Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-09-17

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

All times shown according to UTC.

Time Nick Message
00:10 dmsimard joined #puppet-openstack
00:12 dmsimard1 joined #puppet-openstack
00:21 ari joined #puppet-openstack
00:32 lnxnut joined #puppet-openstack
00:34 dmsimard joined #puppet-openstack
00:42 lnxnut joined #puppet-openstack
00:45 marun joined #puppet-openstack
01:29 andi- joined #puppet-openstack
03:04 badiane_ka joined #puppet-openstack
03:15 otherwiseguy joined #puppet-openstack
03:17 michchap joined #puppet-openstack
03:21 michchap_ joined #puppet-openstack
03:38 michchap joined #puppet-openstack
03:56 dmsimard joined #puppet-openstack
04:21 lnxnut joined #puppet-openstack
06:08 blkperl_ joined #puppet-openstack
06:11 odyi joined #puppet-openstack
06:11 odyi joined #puppet-openstack
06:28 _ilbot joined #puppet-openstack
06:28 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
06:38 francois1 joined #puppet-openstack
07:18 mjeanson joined #puppet-openstack
07:58 mmagr joined #puppet-openstack
08:30 abk joined #puppet-openstack
09:04 rcrit joined #puppet-openstack
09:17 markvoelker joined #puppet-openstack
09:56 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-glance: Adapt to naming change in puppetlabs-mysql  https://review.openstack.org/46915
09:57 openstackgerrit Sandro Mathys proposed a change to stackforge/puppet-cinder: Adapt to naming change in puppetlabs-mysql  https://review.openstack.org/46916
12:28 morazi joined #puppet-openstack
12:44 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-horizon: Let apache module manage Listen directives  https://review.openstack.org/46936
13:16 prad joined #puppet-openstack
13:32 xingchao joined #puppet-openstack
13:48 prad joined #puppet-openstack
13:53 michchap joined #puppet-openstack
14:09 lnxnut joined #puppet-openstack
14:11 fvollero bodepd: ping
14:24 dmsimard joined #puppet-openstack
14:31 openstackgerrit A change was merged to stackforge/puppet-neutron: Rename parameter 'syslog_facility' into 'log_facility'  https://review.openstack.org/45418
14:34 otherwiseguy joined #puppet-openstack
14:41 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-horizon: Let apache module manage Listen directives  https://review.openstack.org/46936
14:58 badiane_ka joined #puppet-openstack
14:59 openstackgerrit Francois Deppierraz proposed a change to stackforge/puppet-openstack: Add syslog support to the openstack module  https://review.openstack.org/45200
14:59 technolo-g joined #puppet-openstack
15:14 ari_ joined #puppet-openstack
15:34 marun joined #puppet-openstack
15:39 abk joined #puppet-openstack
15:45 mmagr joined #puppet-openstack
15:59 elfelvin joined #puppet-openstack
16:00 elfelvin hi, im pretty new to puppet and openstack, but im just trying to do a default installation (controller and conpute nodes) but when installing on the controller node i get this error from puppet ..... Error: Failed to apply catalog: Could not find dependency Exec[mkdir ] for File[wsgi.conf] at /etc/puppet/modules/apache/manifests/mod/wsgi.pp:17
16:00 elfelvin it seems like a variable is not set as its [mkdir <space>]
16:01 elfelvin to install the modules i did "puppet module install puppetlabs/openstack" is this correct?
16:01 blkperl yep
16:02 elfelvin so then im missing something ... a variable or something?
16:02 blkperl yeah
16:03 elfelvin hmmm, im using the sites.pp file and i dont see anything about apache wsgi directories
16:03 elfelvin the test sites.pp file that is
16:04 marun joined #puppet-openstack
16:04 blkperl oh... I know what it is...
16:04 elfelvin is there anything i can do to get further debugging on this error?
16:04 elfelvin yea?
16:04 blkperl elfelvin: try downgrading the apache module one version
16:05 elfelvin âââ puppetlabs-apache (v0.9.0)
16:05 elfelvin thats my current version
16:06 technolo-g ├── puppetlabs-apache (v0.8.1)
16:06 technolo-g ^this is working for me
16:06 blkperl so your ${apache::mod_dir} is not set
16:06 blkperl elfelvin: what OS?
16:06 elfelvin ubuntu 12.04 LTS
16:07 blkperl ok please try apache 0.8.1
16:07 elfelvin ok, just working out how to downgrade
16:07 technolo-g i am on rhel, but iirc had to downgrade that module
16:07 blkperl technolo-g: did you file a bug ticket?
16:07 technolo-g i did not
16:08 technolo-g let me confirm im not talking snit
16:10 technolo-g yeah bug confirmed.
16:10 technolo-g i've never filed a bug before. lemme look around as to how its done
16:12 blkperl If you think its in the apache module you can do it on github, you can also file a bug on launchpad against the openstack modules
16:14 elfelvin okay, downgrading to 0.8.1 has got me past that error
16:15 technolo-g https://bugs.launchpad.net/puppet-openstack/+bug/1226687
16:15 technolo-g bug filed blkperl:
16:16 abk joined #puppet-openstack
16:17 blkperl thanks!
16:17 blkperl elfelvin: can you comment on the bug that it also affects you
16:17 technolo-g np. sorry i didnt do it sooner. ill keep it in mind to report instead of just workaround :)
16:17 blkperl technolo-g: np :)
16:54 rcrit joined #puppet-openstack
17:01 abk joined #puppet-openstack
17:21 bodepd blkperl: should we file something upstream about semantic versioning and backwards compat breakage?
17:24 blkperl bodepd: I haven't investigated why its broken, but I think its a bug and not an intentional backwards compat break
17:24 blkperl s/investigated/investigating
17:25 blkperl bodepd: I'm also hoping it wasn't my fault :)
17:25 blkperl since I added some more params to wsgi
17:25 rcrit joined #puppet-openstack
17:26 blkperl bodepd: are you going to join the data in modules google hangout at 11?
17:26 blkperl link will be posted in #puppet-dev
17:47 rcrit joined #puppet-openstack
17:53 bodepd fvollero: pong
17:53 bodepd blkperl: I could attend
17:53 bodepd blkperl: when was the release?
17:58 blkperl bodepd: 11 days ago
17:59 bodepd yeah. I created that 0.x branch, and assumed they would release a 1.0
18:00 bodepd b/c they broke backwards compat ages ago
18:00 blkperl bodepd: https://plus.google.com/hangouts/_/47515ed684be8edd2fd54a8994d872440dbd6752?authuser=1&amp;hl=en
18:00 blkperl bodepd: the 0.9.x branch wasn't suppose to break anything :(
18:04 rcrit joined #puppet-openstack
18:19 elfelvin another thing ive noticed is that when installing the compute node it seems to set the nova DB address as 127.0.0.1 ... i get this error Notice: /Stage[main]/Nova::Api/Exec[nova-db-sync]/returns: 2013-09-17 19:15:16.428 31210 CRITICAL nova [-] (OperationalError) (2003, "Can't connect to MySQL server on '127.0.0.1' (111)") None None
18:20 elfelvin meanwhile the mysql db is on another host, i dont see any variable to set this in site.pp, am i missing something?
18:27 elfelvin okay, i found that adding "db_host            => $controller_node_internal," to the openstack::compute class solves this issue
18:40 blkperl elfelvin: what version of puppet?
18:41 blkperl technolo-g: ^
18:41 hogepodge joined #puppet-openstack
18:53 technolo-g puppet-3.3.0-1.el6.noarch
18:53 technolo-g puppet-server-3.3.0-1.el6.noarch
18:59 ari joined #puppet-openstack
19:21 otherwiseguy joined #puppet-openstack
19:24 hogepodge I've been occupied with other items internally for a few weeks now, but I'm going to turn my attentions to a new release of the stable/grizzly branch for the remainder of the month.
19:25 hogepodge With the latest apache break I'm thinking of pinning all of the dependencies to known working versions, semver notwithstanding.
19:25 mgagne hogepodge: 2.2.0 ?
19:26 hogepodge I'll also patch up the puppet-quantum/puppetlabs-quantum mixup in the module file.
19:26 mgagne hogepodge: done already
19:26 mgagne hogepodge: https://bugs.launchpad.net/puppet-neutron/+bug/1223533
19:27 hogepodge Ah, good. Thanks.
19:28 mgagne hogepodge: creating milestones in LP
19:29 ari_ joined #puppet-openstack
19:32 hogepodge Thanks. I've been swamped here and turned my attention away. How much cherry picking do you think there needs to be done back to stable/grizzly?
19:33 hogepodge I want to make sure I avoid the mistakes I made last time.
19:34 hogepodge So please don't take any questions or statement out of context wrt positions from PL. I've seen enough bugs come through on the latest module release to the Forge that I want to get an updated version out.
19:34 mgagne hogepodge: we aren't tooled properly to make sure tracking. people are fixing stuff without opening bugs, making it hard to track bugs
19:35 mgagne + is done properly.
19:36 hogepodge That's good to know. I expected to do some manual cherry picking.
19:38 mgagne hogepodge: diff git log would be the best option...
19:38 Hunner joined #puppet-openstack
19:38 hogepodge We're looking at the Apache problem right now. (see Hunner who just joined the room)
19:39 Hunner technolo-g: The stuff in the apache module looks correct, so my bet would be that whatever is declaring the `apache` class and `apache::mod::wsgi` class are doing them in the reverse order
19:40 Hunner In the process of looking for where that was I saw that the swift module declares ripienaar-concat as a dependency; that should be updated to puppetlabs-concat as RI doesn't maintain it anymore so we took it up
19:40 technolo-g ahh ok cool. yeah i didnt really dig into it b/c the previous version worked fine for me.
19:40 Hunner (I haven't filed a launchpad bug yet)
19:41 Hunner technolo-g: The reason the previous version worked (guessing) is because the wsgi.conf file wasn't managed until 0.9.0. Still looking for where exactly this is happening
19:42 hogepodge mgagne: I have a multi-node grizzly installation module that I'm going to try and do some internal integration tests with. I'm hoping that will help (along with the automated gerrit builds) to catch any problems.
19:42 mgagne hogepodge: I'm using stable/grizzly in production. (but without puppetlabs-openstack)
19:42 technolo-g Hunner: line 88/89 in modules/horizon/manifests/init.pp perhaps?
19:43 hogepodge mgagne: With quantum?
19:43 mgagne hogepodge: yes
19:43 mgagne hogepodge: I expect people using or contributing to puppetlabs-openstack to do proper testing. (cross fingers)
19:43 hogepodge I treat puppetlabs-openstack more as an examples file than anything.
19:44 mgagne hogepodge: nova (libvirt/kvm) + cinder + glance + keystone + ceilometer + quantum (linuxbridge) + horizon
19:44 hogepodge We were running deployments before, just without quantum. My manifest uses OVS/GRE.
19:45 hogepodge There's been a pretty big knowledge gap here that we've been trying to fill since Dan left.
19:45 mgagne hogepodge: what I found out is that people sometimes contribute but don't test it properly until it's committed.
19:46 Hunner It's the "community-driven acceptance test model" :)
19:46 hogepodge Yup. It's why I haven't been pushing development through.
19:47 Hunner (not to say that we're working on beefing up testing frameworks to handle this stuff)
19:47 Hunner s/we're/we're not/
19:47 mgagne hogepodge: I'm stealing that one
19:53 hogepodge https://github.com/hogepodge/openstack-deploy/tree/master/external/grizzly
19:54 hogepodge Higher level, it depends on hieradata/common.yaml https://github.com/hogepodge/openstack-deploy/
19:54 hogepodge It's very opinionated. Not well documented yet. It's also missing some post rules on the network node to enable external nat.
19:55 mgagne hogepodge: tldr =(
19:56 mgagne hogepodge: would probably be the same if I had to document my setup =)
19:57 Hunner technolo-g: exactly. The base class should be declared before and sub classes or defines (so swap the order of those lines and that should fix it)
19:57 hogepodge One of my longer-term goals is to produce documentation to give users a better reference to start from.
20:01 Hunner technolo-g: Ticket updated.
20:04 technolo-g excellente :) I've got versions locked up right now as im trying to get this thing to prod, but next iteration it will come into play again
20:11 mgagne Hunner: isn't the order not important in Puppet?
20:12 mgagne Hunner: or is it for variable declaration that order is important?
20:18 ari joined #puppet-openstack
20:19 ari joined #puppet-openstack
20:24 Hunner mgagne: Parsing is in written order; application follows the dependency graph
20:24 mgagne Hunner: thanks
20:25 Hunner So yes, variables are part of parsing and order matters :)
20:26 Hunner It does allow the application ordering on the target machine to be decoupled from the organization of the manifest code though. So manifests can be rearranged without having to also think about how it affects the runtime application
20:32 blkperl Hunner: want me to submit a patch against horizon? then mgagne can review :)
20:39 openstackgerrit William Van Hevelingen proposed a change to stackforge/puppet-horizon: Fixes bug 1226687 apache module 0.9.0 breakage  https://review.openstack.org/47027
20:41 mgagne blkperl: do you mind wrapping your commit message and adding 'Fixes: bug #1226687' at the end?
20:47 blkperl mgagne: sure
20:47 mgagne blkperl: commenting
20:49 openstackgerrit William Van Hevelingen proposed a change to stackforge/puppet-horizon: Fix apache module 0.9.0 compatibility issue  https://review.openstack.org/47027
20:49 mgagne blkperl: tyvm
20:50 mgagne blkperl: and now bug gets updated =) https://bugs.launchpad.net/puppet-horizon/+bug/1226687
20:50 blkperl mgagne: yeah I was wondering about the syntax for that :)
20:50 mgagne blkperl: I also moved the bug to puppet-horizon
20:50 blkperl let me know when I should propose it to the grizzly branch :)
20:51 mgagne blkperl: https://wiki.openstack.org/wiki/StableBranch#Workflow
20:51 mgagne # checkout stable/grizzly
20:51 mgagne git cherry-pick -x XYZ
20:51 mgagne git review stable/grizzly
20:52 mgagne blkperl: that should be it
20:52 blkperl ok
20:52 mgagne blkperl: might want to create a local branch for that =)
20:52 blkperl k
20:53 openstackgerrit A change was merged to stackforge/puppet-horizon: Fix apache module 0.9.0 compatibility issue  https://review.openstack.org/47027
20:55 blkperl hmm why can't I check out stable/grizzly...
20:56 openstackgerrit William Van Hevelingen proposed a change to stackforge/puppet-horizon: Fix apache module 0.9.0 compatibility issue  https://review.openstack.org/47029
20:57 mgagne blkperl: I usually do: git checkout -t origin/stable/grizzly and then branch from there
20:58 blkperl woot learned some new git flags today :)
20:58 mgagne blkperl: =)
20:58 openstackgerrit A change was merged to stackforge/puppet-horizon: Fix apache module 0.9.0 compatibility issue  https://review.openstack.org/47029
21:16 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-cinder: Add ratelimiting parameters to cinder::api  https://review.openstack.org/46672
21:17 mgagne mainly fixing the patch =)
21:17 mgagne and tests
21:27 MxG joined #puppet-openstack
21:40 mgagne releasing milestones and closing bugs atm
21:45 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-cinder: Add ability to configure control_exchange  https://review.openstack.org/47036
21:46 mgagne hogepodge: who should handle the release of puppet-ceilometer?
21:51 ari joined #puppet-openstack
21:53 hogepodge mgagne: Good question. Releasing quantum was sketchy for me because we hadn't tested it. ceilometer would fall into the same boat.
21:53 ari joined #puppet-openstack
21:54 mgagne hogepodge: I do use ceilometer stable/grizzly in production. Obviously, I might not cover all use cases.
21:57 hogepodge We just talked it over here. We'll release it under the namespace and throw it into our todo pile for integration testing, realizing that it's being used and tested _somewhere_.
21:58 mgagne hogepodge: cool =)
21:59 mgagne hogepodge: how does puppetlabs feel about managing those releases? do they feel it's their role to do it? I feel the question becomes more pertinent when "3rd party" modules are involved. (like ceilometer and heat)
22:00 hogepodge We think it's very important. It's tricky, because at this point all of the module development is happening outside of PL and is being driven by interested 3rd parties.
22:02 hogepodge I was brought in as a person who knew Puppet, and a person who knew OpenStack to help address that. It's taken longer than I wanted to get up to speed, in part because I didn't know the modules, and I'm pretty ashamed about that.
22:02 mgagne hogepodge: what would be the best scenario/thing to do? Is PL interested in managing releases in the long term or would it be interested in more people being involved?
22:02 hogepodge There's been a lot of interest in the modules, though, and I want us to be more integrated in development.
22:03 hogepodge We want to release under our namespace, but I'm very open to help doing it right.
22:03 hogepodge I certainly don't want to repeat my mistakes from last time.
22:03 mgagne hogepodge: I was thinking about improving the release process
22:03 hogepodge I'm very open to suggestions.
22:04 mgagne hogepodge: I take for example this document: https://gist.github.com/alex/8881eb1662bedf03a3ba
22:04 hogepodge Moreover, I think that anyone in the community who wants to be involved, should. The development happens there, so there's no reason releases shouldn't happen there either.
22:05 mgagne hogepodge: more that the process could be simplified and automate to some level
22:06 hogepodge Yeah. The Forge team here is working on automated releases based on Git tags.
22:06 mgagne hogepodge: OpenStack already has such process for python and maven projects.
22:07 hogepodge We have automated Forge updates based on metadata information.
22:07 hogepodge I'm getting the docs right now.
22:09 hogepodge It's based on travis. I need to take some time to read over it, and I'll get back to you.
22:09 mgagne hogepodge: no rush, I don't have much time this week for it =)
22:11 mgagne hogepodge: question about hiera. can a value in hiera be referenced by an other value in hiera?
22:12 technolo-g +1 on that question. i have not found a way, but barely looked
22:12 mgagne hogepodge: like this: http://paste.openstack.org/show/47184/
22:12 mgagne hogepodge: in fact, I'll test it now.
22:13 mgagne hogepodge: it would be F**KING awesome if it works
22:13 technolo-g +1M
22:13 mgagne technolo-g: cleaning up my hiera file atm
22:14 * technolo-g ignoring his hiera file atm :)
22:14 technolo-g all the duplicate data is dirty
22:15 mgagne technolo-g: when more people get involved in puppet in your team, you are kind of forced to do things right otherwise you have a hard time explaining why values are duplicated or they have to update 3 layers of abstraction to pass the value. =)
22:17 * mgagne prays
22:17 technolo-g mgagne: iterative releases FTW :) At least the new tree im building is void of dead bodies and #TODOs. compared to our current prod tree, its a gem
22:18 mgagne technolo-g: hehehe working in the same way here ;)
22:19 mgagne technolo-g: saddest panda ever: http://paste.openstack.org/show/47185/
22:20 technolo-g mgagne: damn :(
22:20 technolo-g generate hiera files from a template that calls hiera data? O:-)
22:20 mgagne hogepodge: find someone to fix it! :O
22:21 mgagne technolo-g: =(
22:21 mgagne technolo-g: sed ?
22:25 badiane_ka joined #puppet-openstack
22:25 technolo-g mgagne: http://i.stack.imgur.com/PJw3h.png
22:25 mgagne technolo-g: :D
22:26 dmsimard joined #puppet-openstack
22:28 technolo-g mgagne: the talk at the puppet user group im going to tonight is about hiera. ill see if anyone is doing anything good
22:28 technolo-g slash troll them about dupe data lol
22:29 mgagne technolo-g: might be possible to hack something: http://docs.puppetlabs.com/hiera/1/variables.html#passing-variables-to-hiera
22:29 technolo-g mgagne: then the data is back into the manifests or site.pp. somewhat defeats the purpose imho
22:29 mgagne technolo-g: have a base class which would initialize a variable in puppet
22:30 mgagne technolo-g: but you would be limited to what your base class "support"
22:32 technolo-g mgagne: and the readability of the hiera is degraded significantly
22:32 mgagne technolo-g: =(
22:32 technolo-g i think hiera should support setting basic vars
22:32 technolo-g in hiera itself
22:32 technolo-g but i'm full of terrible opinions normally so...
22:33 mgagne technolo-g: basic variable support is already there =)
22:33 mgagne technolo-g: just not using hiera itself as a possible source when interpolating variables
22:34 hogepodge I'll throw it on the pile. Right now I just write wrapper modules that pass the relevant hiera data to all the places that need it.
22:35 mgagne hogepodge: I know that route, it's messy, see puppet-openstack =)
22:35 ari left #puppet-openstack
22:36 Hunner technolo-g: $datacenter = hiera('datacenter') ; $firewall_rules = hiera('firewall_rules') <-- the second query can use %{datacenter} in its hierarchy
22:36 Hunner I don't know if that's what you mean, but maybe
22:36 hogepodge I cleaned it up a bit by not allowing any class variables. Hiera is the only supported way to set parameters.
22:36 hogepodge For my deployment, at least.
22:36 mgagne technolo-g: http://paste.openstack.org/show/47186/
22:37 hogepodge Ok, home repairs in 20 minutes. see you tomorrow.
22:37 mgagne hogepodge: cya
22:38 technolo-g mgagne: hmmmm
22:43 otherwiseguy joined #puppet-openstack
22:45 technolo-g mgagne: that might be workable. perhaps in a profile. i assume it breaks if the parameter is not defined in yeah::base?
22:45 mgagne technolo-g: I think it could default to undef
23:15 MxG_ joined #puppet-openstack
23:24 michchap joined #puppet-openstack
23:25 michchap joined #puppet-openstack

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