Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-10-23

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

All times shown according to UTC.

Time Nick Message
00:09 rcallawa_ joined #puppet-openstack
00:36 openstackgerrit Tim Goddard proposed a change to stackforge/puppet-horizon: Add selection of Django session engine to this module  https://review.openstack.org/130403
00:45 xarses joined #puppet-openstack
00:47 rcallawa joined #puppet-openstack
01:26 topshare joined #puppet-openstack
01:47 ilbot3 joined #puppet-openstack
01:47 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
02:36 sputnik13 joined #puppet-openstack
02:54 topshare joined #puppet-openstack
03:01 sputnik13 joined #puppet-openstack
04:03 sputnik13 joined #puppet-openstack
04:03 thumpba_ joined #puppet-openstack
05:20 sputnik13 joined #puppet-openstack
06:10 christop1eraedo joined #puppet-openstack
06:10 dmsimardz joined #puppet-openstack
06:11 ingard__ joined #puppet-openstack
06:12 Madkiss joined #puppet-openstack
06:12 Madkiss joined #puppet-openstack
06:13 exed_ joined #puppet-openstack
06:15 trash joined #puppet-openstack
06:19 saneax_-_ joined #puppet-openstack
06:23 xingchao joined #puppet-openstack
06:44 dmsimard_away joined #puppet-openstack
06:52 dilyin joined #puppet-openstack
06:56 ianw joined #puppet-openstack
06:59 beddari_ joined #puppet-openstack
07:00 igajsin2 joined #puppet-openstack
07:08 saneax_-_ joined #puppet-openstack
07:08 arnaud_orange joined #puppet-openstack
07:11 xarses joined #puppet-openstack
07:18 saneax_-_ joined #puppet-openstack
07:30 rwsu_ joined #puppet-openstack
07:30 mmagr joined #puppet-openstack
07:33 saneax_-_ joined #puppet-openstack
07:52 fc__ joined #puppet-openstack
08:00 qba73 joined #puppet-openstack
08:19 exed_ joined #puppet-openstack
08:19 rushiagr_away joined #puppet-openstack
08:35 exed_ joined #puppet-openstack
08:35 rushiagr_away joined #puppet-openstack
08:51 derekh joined #puppet-openstack
09:04 dilyin joined #puppet-openstack
09:49 saneax_-_ joined #puppet-openstack
09:52 exed_ joined #puppet-openstack
09:53 rushiagr_away joined #puppet-openstack
10:25 mmaglana joined #puppet-openstack
10:39 openstackgerrit Sergii Golovatiuk proposed a change to stackforge/puppet-nova: Set force_snat_rang parameter  https://review.openstack.org/130488
11:11 dgurtner joined #puppet-openstack
11:38 EmilienM good morning everyone!
12:04 dgurtner joined #puppet-openstack
12:26 mmaglana joined #puppet-openstack
13:04 rcallawa joined #puppet-openstack
13:20 richm joined #puppet-openstack
13:26 mschmitt joined #puppet-openstack
13:27 mmaglana joined #puppet-openstack
13:44 morazi joined #puppet-openstack
13:47 rcallawa_ joined #puppet-openstack
14:04 holser Morning EmilienM!
14:05 openstackgerrit Sergii Golovatiuk proposed a change to stackforge/puppet-nova: Set force_snat_range parameter  https://review.openstack.org/130488
14:05 holser concerning, I believe it should be in nova_compute
14:05 holser EmilienM, ^^^
14:06 EmilienM holser: will look
14:09 topshare joined #puppet-openstack
14:10 topshare joined #puppet-openstack
14:13 michchap holser: EmilienM what about making a new class for compute/network.pp
14:14 holser michchap: I am with you ;)
14:14 EmilienM michchap: to be honest, nova-network is kind of far from my priorities right now
14:14 EmilienM I even wonder who is using it
14:15 topshare joined #puppet-openstack
14:16 EmilienM also I would like force_snat_range to false by default
14:16 holser I can change to false, that’s not a problem
14:16 EmilienM holser: lgtm otherwise
14:17 EmilienM michchap: really?
14:17 EmilienM michchap: kind of heavy
14:18 michchap EmilienM: yeah, but I have been thinking for a while we need to split more
14:18 michchap EmilienM: like...module/daemon/config-section.pp
14:19 michchap EmilienM: if someone is not using nova-network, do they want to be presented with parameters for it in the compute class? I don't think it's very good. Maybe if the vast majority of installs used it, but otherwise it seems like something we want to section off
14:23 holser michchap, EmilienM the best way is to create compute/network.pp where we can specify setting for compute nodes
14:23 xingchao joined #puppet-openstack
14:24 holser though if I change it to false, will it meet your expectations?
14:24 michchap holser: do you think there will be more?
14:24 michchap holser: are you sure this is correct? force_snatJ_range is a list according to the config reference for trunk
14:25 xingchao joined #puppet-openstack
14:25 michchap holser: "Traffic to this range will always be snatted to the fallback ip, even if it would normally be bridged out of the node. Can be specified multiple times."
14:28 mmaglana joined #puppet-openstack
14:34 holser michchap: I don’t think there will be more changes.
14:34 holser michchap: I think you are right, according to this
14:34 holser https://review.openstack.org/#/c/94112/17/nova/network/linux_net.py
14:35 holser you may have more than one SNAT rule
14:40 michchap holser: I mean, are there more parameters in nova.conf specifically for use with nova-network on compute nodes?
14:40 holser michchap, nope
14:41 holser only this one, we have 4.1.0 manifests and they work pretty fine with nova-network
14:41 michchap holser: then it's probably not worth it making a new class.
14:41 holser though, Juno broke nova-network
14:42 holser I’ll test with Array of parameters
14:42 michchap holser: although compute::neutron has only one parameter as well
14:42 holser yes
14:42 holser it’s like a switches
14:42 holser switcher*
14:43 michchap juno broke nova-network?
14:43 holser I can create compute::nova-network with one parameter ;)
14:43 vb joined #puppet-openstack
14:43 holser michchap, yes
14:43 michchap holser: just network. dashes break things.
14:43 vb Hi guys
14:43 holser yep
14:43 holser michchap, do you want me to move to compute::network ?
14:44 michchap holser: I think so. given that compute::neutron is only a single parameter and it's kind of the same thing - only relevant when doing that kind of networking.
14:44 michchap EmilienM: what do you think? ^^
14:44 michchap EmilienM: is that reasonable? or just chuck it in with everything esle?
14:45 vb I am looking into dependency cycle generated by packstack. Packstack wants  to have vs_bridge that requires neutron-ovs-plugin-service. And it worked grand until bugfix https://bugs.launchpad.net/puppet-neutron/+bug/1298146
14:46 vb now the latter says vs_bridge is a prerequisite for neutron-ovs-plugin-service
14:47 vb michap was the reviewer :)
14:47 vb sorry, michchap :)
14:47 michchap I don't know why packstack would have the dependency around the other way. The bridge should be up before neutron plugin starts.
14:48 vb yes, i have no objections, just trying to fix packstack to use the latest openstack puppet
14:49 ericpeterson joined #puppet-openstack
14:49 vb https://github.com/stackforge/packstack/blob/master/packstack/puppet/templates/neutron_ovs_bridge.pp
14:50 vb ^^^ that's what I mean
14:50 EmilienM michchap: compute::network is confusing to me, but I don't mind
14:51 michchap vb: I'd file a bug against packstack for that.
14:51 michchap EmilienM: prefer compute::nova_network?
14:51 EmilienM michchap: why not
14:51 michchap EmilienM: I'm fine with that.
14:52 mdorman joined #puppet-openstack
14:53 michchap vb: there might be something I'm missing that made it that way, but there was a very similar error in the manifests late last year
14:53 vb michchap: ok, thanks, I believe they won't fix that error because ovs plugin is obsolete since havana
14:54 vb and with ml2 it just works
14:54 michchap vb: that's odd.
14:58 openstackgerrit A change was merged to stackforge/puppet-tempest: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124584
14:59 mfisch I'd like to get an opinion on something. We have some Keystone settings which puppet doesn't know about or care about that have been deprecated in J and will be removed in K (according to log messages). What's everyone's opinion on adding these to the puppet module and adding the right default?
14:59 vb michchap: yes, it's quite a mess :) Anyway, thanks
15:00 mfisch Other than updating the default I have no need to actually set these to anything.
15:00 mfisch mdorman: ^
15:00 mfisch the [revoke] driver is one example
15:01 michchap mfisch: so the packages have the wrong default and that's spitting deprecation errors?
15:01 michchap mfisch: or where do the settings come from if not from puppet?
15:01 mfisch they come from the old packages
15:01 mfisch and the config file is carried forward
15:02 mfisch puppet's not using a template that I could just update...
15:02 michchap mfisch: I knew you were going to say that just after I pressed enter, heh.
15:02 mfisch np ;)_
15:03 michchap mfisch: you could use the keystone::config class to set them to absent so they get the default
15:04 pradk joined #puppet-openstack
15:04 michchap mfisch: or perhaps having an entire class dedicated to deprecated parameters so that people can purge their config when they update?
15:05 mfisch similar to config something like deprecated_config?
15:05 mfisch similar function
15:07 michchap mfisch: More that parameters in that class are deprecated for that release of openstack, or obsolete, and it will ensure => absent on them. So you can include that class and it will clean your config file if you're carrying it forward. Another option is to use the same thing as in the glance module that purges unmanaged config.
15:08 michchap mfisch: https://github.com/stackforge/puppet-glance/blob/master/manifests/api.pp#L422
15:09 mfisch I'm thinking that the config class will meet the needs and wont require updates every cycle
15:11 mfisch let me look at this more after my team call
15:12 michchap mfisch: I vaguely remember adding the purge option to glance after similar issues.
15:12 mfisch https://github.com/stackforge/puppet-glance/commit/c629376782efbc55bcf882543fa5801ddf5f1cfb
15:13 michchap mfisch: ah. different situation I guess.
15:16 mfisch with keystone::config I can just do 'revoke/foo' => { purge => true }
15:17 EmilienM sbadia: and debian ? https://review.openstack.org/#/c/124574/
15:18 sbadia EmilienM: puppet-designate doesn't support debian for the moment
15:19 EmilienM w00t !
15:19 sbadia I've a couple of patch to add debian support
15:19 sbadia but still in testing
15:20 sbadia EmilienM: and since the migration to openstacklib for mysql backend the catalog fail also on rh :D
15:20 EmilienM damn
15:20 sbadia mysql binding are now in binding class
15:20 sbadia https://github.com/stackforge/puppet-designate/blob/master/manifests/db.pp#L20
15:20 sbadia EmilienM: I'm on it ;)
15:21 EmilienM sbadia: I wonder if someone is using this module
15:21 sbadia =)
15:24 mdorman mfisch:  i’m late to the party here
15:24 mdorman catching up
15:29 mmaglana joined #puppet-openstack
15:32 mdorman i think i’d be in favor of the purge => true procedure.  that takes care of the depreciated stuff, and as an added bonus will bring any needed, but unmanaged, settings into the light
15:32 mdorman and that way we don’t have to deal with specifically tracking all the depreciated parameters
15:32 mfisch after some digging it looks like even keystone doesn't have the right default
15:33 mfisch so it needs to be fixed there, they're warning against themselves
15:33 mfisch there's at least 1 more parameter that fits this issue though, I will see if its intentionally set
15:37 cjellick joined #puppet-openstack
15:38 mfisch fyi - https://bugs.launchpad.net/keystone/+bug/1384775
15:41 EmilienM mfisch: nice one
15:43 mdorman good times
15:44 mfisch the other one I have is the same situation
15:46 EmilienM mfisch: reviewing your patch about python-ldappol, not sure it's in RHEL or Fedora
15:47 mfisch it has to be in something if they want ldap to work
15:47 mfisch or we could bug against keystone to not load the module until its enabled
15:47 EmilienM mfisch: gotcha !
15:48 EmilienM mfisch: I found it
15:48 EmilienM mfisch: i was just wondering about the packaging naming
15:48 EmilienM if it's the same in debian/fedora
15:48 EmilienM and yes it is.
15:48 EmilienM +2
15:48 EmilienM oops except there is no test
15:48 EmilienM mfisch: ^
15:50 xarses joined #puppet-openstack
15:51 mfisch what, tests? I'll do it just for you EmilienM
15:51 EmilienM mfisch: lol
15:51 EmilienM mfisch: thanks :P
15:51 mfisch here's my other deprecation: https://bugs.launchpad.net/keystone/+bug/1384789
15:52 EmilienM mfisch: I'm sure you'll take care of all those patches :P
15:52 EmilienM mfisch: let me know, I'll review them quickly
15:52 mfisch I'd be happy to patch keystone, but I suspect there's a reason they didnt switch the default
15:52 mfisch working on tests now mon ami
15:52 EmilienM mfisch: puppet-keystone
15:52 EmilienM if oyu do we need?
15:53 mfisch ???
15:53 EmilienM mfisch: our parameters are good, right?
15:53 mfisch yeah, puppet doesn't do anything with that stuff
15:53 mfisch the latter is in keystone-paste.ini
15:53 EmilienM kk
15:54 EmilienM mfisch: I can see you commit a lot in ldap things. Do you run keystone with pure openldap?
15:54 mfisch no
15:54 EmilienM mfisch: AD?
15:54 mfisch we use a hybrid-auth mechanism between SQL and AD via LDAP
15:54 EmilienM ok
15:55 mfisch for auth, you login against sql first, if it fails it tries LDAP
15:55 mfisch roles, tokens, etc are in sql
15:55 EmilienM ok, very nice
15:55 mfisch LDAP is really bind-only
15:55 EmilienM mfisch: how to you manage token storage at scale?
15:55 EmilienM do you replicate token mysql table or not?
15:55 mfisch yes
15:56 mfisch but more just by default and as we grow I suspect we may reconsider that
15:56 mfisch we do use a global identity for DR purposes
15:56 EmilienM I see
15:57 mfisch dvorak also setup scheduled token purging for us
15:57 mfisch EmilienM: review updated
15:57 EmilienM mfisch: cool, I'll +2
15:57 rcallawa joined #puppet-openstack
15:58 mfisch EmilienM: wait a sec... it didnt go
15:58 mfisch EmilienM: oops... I need to fix what I just did :(
15:59 rcallawa joined #puppet-openstack
16:00 EmilienM mgagne: wdyt about https://bugs.launchpad.net/puppet-neutron/+bug/1384119 ?
16:00 mfisch EmilienM: added, and you'll need to re +2 my other LDAP review since I accidentally added the test there
16:01 EmilienM mfisch: lol ok
16:01 mgagne EmilienM: https://review.openstack.org/#/c/130164/ ?
16:02 EmilienM mgagne: yeah
16:02 EmilienM do I miss something?
16:05 mgagne EmilienM: it's me that should ask this question =)
16:06 EmilienM mgagne: I'm not sure about this dep
16:06 EmilienM I don't like running db-sync easily like this
16:06 mgagne me neither, not much luck with neutron dn sync lately
16:08 mgagne EmilienM: maybe ask about which configs is required before running db sync
16:09 mgagne EmilienM: could be core_plugin, service_providers, etc.
16:09 EmilienM mfisch: what is the IRC of Clayton ?
16:09 mgagne EmilienM: not all of them
16:09 EmilienM mgagne: yeah
16:09 mmaglana joined #puppet-openstack
16:09 mgagne dvorak
16:10 mgagne michchap: ^ dvorak
16:10 mgagne michchap: https://launchpad.net/~clayton-oneill
16:11 EmilienM thanks!
16:11 mgagne michchap: wrong guy :P
16:11 EmilienM lol
16:12 EmilienM dvorak: I've reviewed your patch about nova_admin_tenant_id_setter, I have a question for you inline
16:18 mfisch sorry EmilienM was busy, but yes, its dvorak, he's probably at lunch now
16:18 EmilienM mfisch: np
16:18 rmoe joined #puppet-openstack
16:43 sputnik13 joined #puppet-openstack
16:46 ibravo joined #puppet-openstack
16:49 hogepodge crinkle on rhel 7 I use this line 'ln -s /dev/null /etc/udev/rules.d/80-net-name-slot.rules' from this page https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/Networking_Guide/sec-Disabling_Consistent_Network_Device_Naming.html to get networking to work properly in vagrant
16:50 openstackgerrit joined #puppet-openstack
16:51 crinkle why would you disable consistent network device naming? that seems backwards to me
16:51 crinkle also I think I tried that, maybe should have looked at it more
16:59 sputnik13 joined #puppet-openstack
17:01 sputnik13 joined #puppet-openstack
17:07 namredips joined #puppet-openstack
17:24 vb joined #puppet-openstack
17:34 EmilienM Hunner: crinkle: do you mind to review https://review.openstack.org/#/c/102020/ if you are familiar with acceptance?
17:35 crinkle EmilienM: i want to take a bit to play with it
17:35 EmilienM make sense
17:50 openstackgerrit joined #puppet-openstack
18:03 Mosibi joined #puppet-openstack
18:10 Mosibi joined #puppet-openstack
18:36 hogepodge crinkle Vagrant expects the old style of naming with eth0, eth1, etc.
18:36 hogepodge It won't set up networks properly without them.
18:38 hogepodge crinkle consistent naming is consistent within the box, but not outside of the box. It was meant to address the kernel loading different devices in a non-deterministic order (what might happen if you swap out a bad nic for a good one)
18:40 crinkle I ended up fixing it by modifying some grub configs and network scripts to use eth0
18:42 hogepodge crinkle that's how I started, but the other solution is much simpler and doesn't depend upon kernel parameters.
18:42 crinkle welp
18:42 hogepodge This is my latest centos box https://dl.dropboxusercontent.com/u/48788789/centos-7-64-openstack4.vmware.box
19:06 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Add ability to configure policy_files_path and policy_files  https://review.openstack.org/130621
19:14 openstackgerrit Nicolas Hicher proposed a change to stackforge/puppet-cinder: Create type-key only if it doesn't exist  https://review.openstack.org/112955
19:17 openstackgerrit A change was merged to stackforge/puppet-horizon: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124577
19:18 openstackgerrit A change was merged to stackforge/puppet-ceilometer: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124572
19:20 openstackgerrit A change was merged to stackforge/puppet-trove: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124585
19:22 openstackgerrit A change was merged to stackforge/puppet-swift: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124583
19:22 openstackgerrit A change was merged to stackforge/puppet-ironic: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124578
19:23 crinkle sbadia: I +2'd or commented on the metadata changes
19:23 crinkle are we going to do vswitch, openstacklib, and openstack_extras too?
19:24 openstackgerrit A change was merged to stackforge/puppet-nova: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124581
19:24 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Add selection of Django session engine to this module  https://review.openstack.org/130403
19:24 openstackgerrit A change was merged to stackforge/puppet-heat: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124576
19:25 openstackgerrit A change was merged to stackforge/puppet-glance: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124575
19:26 openstackgerrit A change was merged to stackforge/puppet-cinder: Switch to metadata.json for Puppet forge releases  https://review.openstack.org/124573
19:26 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Always manage local_settings.py  https://review.openstack.org/129731
19:26 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Use concat to manage local_settings.py  https://review.openstack.org/129743
19:27 mgagne crinkle, EmilienM: thanks for taking care of metadata.json reviews
19:28 crinkle finally got around to testing them
19:28 mgagne crinkle: cool, I don't know what to look for and how to test it. I leave it to knowledgeable people
19:29 crinkle puppetlabs has a staging forge that I can test them against
19:29 crinkle makes it easy for me
19:29 mgagne nice
19:31 imcsk8 joined #puppet-openstack
19:42 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Use concat to manage local_settings.py  https://review.openstack.org/129743
19:44 crinkle so I was playing with juno and finding that https://review.openstack.org/#/c/125231 was causing problems for me, since there isn't a way to set DEFAULT/os_region_name in the module except by using glance::backend::cinder
19:45 crinkle did DEFAULT/os_region_name used to have a default value in icehouse?
19:45 crinkle would it be better to add that as a class paramter or make it optional in the provider?
19:53 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-horizon: Add ability to configure simple HORIZON_CONFIG entries  https://review.openstack.org/129724
20:04 rcallawa_ joined #puppet-openstack
20:08 JayJ joined #puppet-openstack
20:12 EmilienM mgagne: crinkle: spredzy and I are thinking about a blueprint to write firewall code in our modules, using puppetlabs-firewall. Of course, that would be optional and flexible, like policy thing. What do you think? Does the code should go in openstacklib or extra? I think extra but not sure though
20:14 mgagne EmilienM: I believe puppet-nova should ship the strict minimum to manage an nova installation, everything else are extras: firewall, ha, monitoring, etc.
20:15 EmilienM mgagne: and lib is oslo for our modules
20:15 EmilienM right?
20:15 mgagne EmilienM: otherwise it opens the door to bundle more and more
20:15 EmilienM yeah
20:15 mgagne EmilienM: that's my opinion on it
20:15 spredzy to second EmilienM, I was seeing the implementation pretty much the sameway we did with policy
20:15 EmilienM so we will push a BP to put it in extra
20:15 spredzy so with an openstacklib::firewall or whatever the name is
20:15 EmilienM spredzy: policy is in lib, we will put it in extra, since it's ouside OpenStack
20:15 EmilienM outsite*
20:15 mgagne spredzy: policy is part of the core openstack service features, not firewalling
20:15 EmilienM damn
20:16 spredzy ah right, well in extra, but with the same paradigm in mind
20:16 mgagne spredzy: I wouldn't add $manage_firewall to nova::api for example
20:17 mgagne spredzy: this composition is better left for a composition module like puppet-openstack of puppet-awesomeopenstack
20:18 EmilienM mgagne: +1
20:18 spredzy mgagne, well basically we expect to have 3 ports open to traffic for nova::api (regular, ec2 and metadata), why can't we add parameter in nova::api to handle a firewall resource
20:18 mgagne spredzy: for the same reason I won't accept someone adding parameters to manage HA resources
20:18 mgagne spredzy: it's not something that should be part of puppet-nova
20:19 EmilienM mgagne: so there is no patch to do in extra then.
20:19 EmilienM mgagne: just in our module consuming firewall module in our classes
20:19 EmilienM mgagne: we wanted to bring the feature in each modules by using common code
20:19 mgagne EmilienM: if someone thinks there is value to bundle it in extras, go for it.
20:20 mgagne EmilienM: you risk going down a dangerous path where everyone will want to add their own 3rd party resources to it: firewall, ha, load-balancing, monitoring, etc.
20:21 spredzy mgagne, got your point
20:21 spredzy mgagne, seems fair about 3rd party resources hell
20:21 mgagne EmilienM: on the same note, we should remove nova::rabbitmq
20:22 mgagne EmilienM: this should be in extras
20:23 mgagne EmilienM: mysql user and permission provisioning is a gray area though. but AFAIK we do not provision THE mysql server in puppet-nova
20:25 EmilienM mgagne: you did not review my BP
20:25 mgagne EmilienM: which one, the one about identity? I'm trying to find time to review it
20:25 mgagne EmilienM: left at 1am yesterday ^^'
20:25 EmilienM mgagne: we agreed that would move to the lib
20:25 EmilienM like mysql
20:26 EmilienM sounds like we disagree here
20:26 crinkle I think the issue is that nova::rabbitmq is not like the other *::rabbitmq classes, it does a lot more
20:26 crinkle iirc
20:26 mgagne EmilienM: have a link to your spec? looks like we are not talking about the same one
20:27 mgagne crinkle: yep. If nova had a dedicated rabbitmq, MAYBE it would be ok but it also provision the server
20:28 EmilienM mgagne: https://review.openstack.org/#/c/129700/
20:28 EmilienM oh
20:28 mgagne EmilienM: ok, I though it was something about firewalling or rabbitmq
20:28 EmilienM ah sorry
20:28 mgagne thought*
20:29 EmilienM why having mysql in lib and rabbitmq in extra?
20:29 EmilienM mgagne: ^
20:29 mgagne EmilienM: mysql user and permission provisioning is a gray area though. but AFAIK we do not provision THE mysql server in puppet-nova
20:29 mgagne ^
20:30 mgagne crinkle: dedicated rabbitmq user*
20:31 EmilienM I'm lost
20:32 EmilienM mgagne: to me, there is no sense then to push some code in extra then for firewalling
20:32 crinkle hmm i'm a little lost too
20:32 mgagne EmilienM: don't provision infrastructure servers/services in puppet-nova like rabbitmq or mysql. it should be in extras as all openstack services will be able to share the same service.
20:32 crinkle nova::rabbitmq and cinder::rabbitmq both try to provision the rabbitmq server
20:32 mgagne crinkle: yes, those manifests should be deprecated and put into the recycle bin
20:32 crinkle glance::notify::rabbitmq sets up a bunch of rabbitmq configs in glance-api.conf
20:32 mgagne crinkle: move code to extras and tell people to use that instead.
20:33 EmilienM mgagne: fyi, mysql is in lib
20:33 crinkle mgagne: makes sense to me
20:33 crinkle EmilienM: individual mysql databases for each service get set up in lib
20:33 EmilienM I don't use those classes
20:33 crinkle but setting up the server itself would go in extras if we wanted to manage that
20:33 mgagne crinkle: +1
20:34 mgagne EmilienM: is this aspect clearer to you?
20:34 EmilienM mgagne: yes
20:34 mgagne EmilienM: about mysql, yes it's in lib, yes there is *::db::mysql is all our modules.
20:35 mgagne EmilienM: it's a gray area for me. but the "good" news is that it doesn't provision the Mysql server, only the users/databases/grants.
20:36 EmilienM I see
20:36 EmilienM ok
20:36 mgagne EmilienM: my fear is to have too much 3rd party stuff in each of our modules
20:36 EmilienM extra is empty anyway lol
20:36 mgagne EmilienM: it opens the door to abuse
20:37 EmilienM mgagne: I think I give up and I'll put the code in our puppet-openstack-cloud module
20:37 EmilienM I wanted to share the code with the community
20:37 mgagne EmilienM: I understand your wish
20:37 mgagne EmilienM: I think it's the same situation with Horizon
20:38 mgagne EmilienM: do you include vendors' dashboards or not? where do you draw the line?
20:39 EmilienM mgagne: indeed
20:46 mgagne EmilienM: damn ruby 1.8. changing the policy_files order at each run...
20:49 EmilienM ...
20:56 bitblt joined #puppet-openstack
21:13 JayJ joined #puppet-openstack
21:13 mgagne not much luck trying to override an apache::vhost resource with the spaceship operator...
21:14 mgagne Apache::Vhost <| title == $::horizon::wsgi::apache::vhost_ssl_conf_name |> { aliases +> $aliases } looks to not work =(
21:15 marcel joined #puppet-openstack
21:16 mgagne alright found it
21:17 mgagne wrong vhost =)
21:59 rcallawa joined #puppet-openstack
22:02 rcallawa joined #puppet-openstack
22:05 mgagne joined #puppet-openstack
22:08 openstackgerrit Xav Paice proposed a change to stackforge/puppet-ceilometer: Add package_ensure where missing  https://review.openstack.org/118281
22:11 marcel joined #puppet-openstack
22:20 rharrison OK, I"m beyond frustration trying to get the mysql_version params to work with the openstack modules. No matter what I set them too the >= test fails and tries to use the 0.x version of the puppetlabs-mysql module.  How do I get this to work?!
22:47 rharrison d0g5ofW@r
22:50 rharrison meh, well that was old and insecure anyway...
23:01 openstackgerrit Colleen Murphy proposed a change to stackforge/puppet-trove: Fix db sync relationships  https://review.openstack.org/130650
23:25 dgurtner joined #puppet-openstack
23:25 dgurtner joined #puppet-openstack
23:50 ashp joined #puppet-openstack

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