Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-09-03

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

All times shown according to UTC.

Time Nick Message
00:01 michchap joined #puppet-openstack
00:14 michchap joined #puppet-openstack
00:33 ari joined #puppet-openstack
00:45 otherwiseguy joined #puppet-openstack
02:36 ari joined #puppet-openstack
03:03 ari joined #puppet-openstack
03:50 marun joined #puppet-openstack
05:24 ikke-t joined #puppet-openstack
05:27 ikke-t joined #puppet-openstack
05:46 ikke-t joined #puppet-openstack
05:47 ikke-t joined #puppet-openstack
06:58 fvollero ikke-t: So funny your hostname :)
07:38 ikke-t fvollero, on top of openstack since my irssiproxy machine hardware melted :)
07:39 ikke-t and yes, that's a nice domain we have, along with the soft.ware.fi :)
07:40 EmilienM fvollero: /
07:40 ikke-t pfeeew, finally got rdo grizzly installed. next thing to get quantum working there, and then to migrate to stackforge puppet modules...
07:40 ikke-t i wonder what's the gap between rdo puppet modules and stackforge, haven't checked yet
07:40 ikke-t any experience?
07:57 fvollero EmilienM: Morning :)
07:57 fvollero ikke-t: Lol :)
07:58 fvollero ikke-t: What's the problem you got ?
08:04 fvollero EmilienM: All fine my friend? The shitstorm is over ? :)
08:04 EmilienM fvollero: never !!
08:05 fvollero EmilienM: Lol :) As far as you know, exist a ml for us ?
08:05 EmilienM fvollero: for puppet ?
08:05 fvollero EmilienM: yep, for infrastructure i mean
08:06 EmilienM fvollero: there is a lot of ML. do you mean openstack-infra ?
08:07 ikke-t fvollero, i set up quantum with CONFIG_NOVA_NETWORK_PUBIF=eth1 and CONFIG_QUANTUM_L3_EXT_BRIDGE=provider
08:07 ikke-t so i assume it should use the eth1 for floating ipe
08:08 ikke-t ips and snat
08:08 ikke-t but i can't see any of that in quantum
08:08 ikke-t i need to figure out how to create a router and floating ip pool on top of eth1.
08:08 ikke-t any ideas welcome :)
08:09 fvollero ikke-t: that's weird. Ping mmagr and he's one of the packstack author
08:09 ikke-t i'll ask the same on rdo
08:11 fvollero ikke-t: sure thing. on rdo you will find him :)
08:11 fvollero ikke-t: you mean on the forum ?
08:12 fvollero ikke-t: and can you publish the logs for sake of completeness?
08:13 mmagr joined #puppet-openstack
08:14 ikke-t fvollero, no i ment the irc, i prefer that :)
08:15 ikke-t but sure, i could also use the forum
08:15 ikke-t I don't know how the rdo packstack is supposed to handle the provider interface
08:16 ikke-t should it create the SNAT config for it automatically or should i do it manually, i need to google first a bit
08:21 derekh joined #puppet-openstack
10:58 michchap joined #puppet-openstack
12:22 dprince joined #puppet-openstack
12:29 morazi joined #puppet-openstack
13:20 michchap joined #puppet-openstack
13:23 openstackgerrit A change was merged to stackforge/puppet-glance: Change keystone tenant and user default  https://review.openstack.org/44573
13:24 dmsimard joined #puppet-openstack
13:34 mgagne joined #puppet-openstack
13:45 openstackgerrit Lucas Alvares Gomes proposed a change to stackforge/puppet-cinder: Check the parameter enabled before db_sync  https://review.openstack.org/44877
13:52 prad joined #puppet-openstack
14:10 openstackgerrit A change was merged to stackforge/puppet-glance: Add syslog support to the glance module  https://review.openstack.org/44003
14:12 openstackgerrit A change was merged to stackforge/puppet-keystone: Add syslog support to the keystone module  https://review.openstack.org/44004
14:14 openstackgerrit A change was merged to stackforge/puppet-nova: Add syslog support to the nova module  https://review.openstack.org/44005
14:28 xingchao joined #puppet-openstack
14:37 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-cinder: Add ability to configure control_exchange  https://review.openstack.org/44885
14:42 otherwiseguy joined #puppet-openstack
15:25 rcrit joined #puppet-openstack
15:28 michchap joined #puppet-openstack
15:44 ari__ joined #puppet-openstack
15:47 openstackgerrit A change was merged to stackforge/puppet-heat: Consolidate config into single file  https://review.openstack.org/43406
16:10 rcrit joined #puppet-openstack
16:24 rcrit joined #puppet-openstack
16:59 bodepd dprince: your upstream stdlib patch is waiting for a response
17:00 dprince bodepd: thanks. I'll have a look.
17:01 dachary joined #puppet-openstack
17:02 bodepd dprince: it looks like all you have to do is rebase the revisions and submit again!
17:03 dprince bodepd: yes. I'm seeing that.
17:06 britthouser joined #puppet-openstack
17:09 rcrit joined #puppet-openstack
17:17 openstackgerrit Dan Prince proposed a change to stackforge/puppet-ceilometer: Inline a custom file_line provider to fix agent.  https://review.openstack.org/44324
17:34 comptona joined #puppet-openstack
17:57 morazi joined #puppet-openstack
18:02 britthouser joined #puppet-openstack
18:17 rcrit joined #puppet-openstack
19:04 mmagr joined #puppet-openstack
19:34 rcrit joined #puppet-openstack
19:49 otherwiseguy joined #puppet-openstack
20:04 technolo-g joined #puppet-openstack
20:08 technolo-g hi guys, i was wondering how you folks recommend changing a nova setting that does not appear to be controlled by the puppet module. ie, i would like to set libvirt_images_type to lvm and specify the volume group in nova.conf
20:08 technolo-g Would augeas on that file be the best way?
20:12 technolo-g or perhaps add the parameter and contribute it back?
20:12 mgagne technolo-g: nova_config would be your best option
20:13 digicloseup joined #puppet-openstack
20:13 technolo-g mhmmm, ok great. thanks :)
20:13 mgagne technolo-g: could be an option too. please understand that nova-volume is going away soon so it will be of limited use.
20:13 mgagne technolo-g: of is it for local disks?
20:14 technolo-g yeah, its for local disks in order to use lv's instead of flat files for ephemeral storage
20:20 technolo-g thanks mgagne: nova_config works like a charm
20:51 rcrit joined #puppet-openstack
21:48 openstackgerrit joined #puppet-openstack
22:03 ari_ joined #puppet-openstack
22:20 rcrit joined #puppet-openstack
23:32 bodepd mgagne: I have been working at mapping how values are set to parameters
23:32 bodepd mgagne: or assigned. and as a result, I have a pretty good idea about class paramerer naming inconsistencies..
23:32 mgagne bodepd: nice, now you can't sleep anymore
23:33 bodepd mgagne: they need to be fixed, but are such a pain to do in a backwards compat way.
23:33 bodepd mgagne: no, I think that's you :)
23:33 mgagne bodepd: true, now I won't sleep anymore
23:33 bodepd mgagne: I'll create a blueprint soon to at least document them
23:33 mgagne bodepd: I was thinking about fixing that but can't find time to actually work on it
23:33 mgagne bodepd: was thinking about creating an helper
23:34 bodepd mgagne: no worries. I was more interested in your opinion about breaking backwards compat to fix them
23:34 mgagne bodepd: that would handle parameter deprecation
23:34 bodepd mgagne: we need a openstack-common module
23:34 mgagne bodepd: about that, this fictional module should contain ruby lib to interact with openstack api
23:34 bodepd mgagne: :)
23:35 mgagne bodepd: still trying to put aside how I hate some stuff in puppet =)
23:35 bodepd mgagne: This whole external YAML experiment has been really interesting
23:35 mgagne bodepd: nice to hear :D
23:35 bodepd mgagne: just about have it working, and it has highlighted a few things
23:35 bodepd 1. inconsistencies in param names
23:36 bodepd 2. inconsistencies in default behaviors acorss modules
23:36 mgagne bodepd: saw your bug about the 2nd
23:36 bodepd 3. that parameters should be more fine grained in the interfact
23:36 bodepd mgagne: yeah, I am seeing way too much too fix. so, I decided to start opening some bugs
23:37 mgagne bodepd: great idea
23:37 bodepd #3. sql_conneciton => bad, db_host + db_name + db_user, etc.. == good
23:37 mgagne bodepd: I also think that some configs should be splited from *::init
23:37 bodepd (at least in my opinion)
23:37 bodepd I'm open to that.
23:37 mgagne bodepd: don't know about that one as sql_connection allows more fine grain tuning
23:38 mgagne bodepd: or we end up with that mysql ssl patch
23:38 bodepd sql_connecton == "%{db_type}://%{cinder_db_user}:%{cinder_db_password}@%{db_host}/%{cinder_db_name}"
23:38 mgagne bodepd: but I'm open to suggestion
23:38 bodepd I think it's better. #1. b/c you can set individual defaults
23:39 mgagne bodepd: true
23:39 bodepd #2. b/c the parameter flow better represents how class parameters actually map to data values
23:39 bodepd (ie: db_host, db_type are re-used)
23:40 bodepd mgagne: I'm still playing with this yaml/hiera model
23:40 * mgagne fights oz to build fedora 19
23:40 bodepd mgagne: you are playing with oz?
23:40 mgagne bodepd: yes, I am
23:41 mgagne bodepd: haven't used/installed fedora in the past 5-6 years =)
23:42 bodepd I feel like i learned my lesson about 5-6 years ago.
23:42 bodepd not being a redhat hacker, I probably shouldn't :)
23:43 mgagne bodepd: we dropped Fedora because cPanel wasn't able to cope with the new accelerated release cycle of Fedora.
23:43 mgagne bodepd: has been centos since
23:43 mgagne bodepd: just trying to build fedora to "annoy" our netadmin
23:46 mgagne bodepd: anyway, I'm slowly trying to move to hiera instead of having a class with 457 parameters
23:47 bodepd mgagne: do you have interest in the data-mapping hiera backend?
23:47 bodepd mgagne: I'll have it all working in the next 2 days or so
23:48 mgagne bodepd: as you know, I'm interested in the ability to reuse value defined in hiera to build hiera values =)
23:48 bodepd mgagne: when you have a chance to look at it, I think either that, or the kickstack style composition layer that I was working on
23:48 bodepd mgagne: is the right answer
23:48 bodepd mgagne: I may wind up converting to a hiera style ENC
23:49 bodepd mgagne: I'm getting a little worried about the debuggability and performance of hirea
23:49 bodepd mgagne: but... first make it work, then make it fast
23:50 mgagne bodepd: I feel people should be able to compose their own manifests/roles with profiles and/or more basic components offered by our modules.
23:50 bodepd mgagne: yes. I agree. I was looking to model that in Puppet.
23:50 mgagne bodepd: puppet-openstack is opinionated on what should be installed on a controller and it does not fit the agility I (we) need.
23:51 bodepd mgagne: for the stuff I am working on atm, I will not be using puppet-openstack
23:51 bodepd mgagne: I will make sure it supports its currented supported confiugrations
23:51 mgagne bodepd: =) that's why I also think our modules need a redesign too
23:51 mgagne bodepd: not sure how but it feels monolitic
23:51 bodepd mgagne: really? I was generally happy with the core modules
23:52 bodepd mgagne: it's huge b/c openstack is huge
23:52 bodepd mgagne: but there should be more code re-use
23:52 bodepd mgagne: and strategies (like auto-doc generation)
23:52 bodepd to make it more maintainable
23:52 mgagne bodepd: just how an other service/component needs a package defined in init and can't define it itself or include the init class without throwing 45 errors for missing parameters.
23:53 bodepd in theroy, this is b/c it needs those parameters from init
23:53 bodepd but, you are correct, that the reality is that not all of the params are needed
23:53 bodepd I guess creating a rpc_client and db_client class for everything would make sense
23:55 mgagne bodepd: or move python-nova (for example) to a more specific class which can be reused by anyone requiring the python code without the configuration nova::init requires
23:55 mgagne bodepd: even the nova user. I can't require this user without installing the whole nova stack
23:59 mgagne bodepd: exploding classes in smaller ones should be easy, considering that most resources are defined without any particular customizable parameters

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