Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-05-06

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

All times shown according to UTC.

Time Nick Message
00:00 gildub crinkle, yeah, AFAIK I think we trying to be able to support masterless approach as muchs as possible
00:00 crinkle yeah
00:02 gildub crinkle, using user's homedir as default path but also offer that to be overridden, but how? Hiera/params.pp
00:02 gildub crinkle, that won't work for "puppet resource" commands
00:03 crinkle gildub: i don't know how it would be overridden, that's sort of the issue with the rc file approach
00:04 crinkle is it could live anywhere and there's no way to say where it is
00:04 gildub crinkle, what about a using a facter?
00:04 crinkle gildub: how would facter detect where the file is?
00:05 gildub crinkle, hmmm
00:05 crinkle gildub: someone could decide the best place for it is /tmp/openstack/stuff/.openrc and no one would be able to guess that
00:06 crinkle gildub: we could maybe have it check a few common places
00:06 crinkle gildub: like, check /root/openrc, /root/.openrc, /root/keystonerc, /etc/openrc.conf, and then give up
00:06 gildub crinkle, well convention can say "we expect file to be there", if not then please make a symlink
00:06 crinkle gildub: that too
00:06 gildub crinkle, yeah, exactly ^^^
00:06 crinkle gildub: then we have to decide on a convention
00:10 gildub crinkle, well openstackrc file contain auth stuff which are sensitive information, normally the file access is controlled (permissions) as well as it's location, it could not be in tmp for instance (too risky), etc. So a user based location approach is appropriate
00:10 crinkle sure, bad example, just saying it's not necessarily predictable
00:10 richm what user does puppet run as?  It has to have root privs
00:10 crinkle it runs as root
00:11 xingchao joined #puppet-openstack
00:11 richm I don't mean to sidetrack this conversation, but you might want to see this:
00:11 gildub crinkle, couldn't be run as any user with sudo?
00:11 richm https://bugs.launchpad.net/pup​pet-openstacklib/+bug/1452059
00:11 richm and this https://bugs.launchpad.net/pyth​on-openstackclient/+bug/1452045
00:11 richm and note that openstack client now has its own file format
00:11 richm which is yaml
00:11 crinkle gildub: maybe
00:12 richm The common idiom in puppet openstack module code is this:
00:12 richm write out a config file containing a bunch of authentication parameters for keystone
00:13 richm read in that config file and convert those authentication parameters to openstack command line arguments
00:13 richm this seems to be common in puppet-glance, puppet-cinder, etc. etc.
00:13 richm In fact glance does this conversion in a two step process
00:14 richm I'm proposing that we just hand off glance-api.conf to openstack client - openstack --os-auth-config-file /etc/glance/glance-api.conf --os-auth-config-section keystone_authconfig
00:15 richm but the osc guys are reluctant to add support for Yet Another Config File Format
00:15 richm so I'm proposing we do this in openstackclient
00:15 richm I've already got some code to test - so my talk is cheap, lets see my review
00:16 gildub richm, I'm confused here, so OSC guys are not going to do it so you're suggesting we do it?
00:17 gildub richm, in openstacklib ^
00:17 richm I've opened a bug against OSC and I've already been told that they do not consider it a priority
00:17 richm So I opened a bug against puppet-openstacklib to do it
00:17 gildub richm, ok
00:17 richm and did it
00:18 gildub richm, what do you mean
00:18 richm I mean I've already got some code to do this and I'm testing it now
00:19 gildub richm, k
00:20 gildub richm, so the idea is to tap directly and only in the configuration files (keystone.conf, glance.conf, etc)?
00:20 gildub richm, right?
00:21 richm directly, but not only - in addition to
00:21 gildub richm, but we're already doing that, as you said
00:22 gildub richm, well we can/should definitely keep that besides a default rc file
00:23 richm yes - I'm proposing to prevent a lot of code duplication in every puppet module
00:23 gildub richm, oh so you want to generalize that and have it in common in openstack lib instead of each module to deal with
00:24 richm yes
00:24 gildub richm, you're going to be writing puppet-oslo mate!
00:24 richm by the time it's over, yes . . .
00:24 gildub richm, :)
00:25 richm what's the opposite of :-) ?
00:25 gildub richm, :D(
00:25 gildub or maybe just :( ;^)
00:26 richm down under (-:
00:26 gildub which I could flip my monitor 90 degre, having neckaches!
00:27 gildub LOL ^^
00:28 gildub richm, Anyway, that sounds good!
00:29 gildub crinkle, richm, so finding credentials workflow could be:
00:30 gildub 1. Environment (ENV[<token>] or ENV[<project>] in short)
00:30 gildub 2. RC file (default location by convention ~user/<something>
00:31 gildub 3. Provider to tap into corresponding provider configuration file, for instance keystone.rc for puppet-keystone
00:31 gildub ^ sorry I meant keystone.conf
00:32 crinkle ++
00:33 crinkle might be worth noting that keystone.conf is the oddball because it uses admin_token and not username/password credentials
00:34 gildub crinkle, ture
00:34 gildub true
00:34 gildub crinkle, but richm, as the answer in his new code ;)
00:35 gildub ^has
00:35 richm yeah, keystone is different enough to not use the new code
00:38 gildub crinkle, richm, ok then I put that in the mailing list to share/discuss it more
00:43 gildub richm, ^ before I do so, I just realized you responded in main list only - I saw the email about dropping puppet-openstack list, but didn't know was effective
00:45 gildub richm, anyway dropping request vs self.request is still a good thing to do and I'd like to propose a patch to remove it, and following workflow describe above
00:46 gildub richm, as we said bad timing, but impact is mostly removing the duplication code on the current reviews
00:50 richm ok
00:59 openstackgerrit Richard Megginson proposed stackforge/puppet-openstacklib: read auth parameters from component config file  https://review.openstack.org/180369
00:59 richm et voila
01:11 aimon joined #puppet-openstack
01:19 ericpeterson joined #puppet-openstack
01:27 ericpeterson joined #puppet-openstack
01:30 gildub richm, ok, will comment the review
01:31 ericpeterson joined #puppet-openstack
01:45 xingchao joined #puppet-openstack
01:46 derekh_ joined #puppet-openstack
01:54 ericpeterson joined #puppet-openstack
01:57 openstackgerrit Sam Morrison proposed stackforge/puppet-neutron: Don't specify a nova region by default  https://review.openstack.org/180378
02:10 fedexo joined #puppet-openstack
02:14 ericpeterson joined #puppet-openstack
02:45 xarses joined #puppet-openstack
02:53 vinsh joined #puppet-openstack
03:11 xarses joined #puppet-openstack
03:19 ericpeterson joined #puppet-openstack
03:32 vinsh joined #puppet-openstack
03:44 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
03:50 fedexo joined #puppet-openstack
04:00 openstackgerrit Merged stackforge/puppet-nova: Update Beaker tests  https://review.openstack.org/174182
04:11 markvoelker joined #puppet-openstack
04:14 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
04:16 openstackgerrit Merged stackforge/puppet-vswitch: Beaker tests  https://review.openstack.org/178198
04:23 xingchao joined #puppet-openstack
04:29 openstackgerrit Colleen Murphy proposed stackforge/puppet-glance: Use OpenstackClient for glance_image auth  https://review.openstack.org/172580
04:39 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
04:55 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
05:12 markvoelker joined #puppet-openstack
05:14 saneax joined #puppet-openstack
05:35 subscope_ joined #puppet-openstack
05:49 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-openstacklib: Predefined source of authentication information  https://review.openstack.org/180407
05:51 markvoelker joined #puppet-openstack
05:59 higgins joined #puppet-openstack
06:13 xingchao joined #puppet-openstack
06:39 stamak joined #puppet-openstack
06:40 alex_bh joined #puppet-openstack
06:46 jayg|g0n3 joined #puppet-openstack
06:52 markvoelker joined #puppet-openstack
06:54 tfz joined #puppet-openstack
07:00 mmagr joined #puppet-openstack
07:06 dgurtner joined #puppet-openstack
07:14 ahcorporto joined #puppet-openstack
07:55 jpena joined #puppet-openstack
07:57 gfidente joined #puppet-openstack
07:59 jistr joined #puppet-openstack
08:06 xingchao joined #puppet-openstack
08:12 rushiagr joined #puppet-openstack
09:00 ddmitriev joined #puppet-openstack
09:00 morazi joined #puppet-openstack
09:00 ianw joined #puppet-openstack
09:16 cdent joined #puppet-openstack
09:19 aimon joined #puppet-openstack
09:31 saneax joined #puppet-openstack
10:03 xingchao joined #puppet-openstack
10:35 igajsin joined #puppet-openstack
10:35 igajsin left #puppet-openstack
10:36 rcallawa joined #puppet-openstack
10:42 starmer joined #puppet-openstack
11:01 openstackgerrit Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_ensure  https://review.openstack.org/180484
11:03 openstackgerrit Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_ensure  https://review.openstack.org/180484
11:05 openstackgerrit Giulio Fidente proposed stackforge/puppet-tripleo: Add support for ensure_enabled  https://review.openstack.org/180484
11:05 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-openstacklib: Predefined source of authentication information  https://review.openstack.org/180407
11:09 xingchao joined #puppet-openstack
11:16 rcallawa_ joined #puppet-openstack
11:16 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-openstacklib: Predefined source of authentication information  https://review.openstack.org/180407
11:28 social EmilienM: sbadia: could we please branch puppet-sahara for stable/juno?
11:35 openstackgerrit Giulio Fidente proposed stackforge/puppet-glance: Decouple sync_db from manage_service and enabled  https://review.openstack.org/180229
11:48 openstackgerrit Gilles Dubreuil proposed stackforge/puppet-openstacklib: Predefined source of authentication information  https://review.openstack.org/180407
11:50 ferest joined #puppet-openstack
11:51 dprince joined #puppet-openstack
11:58 alex_bh2 joined #puppet-openstack
11:59 markvoelker joined #puppet-openstack
12:02 gfidente dprince, EmilienM I pushed https://review.openstack.org/#/c/180229/ to decouple db_sync from service being enabled
12:03 gfidente but I noticed similar thing happens for other modules, eg: https://github.com/stackforge/puppet-key​stone/blob/master/manifests/init.pp#L827
12:03 gfidente am I missing something, or is it just that we want to decouple in the other modules as well?
12:06 morazi joined #puppet-openstack
12:43 dprince gfidente: does this work? In the case of puppet-glance the db-sync exec() resource is set to refreshonly => true... so it will only run if another resource notifies it?
12:45 gfidente dprince, the triggers are unchanged
12:46 gfidente or am I missing how that interacts with the triggers?
12:47 dprince gfidente: right, I guess my point is some of the triggers may not fire if enabled is false and/or manage_service is false right?
12:48 csschwe joined #puppet-openstack
12:50 gfidente dprince, so I don't see any depending strictly on it the two ... but given the purpose of 'enabled' it looks to me they should not
12:50 gfidente dprince, I will report as soon as deployment finished
12:51 rcallawa joined #puppet-openstack
12:52 dprince gfidente: cool. So you are looking for a way to just run the db_sync via these puppet modules?
12:53 gfidente dprince, so run db_sync even though we don't enable the service
12:53 openstackgerrit joined #puppet-openstack
12:54 gfidente dprince, that's why I asked about the similar behaviour found in puppet-keystone
12:54 xavpaice_ joined #puppet-openstack
12:55 gfidente it seems to me it is fair assumption to be able to run db_sync even though we don't want the service enabled on boot
12:57 EmilienM good morning
13:06 EmilienM gfidente: it's on my todo for today
13:06 csschwe joined #puppet-openstack
13:08 xingchao joined #puppet-openstack
13:08 gfidente EmilienM, ^_^
13:21 richm joined #puppet-openstack
13:24 openstackgerrit Giulio Fidente proposed stackforge/puppet-glance: Decouple sync_db from manage_service and enabled  https://review.openstack.org/180229
13:28 gildub joined #puppet-openstack
13:38 openstackgerrit joined #puppet-openstack
13:41 tfz joined #puppet-openstack
13:43 openstackgerrit Richard Megginson proposed stackforge/puppet-openstacklib: read auth parameters from component config file  https://review.openstack.org/180369
13:43 tfz joined #puppet-openstack
13:44 gildub joined #puppet-openstack
13:54 JayJ_ joined #puppet-openstack
13:54 ericpeterson joined #puppet-openstack
13:56 dfisher joined #puppet-openstack
13:57 openstackgerrit Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_manage  https://review.openstack.org/180484
14:17 xingchao joined #puppet-openstack
14:20 openstackgerrit Merged stackforge/puppet-ceilometer: Ensure python-mysqldb is installed  before MySQL db_sync  https://review.openstack.org/177593
14:23 gfidente joined #puppet-openstack
14:33 chem joined #puppet-openstack
14:33 openstackgerrit Giulio Fidente proposed stackforge/puppet-keystone: Decouple sync_db from enabled  https://review.openstack.org/180565
14:35 openstackgerrit Giulio Fidente proposed stackforge/puppet-keystone: Decouple sync_db from enabled  https://review.openstack.org/180565
14:37 stevemar joined #puppet-openstack
14:40 vinsh joined #puppet-openstack
14:44 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
15:06 rcallawa_ joined #puppet-openstack
15:12 jeh joined #puppet-openstack
15:13 jeh left #puppet-openstack
15:22 mdorman joined #puppet-openstack
15:25 openstackgerrit Sebastien Badia proposed stackforge/puppet-horizon: Beaker tests  https://review.openstack.org/177253
15:36 openstackgerrit Giulio Fidente proposed stackforge/puppet-keystone: Decouple sync_db from enabled  https://review.openstack.org/180565
15:37 openstackgerrit joined #puppet-openstack
15:41 openstackgerrit Emilien Macchi proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the user_role resource  https://review.openstack.org/174999
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - keystone and keystone::roles::admin  https://review.openstack.org/175060
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the user resource  https://review.openstack.org/174976
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the project resource  https://review.openstack.org/174468
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - beaker tests  https://review.openstack.org/178828
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - the domain resource  https://review.openstack.org/174439
15:41 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - add domains to service_identity  https://review.openstack.org/180586
15:42 EmilienM richm: I think I have to rebase beaker patch on top of "add domains to service_identity"
15:42 EmilienM right?
15:43 EmilienM richm: though I let you rebase "add domains to service_identity"
15:46 richm EmilienM: I can rebase
15:46 EmilienM thanks
15:47 EmilienM richm: I want beaker patch on top of everything to make sure I test all
15:47 richm ok
15:48 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - add domains to service_identity  https://review.openstack.org/180586
15:48 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - beaker tests  https://review.openstack.org/178828
15:48 richm EmilienM: ^^
15:49 EmilienM richm: nice!
15:50 rcallawa joined #puppet-openstack
15:50 EmilienM richm: should I abandon https://review.openstack.org/#/c/180274/ ?
15:50 EmilienM richm: because of https://review.openstack.org/#/c/180407/
15:50 openstackgerrit Sebastien Badia proposed stackforge/puppet-horizon: Beaker tests  https://review.openstack.org/177253
15:54 richm EmilienM: yes, if we are going to drop resource[:auth] completely
15:54 EmilienM richm: ack
15:54 EmilienM richm: done
15:56 openstackgerrit Emilien Macchi proposed stackforge/puppet-cinder: Clean Gemfile: drop rspec-puppet and rake  https://review.openstack.org/180004
15:58 cwolferh joined #puppet-openstack
16:12 chem joined #puppet-openstack
16:19 openstackgerrit Jiri Stransky proposed stackforge/puppet-cinder: Decouple $sync_db from $enabled  https://review.openstack.org/180610
16:21 Fin1te joined #puppet-openstack
16:33 cdelatte joined #puppet-openstack
16:41 aimon joined #puppet-openstack
16:43 _ryanv_ joined #puppet-openstack
16:51 xingchao joined #puppet-openstack
16:53 xarses joined #puppet-openstack
16:58 openstackgerrit Sebastien Badia proposed stackforge/puppet-glance: beaker: test glance_image provider  https://review.openstack.org/176521
17:13 starmer joined #puppet-openstack
17:16 mfisch anyone know whats up with this review? https://review.openstack.org/#/c/169071/
17:16 mfisch jenkins isnt picking it up
17:18 openstackgerrit Merged stackforge/puppet-swift: Added rsyslog logging support to object-server  https://review.openstack.org/169071
17:21 mfisch vinsh: looks like Jenkins was waiting to make sure I was cool with it ;) https://review.openstack.org/#/c/169071/
17:28 vinsh :p
17:28 vinsh Sweet
17:38 dlb76 joined #puppet-openstack
17:45 openstackgerrit Bradley Klein proposed stackforge/puppet-monasca: Make api server have configurable backend db  https://review.openstack.org/180652
18:04 xingchao joined #puppet-openstack
18:11 ducttape_ joined #puppet-openstack
18:16 xarses joined #puppet-openstack
18:23 Fin1te joined #puppet-openstack
18:27 stamak joined #puppet-openstack
18:32 EmilienM crinkle: hello, did you test your patch with glance provider refacto?
18:33 crinkle EmilienM: no, i WIP that patch
18:33 EmilienM crinkle: it does not work for me
18:33 EmilienM http://logs.openstack.org/21/176521/19/chec​k/gate-puppet-glance-puppet-beaker-rspec/7c​3f8ec/console.html#_2015-05-06_17_14_23_234
18:33 EmilienM but it seems more an auth issue
18:34 crinkle EmilienM: gildub and richm's changes are probably going to affect that patch, so i was going to wait a bit on it
18:35 EmilienM crinkle: cool cool, I was just asking if I should investigate more or wait a bit
18:35 EmilienM I'll #2
18:36 crinkle bodepd: I think you mentioned you were working on a new provider for nova that used the openstackclient stuff, do you have an opinion on us ditching the auth parameter?
18:41 EmilienM richm: this is awesome: https://review.openstack.org/#/c/180​586/2/manifests/resource/authconfig.pp,cm
18:41 EmilienM crinkle, mgagne, I'll like you tell me if you agree with that thing ^
18:41 bodepd crinkle: we were using the auth parameter
18:42 bodepd crinkle: but I'm not even sure that we should have been
18:42 bodepd what would you use at all
18:42 mgagne EmilienM: are we introducing deprecated parameters right off the bat?
18:43 crinkle bodepd: environment variables, an openrc, or keystone_authtoken values in *.conf
18:43 mgagne EmilienM: someone, I feel we should rename the define to reflect its actual purpose. I would include keystone_authtoken somewhere in the name or something similar
18:43 EmilienM mgagne: we need to review the params, indeed. I was happy to see this kind of define
18:43 bodepd I think that is fine for all of our use cases (to require an openrc if it isn't the service credentials
18:44 mgagne EmilienM: the "problem" is that keystonemiddleware isn't tied to keystone itself
18:49 hogepodge joined #puppet-openstack
18:49 jesusaurus joined #puppet-openstack
18:49 asilenkov joined #puppet-openstack
18:49 pasquier-s joined #puppet-openstack
18:49 ducttape_ joined #puppet-openstack
18:49 openstackgerrit joined #puppet-openstack
18:49 vinsh joined #puppet-openstack
18:49 markvoelker joined #puppet-openstack
18:49 ianw joined #puppet-openstack
18:52 EmilienM mgagne: +1
18:52 EmilienM mgagne: I think we should drop the auth_ and use identity_url
18:52 openstackgerrit Merged stackforge/puppet-ironic: Run dbsync as root and don't manage /etc/ironic & ironic.conf ownership  https://review.openstack.org/179531
18:54 chem joined #puppet-openstack
18:54 openstackgerrit Merged stackforge/puppet-neutron: FWaaS: update packaging for Debian & Ubuntu  https://review.openstack.org/179210
18:54 openstackgerrit Merged stackforge/puppet-heat: Engine: validate auth_encryption_key  https://review.openstack.org/178478
18:55 chem left #puppet-openstack
18:55 openstackgerrit Merged stackforge/puppet-trove: Create Trove config before dbsync  https://review.openstack.org/179925
18:55 openstackgerrit Merged stackforge/puppet-ironic: Beaker tests  https://review.openstack.org/178786
18:57 cwolferh joined #puppet-openstack
19:01 openstackgerrit Merged stackforge/puppet-sahara: Allow to deploy on Ubuntu  https://review.openstack.org/179136
19:02 richm I didn't want to put "authtoken" in the name because that implies it is used for token authentication
19:02 openstackgerrit Merged stackforge/puppet-sahara: Fix Sahara installation for Ubuntu  https://review.openstack.org/179276
19:03 richm The parameters are introduced as a deprecated as a convenience.  I've noticed that a lot of code still relies on auth_host, port, etc.
19:03 richm If we don't accept then, that means every puppet module has to have logic to construct the auth_url if not given by the user.
19:03 richm s/then/them/
19:03 richm But if you guys think that's ok, it's fine with me
19:04 richm It is unfortunate that the keystone middleware still uses the name "keystone_authtoken", but I didn't see a need to propagate that
19:06 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - add domains to service_identity  https://review.openstack.org/180586
19:06 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - beaker tests  https://review.openstack.org/178828
19:07 EmilienM mgagne: please look richm's comment on https://review.openstack.org/#/c/180586/
19:12 EmilienM richm: a lot of modules already have the logic : https://github.com/stackforge/puppet-nov​a/blob/master/manifests/api.pp#L242-L252
19:14 richm EmilienM: ok - I can remove the deprecated parameters
19:15 EmilienM richm: I think by the end, we will have to have only identity_url
19:15 richm yes
19:15 richm no need for two separate urls any more
19:15 EmilienM that's nice
19:15 EmilienM richm: I really like this define
19:16 EmilienM mgagne suggested to find another name though,why not 'authtoken' ?
19:17 richm because 'authtoken' implies that it is for token auth
19:17 richm I think the name is unfortunate
19:18 richm I know that keystone middleware uses "keystone_authtoken" - that is also unfortunate
19:18 EmilienM richm: you can keep this name, and if someone is not happy, we have gerrit for reviewing :)
19:18 richm ack
19:19 jesusaurus joined #puppet-openstack
19:19 asilenkov joined #puppet-openstack
19:19 pasquier-s joined #puppet-openstack
19:19 richm should auth_url default to "http://127.0.0.1:35357/"?  Or should it be a required parameter?
19:20 richm I would like to make it required with no default value
19:21 richm Every module I've seen so far can provide this value
19:21 richm I don't think it will cause any problems if there is no default
19:22 EmilienM richm: wfm
19:24 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - add domains to service_identity  https://review.openstack.org/180586
19:24 openstackgerrit Richard Megginson proposed stackforge/puppet-keystone: support for keystone v3 api - beaker tests  https://review.openstack.org/178828
19:36 ducttape_ joined #puppet-openstack
19:39 EmilienM mgagne: "I suggest we wait ~3 months before cutting a stable branch for kilo"
19:39 mgagne EmilienM: hi
19:39 EmilienM mgagne: do you mean we can merge the patches that break juno in master?
19:40 EmilienM and cut master in 3 months?
19:40 mgagne EmilienM: yes, juno is "old", new stable release is kilo
19:40 EmilienM well, I think that the biggest point of the discussion now
19:40 EmilienM that's where you/me and mfisch / clayton disagreed
19:40 mgagne EmilienM: well I think people have unfair expectation from master branch
19:41 mgagne EmilienM: I don't go around complaining that master broke my icehouse installation
19:41 EmilienM mgagne: your comment is good, but I would like to focus on the problems we have first
19:42 EmilienM mgagne: your comment is like all your thoughts in one
19:42 mgagne EmilienM: master was never meant to be used for stable release and now we are stuck with a problem we shouldn't have in the first place
19:43 mfisch I dont see how its unfair when every other Openstack project manages not to break stuff with deprecations
19:44 mgagne EmilienM: if we cut stable/kilo in 3 months (and do not really do any major changes to support Liberty in master), people will be able to still use master for kilo for 3 more months, enough time to stabilize the master branch.
19:44 mgagne mfisch: you know there is a different between OUR api and openstack api?
19:44 maximov joined #puppet-openstack
19:44 mgagne mfisch: we do deprecate our parameters
19:45 clayton he's talking about service config file deprecations
19:46 mgagne clayton: forget about config file. there is more than that
19:46 clayton at least I think so
19:46 mgagne clayton: like service and package name
19:46 clayton perhaps, but config file deprecations were the biggest issue for us with icehouse -> juno, and the most recent issue with master.
19:47 clayton just because it's not the only issue doesn't mean it's not a valid one
19:47 clayton and it's easier to manage in a backwards compat way than the others
19:47 mfisch (sorry I had to get on a call but clayton can speak for me)
19:47 clayton at least potentially
19:47 mgagne because people expect master to work with a previous release of OpenStack, not a previous release of OUR puppet modules configs like we *do* support now
19:48 clayton perhaps I failed, but I tried to make it clear last time this came up in the meeting that I don't expect anyone to hold required kilo changes off the master branch
19:49 clayton I believe matt feels the same way
19:50 mgagne so if we wish to cut stable/kilo, what will happen to this change? https://review.openstack.org/#/c/173495/
19:50 clayton The point matt and I have tried to make is that we'd like see non-required changes (like config file deprecations) held for the time being, and discuss options for better ways to manage them
19:50 clayton I agree with Coleen's comment
19:50 mgagne will it get merged in master? in stable/kilo after the branch is cut? what will happen with master?
19:50 clayton Colleen even
19:53 EmilienM crinkle, mgagne : 'next' branch > -1 on my side
19:53 EmilienM I have a problem with that
19:53 EmilienM nobody in OpenStack projects is doing that AFIK
19:53 mgagne and it's just lying to ourselves
19:53 EmilienM though they have RC
19:53 EmilienM mgagne: Nova has 2015.1.rc1
19:53 EmilienM (tags)
19:53 clayton sure, but that's a tag, not a branch
19:54 mgagne "don't mess too much with master because I use it for stable-2" alright, we will move our dev to next then
19:54 EmilienM clayton: we should be consistent with OpenStack
19:54 mgagne /s
19:54 EmilienM there is a branch milestone-proposed and proposed/juno
19:55 EmilienM clayton: any idea of what is it?
19:55 mgagne EmilienM: it's where rc goes until it's stable and released
19:55 clayton I suspect milestone-proposed would be the dev milestones during the dev cycle
19:55 clayton and proposed/juno would be for RCs
19:55 mgagne AFAIK, milestone-proposed is the old way and has been replaced with proposed/*
19:56 crinkle EmilienM: mgagne okay I'm giving up on the next branch idea
19:56 clayton I'm completely fine with continuing to treat master the way it always has been.  I'd prefer to focus the discussion on how to make upgrades easier than they currently are.
19:56 crinkle I think as long as beaker is able to install packages from somewhere then that should be the date we're allowed to switch over to working on the next release
19:57 mgagne clayton: yes, I would prefer this kind of discussion
19:57 clayton we've got some concrete ideas on how we could improve there, and some more vague ideas that have promise
19:57 mgagne clayton: I'm not against making it possible. I'm -1 for the idea of stopping what we are currently doing in master to support a use case that was never officially supported. that's unfair
19:58 clayton I don't think anyone is saying we should do that.  At least Matt & I aren't
19:58 clayton with the potential exception of config file deprecations
19:58 mgagne stuff got reverted from master. that's a strong statement to me
19:58 clayton yes, that was a month or two ago
19:58 clayton and it was a config file deprecation
19:59 clayton the config file deprecation discussion is more of a upgrade discussion, not a master vs stable branch discussion
20:00 EmilienM clayton: well... it's still how you deal with master
20:00 mgagne for me, it's part of the process
20:00 mgagne we shouldn't deliver deprecated stuff
20:00 clayton EmilienM: only in that master is what will eventually be in the stable branch.
20:00 mgagne in the past, people came from smokestack and packstack updating our stuff because we used a deprecated config
20:01 clayton we should be able to automate a good bit of that from beaker runs
20:01 clayton at least eventually, in theory
20:01 EmilienM clayton: automate what?
20:01 clayton detection of deprecations
20:02 clayton instead of relying on the docs, and people pointing them out
20:02 openstackgerrit Merged stackforge/puppet-monasca: Make api server have configurable backend db  https://review.openstack.org/180652
20:03 clayton I thought Dan made a good point the other day, we do a poor job of documenting deprecations.  I don't think its' reasonable to expect people to grep all the manifests and hope all the deprecations are worded the same way
20:04 EmilienM mgagne, clayton: the goal of my BP now is to track problem.
20:04 clayton dunno, maybe they're all called out in the README.md
20:04 richm is there a standard format for puppet api docs and deprecations?  If so, is that puppet-lintable?
20:05 clayton there is kind of for the api, we already have some lint checks for that
20:05 clayton it's changing (at least somewhat) in puppet 4.x
20:05 mgagne clayton: which deprecation? openstack itself or puppet?
20:06 mgagne clayton: https://wiki.openstack.org/wik​i/Puppet/Deprecated_Parameters
20:07 mgagne clayton: and wording used to be *VERY* consistent
20:08 mgagne The XXX parameter is deprecated, use YYY instead. or
20:08 mgagne The XXX parameter is deprecated and has no effect.
20:09 clayton well, we've definitely not been doing it recently.  None of the identity uri stuff is in there or in the README
20:09 mgagne what we can do, is create a custom function to report deprecation in OUR parameters. And find a way to make them fatal for people that wishes to make them fatal instead of warning.
20:10 clayton I've never seen that page, how does one find that?
20:10 mgagne clayton: google
20:10 clayton mgagne: I like that idea
20:10 clayton should be linked somewhere off the puppet wiki page
20:10 mgagne clayton: you don,t need README or anything humanly maintained. your puppet logs should be enough to catch them
20:11 mgagne clayton: that's what I do and it works very fine. we are about to do the same with openstack deprecations anyway
20:11 mgagne don't know why operators can't do the same
20:11 clayton Well, that's what we did last time, I found it to be a giant pain
20:11 mgagne clayton: define pain. I ran one puppet run on all my boxes, grep deprecate  and updated everything in 15m
20:18 openstackgerrit Merged stackforge/puppet-trove: Beaker tests  https://review.openstack.org/179088
20:19 openstackgerrit Merged stackforge/puppet-designate: Beaker tests  https://review.openstack.org/178366
20:20 openstackgerrit Merged stackforge/puppet-cinder: Beaker tests  https://review.openstack.org/177601
20:20 openstackgerrit Merged stackforge/puppet-cinder: Clean Gemfile: drop rspec-puppet and rake  https://review.openstack.org/180004
20:20 openstackgerrit Emilien Macchi proposed stackforge/puppet-sahara: Beaker tests  https://review.openstack.org/178821
20:23 dgurtner joined #puppet-openstack
20:23 dgurtner joined #puppet-openstack
20:24 openstackgerrit Emilien Macchi proposed stackforge/puppet-horizon: Beaker tests  https://review.openstack.org/177253
20:28 EmilienM crinkle: what you propose with the new branch, is the proposed/<future-stable> ( https://wiki.openstack.org/wiki/Branch_Model ) ?
20:29 crinkle EmilienM: no I give up on that
20:29 crinkle I propose nothing
20:29 EmilienM lol
20:29 crinkle EmilienM: as long as beaker works i'm happy
20:29 EmilienM you made my day
20:29 EmilienM @all
20:30 EmilienM crinkle: because beaker jobs will rely on packaging (last very new stable or very-soon stable) - right?
20:31 crinkle EmilienM: right
20:31 crinkle so as long as there is a repo to point beaker at then that works for me
20:36 bogdando joined #puppet-openstack
20:42 openstackgerrit Adam Vinsh proposed stackforge/puppet-swift: Add support for dedicated replication network.  https://review.openstack.org/177037
20:58 xingchao joined #puppet-openstack
21:07 gildub joined #puppet-openstack
21:10 aimon_ joined #puppet-openstack
21:27 xingchao joined #puppet-openstack
21:27 EmilienM mfisch, clayton: I spent some time to think about your upgrade topic
21:27 EmilienM and I've got an idea
21:28 EmilienM but I'm not sure it's possible with beaker
21:28 EmilienM crinkle: would it be possible to run beaker on stable/kilo (when it's done) and then run again on the same node on master ?
21:28 EmilienM we could test upgrades that way (?)
21:30 crinkle maybe
21:30 crinkle i'm not sure how to make beaker aware of branches
21:33 EmilienM crinkle: well, if there is some dev, I'm all in to push
21:33 EmilienM crinkle: I think at a grenade-like job
21:33 EmilienM is it crazy?
21:33 EmilienM https://wiki.openstack.org/wiki/Grenade
21:43 EmilienM richm: https://review.openstack.org/#/c/178385 is going to change after Gilles's refacto ?
22:06 britthou_ joined #puppet-openstack
22:09 _nick joined #puppet-openstack
22:13 richm EmilienM: yes
22:13 richm I think it should change anyway
22:13 richm I don't like using OS_DOMAIN_NAME and OS_DEFAULT_DOMAIN
22:17 mdorman joined #puppet-openstack
22:21 openstackgerrit Ryan Bak proposed stackforge/puppet-monasca: Add check setup for system checks.  https://review.openstack.org/180777
22:32 ddmitriev joined #puppet-openstack
22:43 openstackgerrit Ryan Bak proposed stackforge/puppet-monasca: Add check setup for system checks.  https://review.openstack.org/180777
23:01 openstackgerrit Merged stackforge/puppet-monasca: Add check setup for system checks.  https://review.openstack.org/180777
23:05 britthouser joined #puppet-openstack
23:18 panda joined #puppet-openstack
23:28 xingchao joined #puppet-openstack
23:43 openstackgerrit Sam Morrison proposed stackforge/puppet-neutron: Don't specify a nova region by default  https://review.openstack.org/180378
23:44 ducttape_ joined #puppet-openstack
23:45 ericpeterson joined #puppet-openstack
23:48 stamak joined #puppet-openstack

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