21 otherwiseguy joined #puppet-openstack
06:21 bauzas1 joined #puppet-openstack
06:46 fandikurnia01 joined #puppet-openstack
06:53 ikkeT joined #puppet-openstack
07:43 dansmith joined #puppet-openstack
07:52 mmagr joined #puppet-openstack
08:17 nightfly joined #puppet-openstack
08:18 derekh joined #puppet-openstack
09:33 mmagr joined #puppet-openstack
09:43 mmagr joined #puppet-openstack
11:58 dprince joined #puppet-openstack
12:07 morazi joined #puppet-openstack
13:07 fandikurnia01 joined #puppet-openstack
13:13 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-horizon: RBAC policy files are now in /etc/openstack-dashboard on osfamily=RedHat  https://review.openstack.org/50623
13:14 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-horizon: Policy file paths on osfamily=RedHat have changed  https://review.openstack.org/50623
13:15 pwiltsey joined #puppet-openstack
13:40 markvoelker joined #puppet-openstack
13:40 dafter joined #puppet-openstack
13:40 dafter joined #puppet-openstack
13:47 fandikurnia01 joined #puppet-openstack
14:04 mjblack joined #puppet-openstack
14:07 dmsimard joined #puppet-openstack
14:07 mjblack joined #puppet-openstack
14:48 otherwiseguy joined #puppet-openstack
15:02 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-neutron: Upstream changed default quota_driver  https://review.openstack.org/50666
15:03 prad joined #puppet-openstack
15:12 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-neutron: Upstream changed default quota_driver  https://review.openstack.org/50666
15:13 ryanycoleman joined #puppet-openstack
15:42 ari joined #puppet-openstack
15:59 ari joined #puppet-openstack
16:03 otherwiseguy joined #puppet-openstack
16:12 badiane_ka joined #puppet-openstack
16:29 hogepodge joined #puppet-openstack
16:37 dmsimard joined #puppet-openstack
16:39 dmsimard joined #puppet-openstack
16:43 hogepodge joined #puppet-openstack
16:43 xingchao joined #puppet-openstack
16:46 rcrit joined #puppet-openstack
16:50 ryanycoleman joined #puppet-openstack
17:01 badiane_ka joined #puppet-openstack
17:04 hogepodge ping any core members?
17:04 hogepodge Any opposition to me just approving the release notes and Modulefile changes in stable/grizzly?
17:05 hogepodge I also think it would be nice to get another +1 (if it's appropriate) on https://review.openstack.org/#/c/47530/
17:05 hogepodge I cut a RC to the Forge, and it's working fine with our tests.
17:21 openstackgerrit A change was merged to stackforge/puppet-neutron: Upstream changed default quota_driver  https://review.openstack.org/50666
17:22 openstackgerrit A change was merged to stackforge/puppet-horizon: Preparing for 2.2.0 release.  https://review.openstack.org/50440
17:22 openstackgerrit A change was merged to stackforge/puppet-keystone: Preparing for 2.2.0 release.  https://review.openstack.org/50480
17:22 openstackgerrit A change was merged to stackforge/puppet-cinder: Prepping module for 2.2.0 release.  https://review.openstack.org/50412
17:23 openstackgerrit A change was merged to stackforge/puppet-nova: Preparing for 2.2.0 release.  https://review.openstack.org/50443
17:23 openstackgerrit A change was merged to stackforge/puppet-swift: Preparing for 2.2.0 release.  https://review.openstack.org/50445
17:24 mgagne hogepodge: https://review.openstack.org/#/c/50442/
17:24 openstackgerrit A change was merged to stackforge/puppet-glance: Preparing for 2.2.0 release.  https://review.openstack.org/50436
17:25 tvb|afk joined #puppet-openstack
17:25 tvb|afk joined #puppet-openstack
17:26 hogepodge mgagne: Good catch on that. I'll update that.
17:27 tvb|afk joined #puppet-openstack
17:28 openstackgerrit A change was merged to stackforge/puppet-openstack: Preparing for 2.2.0 release.  https://review.openstack.org/50444
17:29 openstackgerrit Chris Hoge proposed a change to stackforge/puppet-neutron: Preparing quantum for 2.2.0 release  https://review.openstack.org/50442
17:31 hogepodge mgagne: it's good for them all to be on the same version.
17:32 openstackgerrit A change was merged to stackforge/puppet-neutron: Preparing quantum for 2.2.0 release  https://review.openstack.org/50442
17:32 hogepodge mgagne: thanks
17:32 mgagne np
17:33 otherwiseguy joined #puppet-openstack
17:38 dafter joined #puppet-openstack
17:38 dafter joined #puppet-openstack
17:40 tvb|afk joined #puppet-openstack
17:44 marun joined #puppet-openstack
17:48 marun joined #puppet-openstack
17:49 hogepodge joined #puppet-openstack
17:58 morazi joined #puppet-openstack
18:02 hogepodge joined #puppet-openstack
18:03 openstackgerrit Francesco Vollero proposed a change to stackforge/puppet-ceilometer: Added alarm-notifier and alarm-evaluator services  https://review.openstack.org/50699
18:04 hogepodge mgagne: You had approved this one before, but there was a change to it. https://review.openstack.org/#/c/47530/ Out of order on the release.
18:07 openstackgerrit Francesco Vollero proposed a change to stackforge/puppet-ceilometer: Added alarm-notifier and alarm-evaluator services  https://review.openstack.org/50699
18:24 ryanycoleman joined #puppet-openstack
18:40 badiane_ka joined #puppet-openstack
19:29 tnoor joined #puppet-openstack
19:32 badiane_ka joined #puppet-openstack
19:47 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-openstack: Add syslog support to the openstack module  https://review.openstack.org/45200
19:52 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-openstack: Add syslog support to the openstack module  https://review.openstack.org/47530
19:54 ryanycoleman joined #puppet-openstack
19:58 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-openstack: Add syslog support to the openstack module  https://review.openstack.org/47530
20:16 ari joined #puppet-openstack
20:16 openstackgerrit A change was merged to stackforge/puppet-horizon: Policy file paths on osfamily=RedHat have changed  https://review.openstack.org/50623
20:20 openstackgerrit John Griffith proposed a change to stackforge/puppet-cinder: Add solidfire manifest  https://review.openstack.org/50736
20:25 openstackgerrit John Griffith proposed a change to stackforge/puppet-cinder: Add solidfire manifest  https://review.openstack.org/50736
20:33 ryanycoleman joined #puppet-openstack
20:33 ari joined #puppet-openstack
20:34 hogepodge_ joined #puppet-openstack
20:34 harsha joined #puppet-openstack
20:39 ryanycol_ joined #puppet-openstack
20:39 exed joined #puppet-openstack
20:41 exed Moin, does somebody have a openstack neutron/quantum setup with multiple l3-agents? if so, do you where able to achieve this with the puppet-neutron module or is it handcrafted? :)
20:42 exed can't see any possible configuration with this class to achieve this, but maybe its just to late, that i can't see it.
20:43 EmilienM exed: what's your issue ? i run multiple neutron l3 agent nodes with puppet only
20:44 exed i was thinking about a possible solution to run multiple l3 agents with different configurations on the same host with the help of the puppet-neutron module but i found i answer now, sorry for the spam: https://github.com/stackforge/puppet-neut​ron/blob/master/manifests/agents/l3.pp#L5
20:44 exed thinking about a possible solution and hopefully will be able to provide a solution within a pull-request.
20:52 otherwiseguy joined #puppet-openstack
21:11 ryanycoleman joined #puppet-openstack
21:20 hogepodge joined #puppet-openstack
21:40 mjblack_ joined #puppet-openstack
22:05 dafter joined #puppet-openstack
23:04 bodepd exed what was the issue?
23:16 dafter joined #puppet-openstack
23:16 dafter joined #puppet-openstack

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