Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-08-11

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

All times shown according to UTC.

Time Nick Message
00:10 gildub joined #puppet-openstack
00:15 marun joined #puppet-openstack
01:27 marun joined #puppet-openstack
01:34 sputnik13 joined #puppet-openstack
01:50 rharrison joined #puppet-openstack
02:35 xingchao joined #puppet-openstack
02:36 sputnik13 joined #puppet-openstack
02:52 sputnik13 joined #puppet-openstack
02:59 daneyon joined #puppet-openstack
03:26 sputnik13 joined #puppet-openstack
03:33 sputnik13 joined #puppet-openstack
04:33 sputnik13 joined #puppet-openstack
04:34 mdorman joined #puppet-openstack
04:37 mdorman_ joined #puppet-openstack
04:49 mdorman joined #puppet-openstack
04:55 openstackgerrit Ivan Chavero proposed a change to stackforge/puppet-nova: Disable file injection when using RBD as compute ephemeral storage  https://review.openstack.org/113164
05:03 mdorman joined #puppet-openstack
05:03 mdorman_ joined #puppet-openstack
05:05 mdorman_ joined #puppet-openstack
05:06 mdorman_ joined #puppet-openstack
05:20 gildub joined #puppet-openstack
05:23 openstackgerrit Ivan Chavero proposed a change to stackforge/puppet-nova: Disable file injection when using RBD as compute ephemeral storage  https://review.openstack.org/113164
05:24 daneyon joined #puppet-openstack
05:26 mdorman_ joined #puppet-openstack
05:31 mdorman_ joined #puppet-openstack
05:33 mdorman joined #puppet-openstack
05:37 mdorman joined #puppet-openstack
05:39 mdorman joined #puppet-openstack
05:46 sasik joined #puppet-openstack
05:46 sasik left #puppet-openstack
05:48 mdorman joined #puppet-openstack
05:55 EmilienM good morning
06:22 michchap EmilienM: morning
06:54 saju_m joined #puppet-openstack
07:00 tfz joined #puppet-openstack
07:22 mmagr joined #puppet-openstack
07:34 arnaud_orange joined #puppet-openstack
07:35 walkleyn_ joined #puppet-openstack
07:54 tfz joined #puppet-openstack
07:57 qba73 joined #puppet-openstack
08:22 igajsin joined #puppet-openstack
08:25 kaliya joined #puppet-openstack
08:28 derekh joined #puppet-openstack
08:58 igajsin joined #puppet-openstack
08:58 artem_panchenko1 joined #puppet-openstack
09:29 saju_m joined #puppet-openstack
10:32 openstackgerrit Sylvain Baubeau proposed a change to stackforge/puppet-trove: Add trove_datastore and trove_datastore_version types  https://review.openstack.org/105699
11:04 tdb joined #puppet-openstack
11:15 dgollub joined #puppet-openstack
12:35 dprince joined #puppet-openstack
12:47 dmsimard joined #puppet-openstack
12:53 EmilienM michchap: ping
12:54 EmilienM michchap: I  have a  bad news :)
12:54 sbadia hi here :)
12:54 sbadia EmilienM: https://groups.google.com/forum​/#!topic/puppet-dev/kRtHbKKmgUw
12:54 mtl1 joined #puppet-openstack
12:54 EmilienM michchap: ^^^^^
12:54 morazi joined #puppet-openstack
12:59 michchap EmilienM: oh?
13:00 michchap EmilienM: haha. great =/
13:00 EmilienM ...
13:05 michchap EmilienM: I looked at nan's docco tool at one point, but unless I'm allowed to change every module to that style there's not much point
13:05 michchap EmilienM: because if I am generating doc, I want it to generate doc for every module.
13:07 jgrassler Good afternoon.
13:08 michchap hello jgrassler
13:08 jgrassler I've got a couple of patches for stackforge modules in the pipeline (adding some options). I already read up on the Gerrit workflow, but I'm unsure what the workflow is regarding tickets.
13:08 jgrassler Do I create a bug or a blueprint for the changes in question?
13:09 michchap jgrassler: is it a big feature patch? If it's not then don't worry about filing bugs or creating a spec.
13:09 Mr_D joined #puppet-openstack
13:10 jgrassler michchap: Not too big I guess, adding five options (always the same ones) to puppet-{ceilometer,cinder,heat,keystone,nova}
13:10 michchap jgrassler: We recently moved to the same system as upstream openstack for blueprints, where we write them as specs and review them through gerrit.
13:11 michchap jgrassler: so if you want to write up a spec, you can do it against this repo, and there's osme examples in the queue: https://review.openstack.org/#/q/status:ope​n+project:stackforge/puppet-openstack-specs,n,z
13:11 michchap jgrassler: for something in oslo?
13:12 jgrassler michchap: Well, if the change is small enough I'd rather not write a spec :-)
13:12 jgrassler michchap: Oslo?
13:13 jgrassler michchap: No, not oslo. It's a bunch of logging options that were missing from the puppet modules.
13:13 jgrassler michchap: log_config_append and a bunch of formatting options.
13:15 michchap jgrassler: Perhaps some overlap with this? https://review.openstack.org/#/c/106055/2
13:15 michchap jgrassler: We discussed it last week on the mailing list: https://groups.google.com/a/puppetlabs.com/​forum/#!topic/puppet-openstack/Km60sZROtR0
13:16 jgrassler michchap: Oops. Yes :-)
13:17 michchap jgrassler: so actually if you could make a spec for logging that would be very helpful, since there seem to be some differing opinions.
13:18 jgrassler michchap: The discussion certainly looks like it, yes.
13:21 jgrassler So let's see if I understood this correctly: these plain text specs go into a repository and gerrit is used for discussion?
13:23 michchap jgrassler: Yep.
13:23 jgrassler michchap: Ok. Thanks.
13:27 jgrassler michchap: I'll look into writing a spec then. Not sure whether I'll have the time to do it (it certainly looks like a lot of work to be doing on my employer's time, but we do need log_config_append in our Openstack installation anyway.).
13:28 michchap jgrassler: if you're going to need someone else to take over the patch(es) you can always leave a note. I think there is demand for the logging options so someone will probably take it up if you don't have time.
13:29 mjblack joined #puppet-openstack
13:30 jgrassler michchap: True enough.
13:51 prad joined #puppet-openstack
14:02 gholler joined #puppet-openstack
14:10 mjblack joined #puppet-openstack
14:16 hogepodge joined #puppet-openstack
14:19 sputnik13 joined #puppet-openstack
14:23 hogepodge Good morning!
14:23 arnaud_orange morning !
14:23 mjblack joined #puppet-openstack
14:34 EmilienM hogepodge: hey, good morning Sir.
14:34 rwsu joined #puppet-openstack
14:36 xingchao joined #puppet-openstack
14:40 sputnik13 joined #puppet-openstack
14:46 hogepodge I’m going through outstanding patched to talk about during the meeting. If anyone has any pets they’ve wanted looked at let me know.
14:48 hogepodge EmilienM I pulled that -1. Feel free to merge if there are other +2
14:48 EmilienM hogepodge: ack
14:48 hogepodge on the logging patch. Sorry about the delay on that.
15:00 michchap morning hogepodge
15:00 hogepodge morning michchap
15:00 hogepodge (or evening as it is in your case?)
15:01 michchap it's 1am.
15:02 hogepodge Etherpad for meeting notes. https://etherpad.openstack.org​/p/puppet-openstack-2014-8-11
15:03 hogepodge Ok, #begin meeting
15:04 sbadia hi
15:04 michchap logging?
15:04 hogepodge The agenda is listed in the etherpad. We’ll just work down the list.
15:04 hogepodge #topic logging
15:05 michchap jgrassler has put his hand up to make a spec
15:05 hogepodge There’s been a bit of discussion, and it looks like someone has a patch out there.
15:05 jgrassler More than one, it might appear.
15:06 jgrassler Yanis Guenane
15:06 hogepodge That sounds like a good idea. I’m happy with any solution. I prefer writing less code, and if we can configure logging inside of openstacklib that would make me happy. I wouldn’t want to have openstacklib code that just has thin wrappers around it though. That seems to be missing the point.
15:06 hogepodge I’m not fond of putting everything in init.pp
15:06 jgrassler has a ceilometer patch that's already being processed and I have a Ceilometer one plus others.
15:06 michchap hogepodge: I'm happy with a logging class if it's just a thin wrapper around identical parameters
15:07 michchap and have one for each module
15:07 michchap if there's nothing to consolidate, it doesn't go in openstacklib imo
15:08 michchap consolidating would allow setting a single logging type across classes easily, but we can already do that via hiera/ENC/other composition
15:08 jgrassler I like the idea of dedicated logging classes, too.
15:08 jgrassler (and I dislike the idea of openstacklib for now - that looks like it might take a while)
15:08 michchap so nova::logging etc? hogepodge are you ok with that?
15:08 hogepodge michchap yes
15:08 hogepodge Is everyone else happy with that solution?
15:09 jgrassler I could fairly easily rework my patches to be in dedicated classes.
15:09 jgrassler Yes.
15:09 xingchao +1 for nova::logging
15:10 hogepodge Ok, let’s move forward with that then.
15:10 michchap still need a spec? or not both
15:10 michchap bother*
15:10 sbadia +1 also for logging class
15:10 hogepodge A short spec could capture the decision, but it’s not strictly necessary.
15:10 jgrassler I would mind not having to write one, but I could do it.
15:11 jgrassler s/mind/not mind/
15:11 hogepodge Just send a follow up to the mailing list so we can capture the decision.
15:12 hogepodge Moving on, #topic ml2-plugin status
15:12 jgrassler Ok, so the follow-up to the list is sufficient (no spec required)?
15:12 hogepodge jgrassler yes
15:12 jgrassler Ok. Will do. Thanks.
15:12 hogepodge mgagne and EmilienM have been working pretty hard on how to properly configure the ml2 plugin
15:13 michchap I'm using ml2-ovs. Haven't tried linuxbridge yet
15:13 EmilienM hogepodge: I'm running some tests atm
15:13 hogepodge So I wanted to solicit feedback from them (and anyone else) on what we’re doing
15:13 xarses I'm using ml2 over ovs
15:13 EmilienM with puppet-neutron/master
15:13 EmilienM & ml2-ovs-gre
15:14 hogepodge I’m also using ml2-ove-gre (we don’t have hardware that will support vxlan sadly)
15:14 michchap iirc there is a slight awkwardness around the plugin class when using a provider network
15:14 EmilienM and people from my team is testing vxlan also
15:14 xarses it was mostly working fine 3 weeks ago, except for the bug's I opened
15:14 EmilienM hogepodge: you don't need hardward afik
15:14 EmilienM xarses: which one please?
15:14 EmilienM let's track it here if you don't mind
15:14 xarses https://bugs.launchpad.net/pupp​et-neutron/+bugs?field.tag=fuel
15:14 EmilienM thanks!
15:15 hogepodge EmilienM I think you need to have multicast enabled. I haven’t been able to set up vxlan networks on our network, and our ops guys are reporting that our network doesn’t support it.
15:15 EmilienM hogepodge: ah, indeed
15:15 Aimon joined #puppet-openstack
15:16 hogepodge So, what I was hoping was to have someplace where we could capture the findings on the network settings. All of this came up because of the patch that set the symlink to the plugin.ini file. There still seems to be a lot of confusion surrounding the right way to configure things.
15:17 hogepodge (rhel packaging links plugin.ini to the old ovs monolithic plugin, rather than the ml2 one)
15:17 xarses have we cleared up the plugin.ini linking ovs / vs linuxbridge what goes into where ?
15:17 hogepodge xarses it’s not clear to me :-)
15:17 fc__ joined #puppet-openstack
15:18 michchap I don't think there is an easy solution to this issue. The downstreams won't change their packages, but our users will expect to be able to use the stackforge modules with the dominant downstream packages. So where there is a non-rectified packaging bug we just have to work around
15:18 csharpsteen EmilienM: Did you get that encoding issue sorted out?
15:18 EmilienM csharpsteen: we are in the middle of weekly meeting, can we catch up in ~ 40 min? :)
15:19 aimon joined #puppet-openstack
15:19 hogepodge michchap agreed on that.
15:19 michchap The biggest worry I have is that we end up with conditionals for every version of rhel/ubuntu for each different version of openstack to handle different packaging bugs, but I don't have a solution
15:19 hogepodge If I understand correctly rh doesn’t consider it to be a packaging bug.
15:20 hogepodge michchap if there’s a way to capture that information in a generic way I’d prefer that. For example, consistently configuring symlinks across all of the packaging
15:20 csharpsteen Might be an hour and 40 for me, daily meetings are starting in 40. Could catch up earlier tomorrow as well if that works better for you :)
15:21 xingchao michchap: maybe repackaging rdo packages is a solution, we have maintaining openstack packages for custom use.
15:22 michchap xingchao: I don't think that's what most users expect though? FWIW I totally agree with you that building your own packages is worth the effort.
15:23 michchap xingchao: RedHat are also big users of these modules, so the handling will end up in there anyway
15:23 hogepodge xingchao I don’t think it a viable solution for most people. Even though at my previous job I found managing my own packages easier (mainly because RDO hadn’t been released yet)
15:23 xingchao michchap: you are right.
15:24 xarses cant we just work with the package folks to show them the problem?
15:24 michchap (I don't mean to pick on redhat, there are also ubuntu packing problems.)
15:24 daneyon joined #puppet-openstack
15:24 michchap If you look at http://docs.openstack.org/icehouse​/install-guide/install/yum/content​/neutron-ml2-controller-node.html the symlink is part of the official doc
15:24 sputnik13 joined #puppet-openstack
15:25 michchap I also met with some tripleo folks last week and they mentioned they have a similar situation where the RHEL builds are expected to account for packaging oddities
15:25 mdorman joined #puppet-openstack
15:25 xarses michchap: I didn't know there where ubuntu issues.  I can try to reach out to the canonical folks if you list the problems
15:25 hogepodge xarses https://bugzilla.redhat.co​m/show_bug.cgi?id=1113794 and https://bugzilla.redhat.co​m/show_bug.cgi?id=1113794 and https://bugzilla.redhat.co​m/show_bug.cgi?id=1087647
15:26 michchap xarses: There have been several to do with defaults in neutron init scripts. I am not working on ubuntu atm so I can't recall off the top of my head.
15:26 michchap I think we should move on with the meeting - this problem is not really solvable imo
15:26 hogepodge agree. take it to the mailing list?
15:27 hogepodge Just keep working around it I suppose.
15:27 xingchao ok
15:27 s_albtraum joined #puppet-openstack
15:27 hogepodge #topic backport reviews
15:28 michchap I think this was resolved on the list
15:28 hogepodge This is a quick follow up from the mailing list topic.
15:28 hogepodge Yup, wanted to pint it out in case anyone missed it.
15:28 mmaglana joined #puppet-openstack
15:29 hogepodge topic is here https://groups.google.com/a/puppetlabs.com/​forum/#!topic/puppet-openstack/9yGkMK6UsTU
15:29 hogepodge If there aren’t any more comment on it we can move on.
15:29 hogepodge #topic mid-cycle openstack operators meetup
15:30 hogepodge There’s going to me a mid-cycle OpenStack operators meet-up in San Antonio TX USA on August 25 and 26
15:31 hogepodge I’ll be going the second day to lead the puppet track. Wanted to make everyone aware of it in case you’re close by and wanted to travel to it.
15:31 hogepodge If you have any topics you want me to bring up or feedback you want me to solicit let me know.
15:32 EmilienM mgagne: are you going?
15:33 hogepodge I’m not sure mgagne is here this morning.
15:33 hogepodge moving on
15:33 hogepodge #topic database refactoring
15:34 hogepodge crinkle has a bunch of patches to transition mysql over to the common code in openstacklib
15:34 EmilienM hogepodge: I did reviews, it looks good to me
15:34 EmilienM I did not put +2 with the goal to let other folks reviewing
15:34 EmilienM but it seems I can now...
15:34 michchap apologies - I have an existing module called openstacklib so I need to rename a lot of things before I can do integration tests on the new module.
15:34 hogepodge Yeah, I’ve been avoiding +2 since I’ve been working with her.
15:35 hogepodge michchap urk. Sorry about that. :-/
15:35 michchap hogepodge: lol it's my own fault
15:35 EmilienM hogepodge: so I'll +2 all of them
15:35 EmilienM I'm looking forward to using it
15:35 michchap me too. It looks good.
15:36 hogepodge Yeah, I’ve been happy with the work. There’s postgres work in flight too.
15:37 hogepodge For RHEL 6.5 there are python database bugs that prevent reliable usage (although I think is just hits cinder)
15:37 EmilienM hogepodge: I was worried about the backward compat
15:37 hogepodge Not supporting 0.9?
15:37 mjblack joined #puppet-openstack
15:37 EmilienM maybe I missed something, but if people don't use the lib, the module will fail. Is it intentional ?
15:38 EmilienM hogepodge: no, the dependency to the lib
15:38 michchap I think the dependency is there in master for juno but won't be backported
15:38 sbadia EmilienM: hum, yep but it's master :)
15:38 EmilienM michchap: ok, perfect then
15:38 hogepodge EmilienM Yes, lib is going to be required going forward
15:38 michchap I don't think we should wait another cycle
15:38 hogepodge It won’t be backported to Icehouse.
15:38 EmilienM I know there are backports ninjas around here :)
15:38 sbadia ^^'
15:39 hogepodge soren has some outstanding mysql patches, they probably should land first so we can cherry pick against those
15:40 sbadia hum indeed
15:40 michchap changing the default?
15:40 EmilienM michchap: it's good, forget about it
15:40 sbadia adding 2.2 support in ironic?
15:40 michchap are we talking about https://review.openstack.org/#/c/107368/
15:41 hogepodge michchap Yes. It felt like extra work in light of the current patches, but it gives a consistent backport poing
15:41 hogepodge https://review.openstack.org/#/q/owner:Soren.Hansen%2540ril.com+status:open,n,z
15:41 michchap to be honest while I am happy with the change this is trivially easy to handle in hiera, which I'm confident Soren is using...
15:42 michchap backporting a change in default module requirement seems like a pretty bad idea
15:42 mjblack joined #puppet-openstack
15:42 hogepodge michchap. Heh. of course, yes you’re right.
15:42 * hogepodge slaps forehead
15:42 michchap I think we should just land the openstacklib ones and get master running the way we want it asap
15:43 hogepodge michchap yes, agreed.
15:44 hogepodge any other comments?
15:44 hogepodge #topic new spec reviews
15:45 hogepodge A few specs out there. The Aviator work that crinkle is working on, a pacemaker spec, and a common configuration provider spec.
15:45 hogepodge This is just a reminder (to myself included) to take a look over them
15:45 hogepodge https://review.openstack.org/#/c/107546/ https://review.openstack.org/#/c/104861/ https://review.openstack.org/#/c/99524/2/sp​ecs/common-openstack-configuration-provider.rst,cm
15:46 hogepodge Any comments or questions about this?
15:47 michchap common config provider I think it's going to be more about implementation than design, particularly judging from bodepd's comment
15:47 michchap I think we're in agreement it should be attempted
15:47 EmilienM yep
15:48 hogepodge ok
15:49 hogepodge #topic openstack_extras
15:50 hogepodge More of a reminder to think about code to put there. I like that there’s a pacemaker spec out there now.
15:50 hogepodge Repo code is there for review, mostly just a port of existing code.
15:50 michchap I have a ton of code I want to put there, but it needs refactoring
15:50 hogepodge michchap +1
15:51 michchap the repo code I'm not sure about. I made a note on the etherpad
15:51 hogepodge I saw that
15:51 michchap I ended up writing my own repo class. I want to make my own implementation of yum repo type, based on concat and file, so that I can purge the yum.repos.d directory in a sensible way
15:52 michchap but the downside is we start diverging from how most people in the puppet community does yum repos (use the builtin type)
15:52 michchap so I thought I'd ask here what people think about the idea
15:53 michchap the underlying issue is here: https://projects.puppetlabs.com/issues/9293
15:53 xarses michchap: fuel already is running a custom(ized) yum and apt provider to allow us for enforce package versions
15:54 michchap xarses: which subdir? apt and ?
15:54 hogepodge updated commentary on that issue here https://tickets.puppetlabs.com/browse/PUP-1066
15:54 michchap package?
15:54 hogepodge Since extras code is optional, I don’t see a problem with either solution, and actually prefer more robust code.
15:56 michchap hogepodge: Right, I guess I just want to write something that others can use, rather than just throwing my particular thing in there
15:57 hogepodge michchap xarses I think that is people are running things other than openstack on their openstack nodes they’re just asking for trouble, so I don’t see a particular solution to necessarily be a bad thing.
15:57 michchap hogepodge: well...they are running puppet :)
15:58 mmagr joined #puppet-openstack
15:58 hogepodge michchap I think a lot of people use masterless for their installs. ;-)
15:59 hogepodge michchap can you drop links to your solutions in the etherpad?
15:59 hogepodge xarses too
15:59 xarses i see they are slightly different problems now, but sure
15:59 hogepodge We’re coming up on the end of the hour. Any other comments or issues?
16:01 michchap I'm good. Thanks all.
16:01 EmilienM thanks, same
16:01 hogepodge I’m going to close out the meeting now. On a side note, if anyone has feedback or comments on how they’re being run let me know. I’m kind of figuring this out as I go along.
16:02 hogepodge Thanks everyone!
16:02 hogepodge #end meeting
16:02 xarses have a good rest of your day
16:02 EmilienM you too!
16:03 qba73 joined #puppet-openstack
16:03 sbadia thank :)
16:04 xarses hogepodge: we've been using a non-dated lp for our meeting minutes. That way we don't have to change the link for every meeting
16:05 xarses otherwise +1
16:05 hogepodge xarses do you copy the notes over to a new pad, or do you just keep appending?
16:07 xarses appending for the most part as we leave most of the details in the meeting minutes so they dont get too long. In this case I could see moving the old notes to a separate pad and linking it back to the re-occurring one.
16:07 xarses https://etherpad.openstack.org​/p/fuel-weekly-meeting-agenda
16:09 xingchao joined #puppet-openstack
16:35 xarses joined #puppet-openstack
16:39 thumpba joined #puppet-openstack
16:39 imcsk8 joined #puppet-openstack
16:44 hogepodge One thing I forgot in the meeting, this epic patch to vswitch.
16:44 hogepodge https://review.openstack.org/#/c/94504/
17:11 hogepodge joined #puppet-openstack
17:11 openstackgerrit A change was merged to stackforge/puppet-neutron: Hide secrets from puppet logs  https://review.openstack.org/106524
17:23 mmagr joined #puppet-openstack
17:24 marun joined #puppet-openstack
17:43 mmaglana joined #puppet-openstack
18:08 dgollub joined #puppet-openstack
18:18 newuser joined #puppet-openstack
18:37 mjblack joined #puppet-openstack
18:38 mjblack joined #puppet-openstack
18:48 tfz joined #puppet-openstack
18:52 openstackgerrit Mathieu Gagné proposed a change to stackforge/puppet-neutron: Add ability to manage Neutron ML2 plugin package state  https://review.openstack.org/113344
19:00 mjblack joined #puppet-openstack
19:39 marun joined #puppet-openstack
20:04 mmaglana joined #puppet-openstack
20:28 xarses joined #puppet-openstack
20:41 hogepodge joined #puppet-openstack
21:25 mmaglana joined #puppet-openstack
21:27 richm joined #puppet-openstack
21:28 richm hello - need help figuring out why rake spec test fails with PUPPET_SPEC_VERSION="~> 2.7.0" but passes with other versions
21:28 richm https://review.openstack.org/#/c/109676/
21:29 richm http://logs.openstack.org/76/109676/9/check/​gate-puppet-keystone-puppet-unit-2.7/d123c28
21:33 richm I think the problem is at lib/puppet/util/package.rb:4 called from ssl.conf.erb:16
21:33 richm for some reason @apache_version is nil
21:37 richm I have tried various ways to try to make sure it is set - including ::apache::version before anything else in keystone::wsgi::apache, even passing in apache_version => 2.2 to class { '::apache': .....}
21:38 mmaglana joined #puppet-openstack
21:45 hogepodge richm sometimes trailing commas can cause problems for older versions of puppet
21:45 richm ok
21:46 hogepodge not very intuitive, and it may not be the problem.
21:52 richm http://paste.openstack.org/show/93541/
21:52 richm does not seem to be the problem
21:53 richm I hacked ssl.conf.erb to print out the contents of scope
21:53 richm apache_version is not in there
21:53 richm doing <% puts scope.lookupvar('::apache::apache_version') -%>
21:53 richm prints "undefined"
21:53 richm so apache version seems not to be set at this point in the processing
21:54 richm The only thing that makes it work is using "include ::apache" instead of trying to use the class with purge_configs => false
21:59 imcsk8 richm: but you have keep the configs, purging them can give a lot of problems
22:00 imcsk8 richm: (i'm ichavero)
22:02 richm which is why I can't use include ::apache in keystone (or horizon, or nagios, or ...)
22:03 imcsk8 richm: yeah it always has to be with purge_configs => false
22:04 richm here is the scope dump output - http://paste.openstack.org/show/93546/
22:07 imcsk8 so at some point the apache_version is being lost
22:08 richm I'm wondering if it is being set at all
22:09 richm at least, if it is being set before apache::mod::ssl is processed
22:09 richm I tried doing this:  class keystone::wsgi::apache { ... } inherits ::apache::version { ...
22:10 richm and now ::apache::version::default is set, but still not ::apache::apache_version
22:17 imcsk8 it's suppoused to be set in the version.pp manifest
22:19 richm imcsk8: it == ::apache::version::default?
22:19 richm if so, then yes
22:20 richm This is what I don't understand - apache::version is not included anywhere
22:20 richm however, class apache depends on it in the parameter list
22:21 richm class apache ( ..., $apache_version       = $::apache::version::default, ...)
22:21 richm I thought this was verboten
22:22 richm there are other parameters whose default values are set by other classes:
22:22 richm $user                 = $::apache::params::user,
22:22 richm but this is ok because apache inherits from ::apache::params
22:24 imcsk8 richm: maybe apache::version is not being loaded
22:24 richm Is there some way I can trace the puppet module loading when using rake spec?
22:25 imcsk8 i'm not sure
22:25 richm I've tried doing "include apache::version" as the very first line in class keystone::wsgi::apache
22:25 richm I
22:25 richm I've even tried doing class keystone::wsgi::apache (...) inherits ::apache::version { ....
22:26 richm thinking that would force ::apache::version to be loaded before evaluating apache and apache::mod::ssl
22:27 imcsk8 one question, does that only causes problem with the spec files or also has problems when using the manifests?
22:27 richm you mean, when actually using packstack/puppet to do an install?
22:28 imcsk8 yeah
22:28 richm packstack works fine on F20
22:28 richm with the default packages for puppet, ruby, etc.
22:28 richm I don't know how to make it force the use of 2.7.0
22:29 richm that is, this is only a problem with PUPPET_SPEC_VERSION="~> 2.7.0"
22:29 richm with other versions, spec tests work fine
22:36 imcsk8 richm: what if you add apache::version here? https://github.com/richm/puppet-keystone/blob/mas​ter/spec/classes/keystone_wsgi_apache_spec.rb#L20
22:37 imcsk8 that might be the source of the problem, after all it's not a problem with the manifest, it's a problem with the spec file
22:38 richm you mean, add it { should contain_class('apache::version') } before the 'apache' line?
22:43 imcsk8 richm: yes
22:43 richm running now
22:43 richm nope - same erropr
22:43 richm error
22:47 imcsk8 damn!
22:47 richm egads - I hacked apache init.pp to explicitly set apache_version = 2.2 in the class parameters list - I still get the error!
22:47 richm it is as if apache::mod::ssl is being processed before apache
22:48 richm but I don't see how that is possible given the requirements/dependencies in ssl.pp
22:49 imcsk8 the problem is in the spec file
22:59 richm ok - what should I do now?
23:02 imcsk8 richm: i woild comment the offending line in the spec file to see if the same error appears
23:03 imcsk8 richm: i think it's the hay the spec file is handling the apache stuff
23:04 richm the offending lines?  That is, if you look at http://logs.openstack.org/76/109676​/9/check/gate-puppet-keystone-puppe​t-unit-2.7/d123c28/console.html.gz
23:04 richm lines 20-24 fail, and lines 28, 35, 45, 55, 69 fail
23:05 richm basically, everything that should have been set up by apache, apache::vhost, etc.
23:08 badiane_ka joined #puppet-openstack
23:21 s_albtraum left #puppet-openstack
23:21 imcsk8 richm: does this fail even without the changes you made?
23:29 hogepodge joined #puppet-openstack
23:29 gildub joined #puppet-openstack
23:57 daneyon joined #puppet-openstack

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