Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-09-30

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

All times shown according to UTC.

Time Nick Message
00:02 xingchao joined #puppet-openstack
00:02 openstackgerrit joined #puppet-openstack
00:04 openstack joined #puppet-openstack
00:07 openstackgerrit Gilles Dubreuil proposed openstack/puppet-keystone: Deletes tenant parameter from keystone_user  https://review.openstack.org/228744
00:10 Guest62975 joined #puppet-openstack
00:21 xingchao joined #puppet-openstack
00:47 TC01 joined #puppet-openstack
01:25 mfisch _ody: clayton and I are there
01:25 Guest35835 joined #puppet-openstack
01:32 openstackgerrit Matt Fischer proposed openstack/puppet-nova: redefine the virsh secret if needed  https://review.openstack.org/229167
01:45 openstackgerrit Matt Fischer proposed openstack/puppet-heat: Make the role for heat_stack_user configurable  https://review.openstack.org/225533
01:50 richm gildub: hello
02:07 jfluhmann joined #puppet-openstack
02:17 gildub richm, never mind, just saw your comment about doing indirection fix before, for kilo
02:18 richm gildub: yeah, I forgot about that :P
02:18 gildub richm, no worries, although that would have been *way* less painful ;)
02:20 richm yeah
02:28 openstackgerrit Gilles Dubreuil proposed openstack/puppet-keystone: Replace indirection calls  https://review.openstack.org/226624
02:37 openstackgerrit Gilles Dubreuil proposed openstack/puppet-keystone: Replace indirection calls  https://review.openstack.org/226624
02:49 gildub richm, chem, I looked at 226919, this is nice.
03:27 gildub michchap, hi - You're doing such a good review job I can't stop coming back to you :)
03:28 gildub michchap, could you please review those: https://review.openstack.org/226624 and  https://review.openstack.org/228744 ?
03:28 michchap gildub: sure :)
03:29 gildub michchap, Thanks! The reason 226624 was done before 228744 is because it might be back ported to kilo, the puppet indirection is *bad* (efficiency and not apparently not supported)
03:30 gildub michchap, I'm saying that because the second patch removes most of code fixed in the first one. It sounds weird :(
03:30 michchap gildub: that makes sense, thanks.
03:31 gildub michchap, cool :)
03:38 v1k0d3n joined #puppet-openstack
03:59 rahulait joined #puppet-openstack
04:18 sanjayu joined #puppet-openstack
04:27 gildub joined #puppet-openstack
04:40 sanjayu joined #puppet-openstack
05:20 LimorStotland joined #puppet-openstack
05:24 sanjayu joined #puppet-openstack
06:04 itsuugo joined #puppet-openstack
06:05 leanderthal joined #puppet-openstack
06:08 vinsh joined #puppet-openstack
06:09 LimorStotland joined #puppet-openstack
06:20 xarses joined #puppet-openstack
06:22 xarses joined #puppet-openstack
06:23 xarses joined #puppet-openstack
06:24 xarses joined #puppet-openstack
06:32 sergmelikyan joined #puppet-openstack
06:38 sanjayu joined #puppet-openstack
06:39 openstackgerrit Gilles Dubreuil proposed openstack/puppet-keystone: Add semantic title for keystone_tenant.  https://review.openstack.org/226919
07:03 zigo Hi guys !
07:04 zigo I have Liberty rc1 fully packaged and fully working, my CI passes tempest tests, minus some setup issues.
07:04 zigo I wonder how I can help to test Debian more using upstream puppet stuff.
07:04 zigo I just read that, once more, there's some work needed to be done on upstream puppet things, so that it can support both Debian & Ubuntu.
07:05 zigo EmilienM: Hello ! ^
07:08 arnaud_orange joined #puppet-openstack
07:09 rahulait joined #puppet-openstack
07:14 khushbu_ joined #puppet-openstack
07:33 sergmelikyan joined #puppet-openstack
07:33 dgurtner joined #puppet-openstack
07:33 dgurtner joined #puppet-openstack
07:34 itsuugo joined #puppet-openstack
07:36 jpena joined #puppet-openstack
07:53 xarses joined #puppet-openstack
07:53 openstackgerrit David Gurtner proposed stackforge/puppet-ceph: Be more specific with CentOS7 package exclude  https://review.openstack.org/214093
07:55 xarses_ joined #puppet-openstack
07:57 sergmelikyan joined #puppet-openstack
07:59 gfidente joined #puppet-openstack
08:02 jistr joined #puppet-openstack
08:09 openstackgerrit Javier Peña proposed openstack/puppet-swift: Add support for swift-object-expirer service  https://review.openstack.org/211549
08:15 khushbu_ joined #puppet-openstack
08:16 derekh joined #puppet-openstack
08:21 rochaporto joined #puppet-openstack
08:28 linkedinyou joined #puppet-openstack
08:43 paramite joined #puppet-openstack
08:53 sanjayu joined #puppet-openstack
09:01 asabbas joined #puppet-openstack
09:02 khushbu joined #puppet-openstack
09:03 openstackgerrit Maksym Yatsenko proposed openstack/puppet-nova: [WIP]  https://review.openstack.org/229308
09:21 markvoelker joined #puppet-openstack
09:47 khushbu joined #puppet-openstack
09:56 itsuugo joined #puppet-openstack
10:11 khushbu joined #puppet-openstack
10:23 sergmelikyan joined #puppet-openstack
10:36 LimorStotland joined #puppet-openstack
11:22 markvoelker joined #puppet-openstack
11:25 mjblack joined #puppet-openstack
11:30 khushbu_ joined #puppet-openstack
11:31 sergmelikyan joined #puppet-openstack
11:38 aarefiev joined #puppet-openstack
11:39 cdelatte joined #puppet-openstack
11:53 iurygregory joined #puppet-openstack
11:53 chem joined #puppet-openstack
11:55 iurygregory good morning :D
11:56 morazi joined #puppet-openstack
12:01 arnaud_orange joined #puppet-openstack
12:03 markvoelker joined #puppet-openstack
12:04 jtomasek joined #puppet-openstack
12:08 dprince joined #puppet-openstack
12:12 iurygregory anyone can take a look in: https://review.openstack.org/#/c/223777/ https://review.openstack.org/#/c/208054/ =)
12:13 ducttape_ joined #puppet-openstack
12:16 LimorStotland joined #puppet-openstack
12:17 EmilienM good morning
12:19 iurygregory o/
12:40 openstackgerrit Adam Vinsh proposed openstack/puppet-swift: Manage swift with swiftinit service provider  https://review.openstack.org/203220
12:41 jtomasek joined #puppet-openstack
12:46 itsuugo joined #puppet-openstack
12:48 vinsh_ joined #puppet-openstack
12:51 v1k0d3n joined #puppet-openstack
12:51 richm joined #puppet-openstack
13:05 IBerezovskiy joined #puppet-openstack
13:19 xingchao joined #puppet-openstack
13:21 vinsh joined #puppet-openstack
13:35 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: Cleanup some legacy code  https://review.openstack.org/229071
13:35 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: run_tests: drop dupplicated code  https://review.openstack.org/229426
13:37 openstackgerrit Merged openstack/puppet-openstack-integration: scenario001: enable object service  https://review.openstack.org/228450
13:44 v1k0d3n joined #puppet-openstack
13:47 nihilifer joined #puppet-openstack
13:47 nihilifer joined #puppet-openstack
13:51 ducttape_ joined #puppet-openstack
13:53 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-cookiecutter: Updates on config provider  https://review.openstack.org/229439
13:54 v1k0d3n joined #puppet-openstack
13:56 openstackgerrit Emilien Macchi proposed openstack/puppet-aodh: puppet-aodh: Initial commit  https://review.openstack.org/225015
13:56 arnaud_orange joined #puppet-openstack
14:03 openstackgerrit Emilien Macchi proposed openstack/puppet-tempest: Allow to enable Trove tests  https://review.openstack.org/229445
14:04 sergmelikyan joined #puppet-openstack
14:06 dmsimard EmilienM: is aodh available in liberty ?
14:06 dmsimard I haven't been following it too much - just saw your comment about it not being packaged yet
14:07 dfisher joined #puppet-openstack
14:08 itsuugo joined #puppet-openstack
14:08 EmilienM dmsimard: yeah, it's in progress by pradk
14:09 dmsimard EmilienM: ok
14:09 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: scenario001: deploy & test Trove  https://review.openstack.org/229450
14:20 ferest joined #puppet-openstack
14:23 ducttape_ joined #puppet-openstack
14:24 openstackgerrit Athlan-Guyot sofer proposed openstack/puppet-keystone: Add semantic title for keystone_tenant.  https://review.openstack.org/226919
14:37 openstackgerrit Athlan-Guyot sofer proposed openstack/puppet-keystone: Add semantic title for keystone_tenant.  https://review.openstack.org/226919
14:57 bklei joined #puppet-openstack
14:57 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: scenario001: deploy & test Trove  https://review.openstack.org/229450
15:08 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: scenario001: use 2 workers for sahara-api  https://review.openstack.org/229470
15:12 openstackgerrit Jay_Clark proposed openstack/puppet-nova: Fix for Bug #1484402  https://review.openstack.org/229472
15:12 openstack bug 1484402 in puppet-nova "no interface for modify the port of nova-api " [Low,Confirmed] https://launchpad.net/bugs/1484402
15:15 mdorman joined #puppet-openstack
15:17 openstackgerrit Jay_Clark proposed openstack/puppet-nova: Removing merge kruft.  https://review.openstack.org/229475
15:54 EmilienM crinkle: https://review.openstack.org/#/c/229071/ is tested against all modules (see https://goo.gl/7AjoiR ) - I think we can safely merge it
15:55 EmilienM crinkle: which means we don't have the legacy code as blocker anymore for the fixtures topic.
15:55 dmsimard EmilienM: hmmm, doesn't look like that worked, still no /etc/ceph in http://logs.openstack.org/99/229099/2/check/gate-puppet-ceph-puppet-beaker-rspec-dsvm-centos7/5f41541/
15:55 EmilienM dmsimard: the nodepool images are nightly built
15:55 EmilienM check on http://nodepool.openstack.org/
15:56 dmsimard oh, so tomorrow then
15:56 EmilienM maybe
15:56 EmilienM if build works
16:01 bkero dmsimard: Ever figure out those swift issues? I didn't see anything new in the logs.
16:02 dmsimard bkero: no, the improved log fetch won't be in until tomorrow and I haven't looked tbh
16:02 dmsimard there might be a legit issue
16:03 bkero yeah
16:03 bkero I saw your recheck earlier, so I thought you might have encountered the issue on something else or it might have been transient.
16:06 bkero Something like "[Wed Sep 30 14:59:50.371504 2015] [core:warn] [pid 22501] (13)Permission denied: AH00056: connect to listener on [::]:8080"
16:07 EmilienM dmsimard: just fyi - we're about to re-enabling selinux (enforced mode)
16:07 dmsimard EmilienM: ok
16:07 dmsimard EmilienM: the new package is already in
16:07 EmilienM I know and it works
16:12 EmilienM if our team could review https://goo.gl/tljZvW and revisit https://review.openstack.org/226830 - thanks
16:13 EmilienM nibalizer: I don't get your comment on https://review.openstack.org/226830 - have you read the thread about it?
16:20 rahulait joined #puppet-openstack
16:21 mentat joined #puppet-openstack
16:22 EmilienM I dropped a comment
16:52 ilbot3 joined #puppet-openstack
16:52 Topic for #puppet-openstack is now Puppet OpenStack modules - https://wiki.openstack.org/wiki/Puppet
16:53 wabu_ left #puppet-openstack
17:00 degorenko EmilienM, hey, take a look please https://review.openstack.org/#/c/219275/
17:13 LimorStotland joined #puppet-openstack
17:14 crinkle EmilienM: why are the puppet-openstack-integration tests failing? these don't look like packaging issues to me http://logs.openstack.org/26/229426/1/check/gate-puppet-openstack-integration-dsvm-trusty/53c4af5/console.html#_2015-09-30_16_16_46_035 http://logs.openstack.org/26/229426/1/check/gate-puppet-openstack-integration-dsvm-centos7/f41d774/console.html#_2015-09-30_14_06_27_345
17:18 EmilienM crinkle: the first link is a race condition I also discovered yesterday
17:18 EmilienM I have no idea why ceilometer/eventlet is started before apache, since we disable it in puppet
17:19 crinkle ubuntu starts its services at boot
17:19 EmilienM https://github.com/openstack/puppet-ceilometer/blob/master/manifests/api.pp#L134-L138
17:19 crinkle er at installation
17:19 EmilienM crinkle: this it what I thought too, but I could not reproduce in my env
17:20 EmilienM I tried to remove/install package and the service was not started.
17:20 EmilienM anyway, we can make sure the service is stopped in Puppet but this is already the case here: https://github.com/openstack/puppet-ceilometer/blob/master/manifests/api.pp#L134-L138
17:20 EmilienM so I don't know how to deal with that issue yet.
17:20 EmilienM crinkle: the second link is happenning randomly.
17:21 EmilienM our integration jobs are pretty unstable now, because of some tempest tests/scenarios that sometimes fail or work
17:21 EmilienM it's probably bugs in OpenStack, I would investigate later
17:21 crinkle these issues should have been discovered before that code was merged, now i can't trust the test results when reviewing patches for that repo
17:23 EmilienM crinkle: they used to pass CI
17:23 openstackgerrit Athlan-Guyot sofer proposed openstack/puppet-keystone: Add semantic title for keystone_tenant.  https://review.openstack.org/226919
17:25 EmilienM crinkle: I might be wrong, but I think our job is installing too much services now
17:25 EmilienM and it could lead to unstability
17:28 EmilienM crinkle: I tried this: https://review.openstack.org/#/c/228597/ and it failed because of dependency cycle issue
17:28 EmilienM I should restore it
17:30 openstackgerrit Emilien Macchi proposed openstack/puppet-ceilometer: wsgi: make sure eventlet process is stopped before httpd  https://review.openstack.org/228597
17:31 openstackgerrit Colleen Murphy proposed openstack/puppet-ceilometer: Ensure ceilometer is stopped before starting httpd  https://review.openstack.org/229538
17:31 xingchao joined #puppet-openstack
17:32 EmilienM crinkle: we have the same patch ^ lol
17:32 crinkle oh i'll drop mine
17:33 EmilienM crinkle: so I tried this on my env, and it failed (I used before =>) will see with ->
17:36 EmilienM Gem::InstallError: fog-google requires Ruby version >= 2.0.
17:36 EmilienM https://jenkins01.openstack.org/job/gate-puppet-ceilometer-puppet-beaker-rspec-dsvm-trusty/54/console
17:37 EmilienM we have a problem \o/
17:38 EmilienM "Houston..."
17:38 EmilienM our CI is broken
17:38 EmilienM crinkle: https://github.com/fog/fog-google/commit/a66b16fa7c2373f9c8be2e80bc942ad8d13ece3f
17:39 crinkle >.<
17:39 EmilienM we need to pin the gem I guess
17:39 EmilienM patch all our modules
17:40 EmilienM trusty is running 1.9
17:40 EmilienM trusty is running ruby 1.9
17:41 dgurtner joined #puppet-openstack
17:41 dgurtner joined #puppet-openstack
17:42 EmilienM as a workaround until Trusty uses ruby 2.0, I suggest we pin fog-google
17:42 crinkle trusty is never going to use 2.0
17:43 crinkle i'm going to file a bug with beaker so they can deal with it
17:44 EmilienM crinkle: the only requirement is https://github.com/puppetlabs/beaker/blob/master/beaker.gemspec#L54
17:45 EmilienM so I guess we can pin fog to ~> 1.25 but not using too recent fog-google
17:46 EmilienM even here, the constraint is too large: https://github.com/fog/fog/blob/v1.34.0/fog.gemspec#L60
17:47 EmilienM I wonder if we can pin fog-google then
17:47 nibalizer EmilienM: responded
17:48 EmilienM nibalizer: we need unit tests dependencies because we have libraries
17:48 EmilienM like openstacklib
17:48 EmilienM openstack has the same 'issue' with Oslo, they also use Depends-On for unit testing
17:49 EmilienM thanks to Global Requirements, we keep consistent dependencies
17:50 EmilienM nibalizer: if you patch openstacklib with a new Provider that will be consummed in puppet-keystone, your patch in puppet-keystone will fail because unit test can't pass since the patch in openstacklib is not merged
17:50 openstackgerrit Denis Egorenko proposed openstack/puppet-nova: Rewrite nova_network provider with using only nova client  https://review.openstack.org/229548
17:51 EmilienM nibalizer: I agree with you that this pattern I propose is not used by Puppet community. Do you have another proposal then?
17:53 nibalizer well you can't merge it even with a depends-on
17:53 nibalizer you still have to wait until the patch adding a provider to openstacklib lands
17:55 EmilienM nibalizer: it's not a merge issue.
17:55 EmilienM nibalizer: it's about testing a feature or bugfix
17:55 EmilienM this feature is a "nice to have", we could probably live without it
17:56 EmilienM it's just something our community is asking - cc mwhahaha and some others
17:56 EmilienM and I agree it's a nice feature, very useful
17:58 openstackgerrit Emilien Macchi proposed openstack/puppet-ceilometer: Gemfile: pin fog-google to ~> 0.1.0  https://review.openstack.org/229552
17:58 EmilienM crinkle: in the meantime ^
17:58 mwhahaha you can workflow a depends-on even if the item it depends on hasn't been merged yet. it just won't land until the other change. the issue is being able to do cross module CI testing for inflight changes
17:59 EmilienM yes, like we like doing with our beaker jobs, and recently with integration jobs
17:59 mwhahaha since there's so many interdependencies between openstack stuff
17:59 EmilienM crinkle: do you think the dependency can be changed in puppetlabs/beaker ?
18:01 nibalizer EmilienM: do you have a link to how the openstack folk do it with oslo and depends-on?
18:01 nibalizer EmilienM: i see the value in the feature
18:02 nibalizer as in that ML thread, its a tradeoff between features and complexity
18:02 EmilienM nibalizer: https://rbtcollins.wordpress.com/2015/04/28/dealing-with-deps-in-openstack/
18:02 nibalizer complexity hurts the project by making it harder to debug and harder for newcommers
18:02 crinkle EmilienM: I filed https://tickets.puppetlabs.com/browse/BKR-564 and pinged anode
18:03 EmilienM nibalizer: otherwise, let's create a rake task for OpenStack Infra
18:03 EmilienM nibalizer: and change our jobs to run this task
18:03 EmilienM nibalizer: so we keep the fixtures and everyone is happy
18:05 nibalizer so that blog isn't about this
18:06 nibalizer thats about python transitive dependencys and dependency solving
18:06 nibalizer also its about installing into a virtualenv and running i think devstack/tempest, not e.g. the nova unit tests
18:07 EmilienM nibalizer: right, I pasted the wrong link
18:07 EmilienM crinkle: thanks for the bug
18:07 dprince joined #puppet-openstack
18:07 EmilienM nibalizer: so, what's your suggestion ?
18:08 nibalizer what's wrong with doing nothing?
18:10 EmilienM nibalizer: seriously?
18:11 EmilienM nibalizer: in my comment on https://review.openstack.org/#/c/226830/ - I explained you why I did that, we can't just ignore the facts, can we?
18:12 EmilienM 1/ unit tests are not run against the same dependencies as functional tests
18:12 EmilienM this is a big issue, last week I ran a bug where puppet-qpid code in puppet-nova was too old - we did not track it because fixtures pin an old release, beaker a new one but we don't test QPID
18:13 EmilienM that means our code is potentially not passing unit tests when using the same dependencies as beaker/integration
18:13 nibalizer well the definition of a bug is a problem in code that has landed imho
18:13 nibalizer if you need depends-on its because there is code that hasn't landed yet
18:13 EmilienM it's not about depends-on now
18:14 EmilienM 2/ unit tests are not using the same dependencies (fixtures.yaml is not consistent across all modules !)
18:14 nibalizer what is it about
18:14 EmilienM nibalizer: I wrote it in Gerrit
18:14 EmilienM and 3/ because we use fixtures, our patches can't use Depends-On zuul feature for unit tests
18:14 nibalizer in the commit messaeg?
18:14 nibalizer or in the discussion
18:14 EmilienM no, in PatchSet 24 comment
18:14 EmilienM I wrote a comment for you
18:14 EmilienM to explain our motivation
18:14 EmilienM to make it
18:15 EmilienM nibalizer: how do you solve 1/ and 2/ ?
18:16 nibalizer i have to get lunch
18:16 nibalizer ill pop on after and we can continue this
18:16 EmilienM me too, it's 2.15pm
18:16 openstackgerrit John Trowbridge proposed openstack/puppet-ironic: Add ironic-inspector support  https://review.openstack.org/223690
18:16 EmilienM nibalizer: i would like to continue that discussion after
18:17 openstackgerrit joined #puppet-openstack
18:31 openstackgerrit John Trowbridge proposed openstack/puppet-ironic: Add ironic-inspector support  https://review.openstack.org/223690
18:31 ibravo joined #puppet-openstack
18:44 _ducttape_ joined #puppet-openstack
18:47 openstackgerrit joined #puppet-openstack
18:50 mwhahaha EmilienM: I remember you updated the apt requirements to 2.1.x in opesntack-integration, what was that for?
18:53 crinkle EmilienM: looks like puppetlabs is going to fix and release beaker soonish
18:53 crinkle so we might not have to do anything
18:55 EmilienM crinkle: they are very responsive, it's awesome
18:55 ducttape_ joined #puppet-openstack
18:55 EmilienM mwhahaha: I updated the Puppetfile yes, so we can test puppetlabs-apt 2.x
18:56 EmilienM mwhahaha: any issue?
18:56 mwhahaha No just wondering if we needed to do the same
18:56 mwhahaha didn't know if there was a technical reason
18:56 EmilienM crinkle: https://github.com/puppetlabs/beaker/pull/973
18:56 mwhahaha we're using 1.8.0 and i just happen to be messing with our puppetfile
18:57 linkedinyou joined #puppet-openstack
18:58 EmilienM mwhahaha: it was a major update for the module, we wanted our modules using it so we stay on the edge
18:58 mwhahaha ok
18:58 mwhahaha thanks
18:58 EmilienM mwhahaha: because it's important for us to make sure our module work with our dependencies
18:59 EmilienM the Puppetfile is willing to be updated quite often I guess
18:59 EmilienM _ody: it's odd we can't see your CI http://jenkins-beaker.delivery.puppetlabs.net/job/qe_beaker_btc-intn/1587/
19:07 _ody EmilienM: Nigel sent out an email to puppet-dev a while back, we had to shut off public access temporarily.  https://groups.google.com/forum/#!searchin/puppet-dev/jenkins/puppet-dev/24WeTcFk8Co/tONeSrl9BQAJ
19:08 EmilienM _ody: thx
19:10 ducttape_ joined #puppet-openstack
19:11 _ody I need a couple Ansible users or ex-users willing to volunteer to be interviewed by our UX and product teams.  Please just PM me if you're willing.
19:12 xarses_ joined #puppet-openstack
19:12 _ody Actually...doesn't have to be Ansible, things like fabric or rundeck are valid too.
19:13 openstackgerrit John Trowbridge proposed openstack/puppet-ironic: Add ironic-inspector support  https://review.openstack.org/223690
19:14 _ducttape_ joined #puppet-openstack
19:20 EmilienM crinkle: re: integration job failures - on centos, it's seems random though, like this one: http://logs.openstack.org/70/229470/1/check/gate-puppet-openstack-integration-dsvm-centos7/e374fd1/console.html#_2015-09-30_18_38_48_171
19:21 EmilienM it looks like it comes from here: http://logs.openstack.org/70/229470/1/check/gate-puppet-openstack-integration-dsvm-centos7/e374fd1/logs/nova/nova-compute.txt.gz#_2015-09-30_18_38_08_067
19:22 crinkle EmilienM: the difference I see between that http://logs.openstack.org/71/229071/4/check/gate-puppet-openstack-integration-dsvm-centos7/482718e/console.html is hpcloud vs rax
19:22 crinkle might be related
19:22 EmilienM crinkle: we have the same flavors, right?
19:23 crinkle i don't know what's different between them
19:24 EmilienM well, we can easily find out
19:24 EmilienM crinkle: to me, it's still random because the failures ^^^^ are for a scenario that used to work, and it's failing randomly now
19:25 EmilienM both flavors have 8GB of Ram and 8 VCPU
19:25 crinkle okay
19:26 crinkle i guess you're right, maybe just putting too many things on one node
19:26 EmilienM crinkle: I should stop pushing more tests on integ
19:26 crinkle need to figure out how devstack gets away with it
19:26 EmilienM yeah
19:26 EmilienM crinkle: actually they don't have this issue because they don't install everything like we do
19:27 EmilienM ie: glance CI installs devstack with basic components, but not sahara for example
19:27 EmilienM in our case, we install everything
19:27 crinkle i thought that was the point of the integrated gate?
19:27 EmilienM maybe I missed something then
19:27 crinkle ¯\_(ツ)_/¯
19:27 EmilienM crinkle: do you have an example of job?
19:28 crinkle EmilienM: ?
19:28 EmilienM crinkle: do you have an example of integration job for OpenStack?
19:28 crinkle EmilienM: no i don't have more knowledge than you
19:28 crinkle maybe mtreinish can help us out?
19:28 EmilienM crinkle: look Glance for example: https://review.openstack.org/#/c/218869/
19:29 EmilienM they have gate-tempest-dsvm-full
19:29 EmilienM and it's only installing cinder, glance, keystone nova and swift;
19:29 mtreinish crinkle: with?
19:29 EmilienM mtreinish: we are having a lot of random failures in tempest
19:29 EmilienM http://logs.openstack.org/70/229470/1/check/gate-puppet-openstack-integration-dsvm-centos7/e374fd1/console.html#_2015-09-30_18_38_48_171 is an example
19:30 EmilienM they are random
19:30 EmilienM we think it could be because we're installing too much services on the same node
19:30 EmilienM mtreinish: we're installing/testing nova/neutron/keystone/cinder/glance/swift/sahara/heat/ceilometer (trove & horizon are wip)
19:31 mtreinish EmilienM: so my gut feeling is that you're putting too much on a single slave
19:31 EmilienM yes
19:31 mtreinish having cinder tests timeout at 300 sec might just mean things are moving too slowly
19:31 mtreinish same with the nova failure from the scheduler
19:31 EmilienM it makes sense
19:32 EmilienM we have several options
19:32 EmilienM 1/ increase timeout where we can and continue to live with that
19:33 EmilienM 2/ drop some services
19:33 EmilienM 3/ split scenario001 and have scenario002 - share services across 2 scenarios
19:36 EmilienM I'm exploring 1/ now, because 2/ is not what we want, and 3/ will consumme more CI jobs (we're already using an important number of jobs...)
19:39 openstackgerrit John Trowbridge proposed openstack/puppet-ironic: Add ironic-inspector support  https://review.openstack.org/223690
19:40 openstackgerrit Ramkumar Gowrishankar proposed openstack/puppet-neutron: Support for Nuage Neutron plugin and Nuage VRS in puppet-neutron  https://review.openstack.org/214798
19:57 ducttape_ joined #puppet-openstack
20:01 rahulait joined #puppet-openstack
20:04 EmilienM crinkle: beside nova timeouts, I noticed neutron server sometimes fails to create ports because it thinks ovs agent is down (heartbeat)
20:14 mtreinish EmilienM: so I actually think splitting the services is what you want to do
20:14 mtreinish I'd recommend something a bit more layered
20:14 mtreinish like base IaaS and then things like sahara, trove and  heat seperately
20:15 mtreinish it also seperates concerns a bit more, because the higher level stuff depends on a working IaaS cloud
20:15 mtreinish sry, I'm split like 4 ways right now my feedback loop might be a bit slow
20:16 EmilienM mtreinish: it's going to consume more CI resources though
20:17 mtreinish so
20:17 mtreinish EmilienM: tempest runs like 30 jobs :)
20:17 mtreinish you're still a far way off from that
20:17 EmilienM it's something we might consider, indeed
20:17 nibalizer EmilienM: ok i am back
20:18 EmilienM nibalizer: hope you had great lunch, I took some energy for this rainy day
20:18 nibalizer cool
20:19 EmilienM nibalizer: so let me summarize again: we have 3 problems
20:19 EmilienM documented in https://review.openstack.org/#/c/226830/ comments (see my reply to you)
20:20 EmilienM Depends-On feature is a "nice to have", which is one of the 3 problems I'm trying to solve.
20:20 EmilienM we still need to test modules against same dependencies in unit & functional tests
20:21 EmilienM and we still need to keep consistent the dependencies
20:21 nibalizer EmilienM: so for both 1 & 2 I think that's fine
20:21 nibalizer unit tests are local to the project
20:21 nibalizer they only matter to the project being tested
20:21 EmilienM "that's fine" == we can ignore the issues?
20:22 nibalizer that's what I am thinking
20:22 EmilienM I disagree here
20:22 nibalizer unit tests and acceptance tests are different tests
20:23 EmilienM they are different in the result, but using the same dependencies just make sense.
20:24 ducttape_ joined #puppet-openstack
20:25 EmilienM nibalizer: so for you, it does not matter to test puppet-keystone with different versions of puppetlabs-apache2 in unit & functional?
20:25 nibalizer so how does that happen?
20:26 nibalizer https://review.openstack.org/#/c/226830/24/.fixtures.yml it looks like nova doesn't even use apache?
20:26 nibalizer or didn't prior to your patch?
20:26 EmilienM nibalizer: functional testing is pulling Puppetfile that contains the version we really want to test. unit testing is using fixtures file in repos, that is does not contain the same version of functional testing, and is hard to control (not centralied)
20:26 EmilienM centralized*
20:27 EmilienM nibalizer: in progress: https://review.openstack.org/213315
20:28 nibalizer ok
20:28 nibalizer ya thats fair
20:28 nibalizer I mean it's not bad
20:29 nibalizer it's better
20:29 EmilienM I don't see what I'm doing wrong here
20:30 nibalizer I think a) it's fine to do it the old way
20:30 nibalizer but b) your way is better
20:30 nibalizer and c) the complexity I don't like
20:31 EmilienM nibalizer: since I'm doing puppet openstack, it very rarely happened to me to have to edit fixtures file locally
20:32 EmilienM this is the only "complex" thing we're changing
20:32 openstackgerrit Jay_Clark proposed openstack/puppet-nova: Fix for Bug #1484402  https://review.openstack.org/229475
20:32 openstack bug 1484402 in puppet-nova "no interface for modify the port of nova-api " [Low,Confirmed] https://launchpad.net/bugs/1484402 - Assigned to Jay_Clark (jay-clark)
20:32 nibalizer EmilienM: ok
20:32 nibalizer i'll stop objecting
20:32 EmilienM people will have to take Puppetfile from our repo, modify it at their wish and change env
20:32 nibalizer ok
20:33 nibalizer we'll document it
20:34 EmilienM let's give a try
20:34 EmilienM if we have bad feedback from our users, we can still revert
20:34 EmilienM or improve it on the fly
20:35 EmilienM mtreinish: I'm considering splitting now, I notices 4 kinds of timeouts that are going to be hard to fix...
20:35 EmilienM mtreinish: I'm just sending an emailt to ML to get feedback before
20:36 mtreinish EmilienM: yeah, you don't really want to up the timeouts either because you'll likely hit the per test timeout
20:36 nibalizer ya
20:36 mtreinish or the jenkins job timeout
20:36 EmilienM ++
20:36 EmilienM we knew that would happen
20:36 nibalizer im considering a ruby gem/function that is basically detect_and_zuul_clone_repo <reponame>
20:36 EmilienM it's just until now it was green :)
20:36 nibalizer that way some of the stuff in both this rakefile and all over spec_helper_acceptance can be shortened
20:37 nibalizer also google-fog/beaker is :(
20:37 EmilienM nibalizer: crinkle had a great idea, to create a gem, see https://review.openstack.org/229126
20:38 nibalizer oh nice
20:38 EmilienM yeah, it will help a lot for consistency
20:39 sergmelikyan joined #puppet-openstack
20:40 ibravo joined #puppet-openstack
20:40 _ducttape_ joined #puppet-openstack
20:44 nibalizer ya
20:45 nibalizer crinkle: see andreas on 229126
20:45 nibalizer EmilienM: too
20:46 EmilienM nibalizer: I replied, and it should not block. jeblair already told me in the past we could add repos without governance merge. because the proejct is already big tent
20:49 nibalizer ok
20:53 itsuugo joined #puppet-openstack
21:00 apuimedo joined #puppet-openstack
21:06 gildub joined #puppet-openstack
21:07 EmilienM crinkle: I'm seriously wondering if we should keep upgrade jobs for now
21:07 gildub EmilienM, hi - Please review https://review.openstack.org/226624
21:07 EmilienM crinkle: do you think their current state is useful to have in our CI ?
21:08 crinkle EmilienM: why are they broken?
21:08 EmilienM crinkle: most of them actually work
21:09 EmilienM crinkle: I'm just wondering how useful they are at this time
21:09 EmilienM crinkle: I was thinking, dropping them and replace by a second scenario if we decide to split our integration manifest
21:09 EmilienM crinkle: I haven't taken time in this cycle to work on upgrades - the upgrade jobs were really experimental
21:10 gildub Actually, any core member please review https://review.openstack.org/226624
21:10 crinkle EmilienM: i'm fine with removing the upgrade job
21:10 crinkle EmilienM: but i don't think it's an either-or necessarily
21:10 EmilienM crinkle: do you think it's acceptable to have 2 more jobs?
21:11 gildub and https://review.openstack.org/228744 too, please.
21:11 EmilienM for scenario002 (if we decide the split)
21:11 crinkle maybe?
21:12 crinkle ¯\_(ツ)_/¯
21:13 Guest37880 joined #puppet-openstack
21:22 openstackgerrit Jay_Clark proposed openstack/puppet-nova: Fix for Bug #1484402  https://review.openstack.org/229634
21:22 openstack bug 1484402 in puppet-nova "no interface for modify the port of nova-api " [Low,Confirmed] https://launchpad.net/bugs/1484402 - Assigned to Jay_Clark (jay-clark)
21:23 jasondotstar EmilienM: think I've fixed the issues, and added the spec tests. I'm sure you'll let me know if there's something else required... :-)
21:24 EmilienM jasondotstar: your commit message is a mess
21:24 EmilienM https://review.openstack.org/#/c/229634/1//COMMIT_MSG,cm
21:24 jasondotstar EmilienM: ok...
21:24 EmilienM you have 3 change-id
21:24 EmilienM your titile is not good
21:24 EmilienM let me give you an example
21:25 jasondotstar plz
21:26 EmilienM jasondotstar: https://review.openstack.org/#/c/226624/
21:26 EmilienM a title, a description and the "Closes-Bug: #XXX"
21:26 EmilienM the "Change-Id" thing is generated by git
21:26 crinkle jasondotstar: did you squash instead of fixup in your rebase perhaps?
21:26 jasondotstar crinkle: i did squash, yes.
21:27 jasondotstar hence the messy commit msg I assume :-/
21:27 crinkle that would mix the commit messages all together, if you choose fixup in the interactive rebase menu it will discard the commit message
21:27 * jasondotstar needs better git-fu
21:27 EmilienM jasondotstar: you have 3 patches also
21:27 jasondotstar git rebase -i HEAD~2  ?
21:28 crinkle jasondotstar: looks like HEAD~3 but yep
21:28 jasondotstar EmilienM: the last one is the only one needed tho, right? it includes the others
21:29 jasondotstar crinkle: i did two squashes trying to fix it....
21:29 jasondotstar EmilienM: other than tidying up the commit msg, are the tests correct?
21:30 sergmelikyan joined #puppet-openstack
21:33 EmilienM jasondotstar: yes
21:34 jasondotstar EmilienM: fair enough. I'll make fix the commit msg and abandon those earlier patches
21:34 jasondotstar s/make//
21:34 angdraug joined #puppet-openstack
21:34 EmilienM jasondotstar: any progress on zaqar?
21:35 jasondotstar EmilienM: indeed
21:35 jasondotstar it's WIP but take a look: https://review.openstack.org/#/c/229036/
21:35 jasondotstar EmilienM: ^
21:35 jasondotstar no tests yet.
21:35 EmilienM jasondotstar: I saw that
21:36 jasondotstar oh ok
21:36 jasondotstar the one nut to crack is fixing the pymongo update
21:36 EmilienM we have a bunch of modules without functional testing, i'm currently dressing a list and making sure we are working on that.
21:37 jasondotstar not much of a rubyist, but I followed the spec written for nova
21:37 jasondotstar so I think I can write the tests needed for zaqar
21:38 jasondotstar everything lands on a rpm-based node.... need to work on the deb stuff next for that module.
21:38 jasondotstar and finish building up the conf file.
21:39 jasondotstar re: writing tests - +1. perhaps I can cut a few teeth helping w/ that.
21:39 jasondotstar EmilienM: ^
21:48 openstackgerrit Jay_Clark proposed openstack/puppet-nova: Allows the modification of the nova-api ports  https://review.openstack.org/229644
21:50 openstackgerrit Guilherme Balzana proposed openstack/puppet-swift: Add swift-informant middleware support  https://review.openstack.org/229646
21:53 ducttape_ joined #puppet-openstack
21:56 EmilienM crinkle: wdyt about my comment on https://review.openstack.org/#/c/229646 ?
21:58 crinkle EmilienM: I think I agree, it looks like swift-informant isn't even an openstack project?
21:59 crinkle based on the github url it looks like maybe a rackspace project?
21:59 EmilienM crinkle: right
21:59 EmilienM I googled it and could not find any reference to upstream work
21:59 EmilienM we should not accept this kind of patch, because it's not encouraging upstream collaboration.
22:00 crinkle agree
22:03 EmilienM crinkle: the middleware is 3 years old - /me sighs
22:03 EmilienM it should be in https://github.com/openstack/swift/tree/master/swift/common/middleware
22:05 EmilienM crinkle: https://github.com/pandemicsyn/swift-informant/issues/17
22:16 EmilienM please review https://review.openstack.org/#/c/229426/ and https://review.openstack.org/#/c/229071/ -  Since now, we don't expect integration jobs "stable" anymore... (see thread on ML)
23:14 EmilienM angdraug: not much feedback about puppet-neutron/third party stuffs
23:14 EmilienM I'm preparing a patch now, there are a lot of third party resources (repo/packages/services)
23:26 angdraug EmilienM: apologies about taking so long to respond, I agree with your conclusion on the thread
23:27 openstackgerrit Emilien Macchi proposed openstack/puppet-neutron: disable third-party software management  https://review.openstack.org/229675
23:27 EmilienM angdraug: ^
23:29 angdraug +1
23:31 gildub crinkle, EmilienM, please review https://review.openstack.org/226624 and https://review.openstack.org/178456, so we get those boulders out of the way, thanks!
23:33 EmilienM not sure what we can do with https://review.openstack.org/#/c/178456/
23:33 angdraug added a link from 209997 to 229675
23:34 EmilienM gildub: https://review.openstack.org/#/c/226624/13/spec/acceptance/keystone_wsgi_apache_spec.rb,cm
23:34 EmilienM gildub: how can we know you're backward compatible?
23:34 EmilienM you're changing the interface
23:35 EmilienM raise(Puppet::Error, "Missing domain name for resource #{name}")
23:35 EmilienM iirc, if you use the previous interface, you have this error ^
23:36 gildub crinkle, EmilienM, sorry second link is wrong, I meant https://review.openstack.org/228744
23:36 EmilienM the 2 patches need a lot of attention
23:36 gildub EmilienM, that said a Keystone patch to support V3 endpoints in v2 seems on its way
23:36 EmilienM it's EOD for me
23:37 EmilienM I'm not sure we can merge https://review.openstack.org/#/c/226624/ right now
23:37 EmilienM is it backward compatible?
23:37 EmilienM gildub, richm ^
23:38 gildub EmilienM, the resource title has to be name::domain because:
23:39 gildub EmilienM, 1. The tenant parameter for ::keystone::resource::service_identity requires it
23:40 gildub EmilienM, 2. The title must be consistent across the catalog
23:41 EmilienM gildub: can you reply to my question?
23:41 gildub EmilienM, I'm talking for the acceptance test to be running green. Now the keystone_tenant/user provider can still use just <name> and pass domain as a parameter
23:41 EmilienM gildub: is it backward compatible?
23:41 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: test old keystone_tenant interface  https://review.openstack.org/229681
23:41 EmilienM testing it ^
23:42 gildub EmilienM, well until we got the upgrade test running nobody is going to be sure isn't?
23:43 EmilienM gildub: the what?
23:43 EmilienM my question is simple: can I use keystone_tenant old interface?
23:44 EmilienM gildub: like I'm testing in my patch ^
23:44 EmilienM if not, your work is not backward compatible and can't be merged.
23:44 gildub EmilienM, yes its backward compatible
23:45 EmilienM i've -2'ed the patch until my test is green
23:45 gildub EmilienM, great, I go to the beach for the rest of the day then.
23:45 EmilienM gildub: maybe but I don't see (or I missed it) tests that test it.
23:45 EmilienM gildub: ?
23:46 gildub EmilienM, It's not because the acceptance test change that it means anything to the backward support, the unit tests do
23:48 EmilienM gildub: maybe, I just don't see it, maybe you can highlight it?
23:51 EmilienM gildub: ?
23:51 gildub EmilienM, lines 165-207 => https://review.openstack.org/#/c/226624/13/spec/unit/provider/keystone_tenant/openstack_spec.rb
23:52 EmilienM indeed, unit tests seem good
23:52 EmilienM I have other concerns, I'm writting a review in Gerrit
23:53 gildub EmilienM, ok thanks, just looking forward to get that one out of the way to be able to move on.
23:53 EmilienM gildub: me too, don't worry
23:54 gildub EmilienM, k
23:54 EmilienM gildub: this code is really critical, since it's used everywhere in our modules - I just pay x2 attention
23:56 gildub EmilienM, fair enough!
23:56 EmilienM gildub: see my comment
23:56 gildub EmilienM, looking

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