Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-01-30

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

All times shown according to UTC.

Time Nick Message
00:05 cwolferh joined #puppet-openstack
00:25 fandi joined #puppet-openstack
00:28 xarses joined #puppet-openstack
00:39 derekh joined #puppet-openstack
01:35 markvoelker joined #puppet-openstack
01:46 rmoe joined #puppet-openstack
01:57 mdorman joined #puppet-openstack
02:04 xarses joined #puppet-openstack
02:11 nosleep77 joined #puppet-openstack
02:11 xingchao joined #puppet-openstack
02:14 fandi joined #puppet-openstack
03:29 badiane_ka joined #puppet-openstack
03:44 mfisch EmilienM: can I get an opinion on this small change?  https://review.openstack.org/#/c/150139/
04:00 rushiagr joined #puppet-openstack
04:04 sanjayu joined #puppet-openstack
04:33 gildub joined #puppet-openstack
04:40 openstackgerrit Matt Fischer proposed stackforge/puppet-neutron: Notify the ovs-agent service if the config changes  https://review.openstack.org/151534
05:31 fandi joined #puppet-openstack
06:32 xarses joined #puppet-openstack
06:37 fandi joined #puppet-openstack
07:09 fandi joined #puppet-openstack
07:23 e1mer joined #puppet-openstack
07:37 xarses joined #puppet-openstack
08:07 dgurtner joined #puppet-openstack
08:07 dgurtner joined #puppet-openstack
08:08 igajsin joined #puppet-openstack
08:09 dvorak joined #puppet-openstack
08:17 cwolferh joined #puppet-openstack
08:20 jpena joined #puppet-openstack
08:27 tfz joined #puppet-openstack
08:27 mmagr joined #puppet-openstack
08:50 dvorak joined #puppet-openstack
09:19 derekh joined #puppet-openstack
09:31 jistr joined #puppet-openstack
09:38 ebogdanov joined #puppet-openstack
09:40 stamak joined #puppet-openstack
09:59 cwolferh joined #puppet-openstack
10:10 dalgaaf joined #puppet-openstack
11:31 dgurtner xarses: dmsimard_away dachary CI is back :) https://review.openstack.org/#/c/140746/
11:36 rcallawa joined #puppet-openstack
11:40 cdent joined #puppet-openstack
11:41 rcallawa joined #puppet-openstack
11:47 openstackgerrit Bogdan Dobrelya proposed stackforge/puppet-openstack_extras: Fixes for new pacemaker versions  https://review.openstack.org/151609
11:50 openstackgerrit Bogdan Dobrelya proposed stackforge/puppet-openstack_extras: Fixes for new pacemaker versions  https://review.openstack.org/151609
11:51 rcallawa_ joined #puppet-openstack
11:59 openstackgerrit Bogdan Dobrelya proposed stackforge/puppet-openstack_extras: Fixes for new pacemaker versions  https://review.openstack.org/151609
12:16 rcallawa joined #puppet-openstack
12:26 dachary dgurtner: \o/\o/
12:52 flaviof joined #puppet-openstack
12:56 ebogdanov joined #puppet-openstack
13:02 berendt joined #puppet-openstack
13:15 openstackgerrit Benedikt Trefzer proposed stackforge/puppet-vswitch: remove openvswitch-datapath-dkms package.  https://review.openstack.org/151629
13:17 arnaud_orange joined #puppet-openstack
13:17 rcallawa joined #puppet-openstack
13:31 EmilienM hola
13:32 dprince joined #puppet-openstack
13:35 ebogdanov joined #puppet-openstack
13:37 markvoelker joined #puppet-openstack
13:52 markvoelker joined #puppet-openstack
14:02 morazi joined #puppet-openstack
14:13 richm joined #puppet-openstack
14:47 mschmitt joined #puppet-openstack
14:58 dmsimard dgurtner: Awesome. Good job and thanks
15:07 dmsimard I'll try and go through the review queue today to run the CI tests and merge stuff.
15:09 bogdando joined #puppet-openstack
15:12 EmilienM|afk dmsimard: I spent some time on stackforge/puppet-ceph, it's a very good work
15:16 dmsimard EmilienM|afk: :)
15:18 prad joined #puppet-openstack
15:22 markvoelker joined #puppet-openstack
15:25 pradk joined #puppet-openstack
15:25 markvoelker joined #puppet-openstack
15:30 kbyrne joined #puppet-openstack
15:31 kbyrne joined #puppet-openstack
15:35 openstackgerrit Merged stackforge/puppet-ceph: This fixes various issues with the centos CI tests  https://review.openstack.org/140746
15:36 openstackgerrit Merged stackforge/puppet-ceph: Rewrite the usecases to actually working scenarios  https://review.openstack.org/98730
15:41 openstackgerrit Merged stackforge/puppet-ceph: Enabling and fixing the rspec tests for profiles  https://review.openstack.org/143058
15:43 tfz joined #puppet-openstack
15:43 richm puppet api question - there is an autorequire method in types e.g. autorequire(:keystone_tenant) do name_of_tenant end
15:43 richm this will cause Keystone_tenant[name_of_tenant] to be a requirement
15:44 richm However, in some cases, I want there to be a requirement on a tenant OR a domain - that is, I want to have a conditional requirement
15:44 Robe joined #puppet-openstack
15:45 Robe hi
15:45 Robe we had quite some fun with the new openstack-client in the puppet-keystone module
15:47 Robe http://nopaste.narf.at/show/16396/ - runtime seems to be exponential to the amount of roles that exist
15:49 trash And users and tenants that exists.
15:49 richm Robe: how many users, tenants, and roles do you have?
15:50 trash https://github.com/stackforge/puppet-keystone/blame/stable/juno/lib/puppet/provider/keystone_user_role/openstack.rb#L123 is the part which takes the longest
15:51 Robe richm: 10 users, 3 tenants, 5 roles
15:51 Robe this is our dev environment
15:51 Robe in production we had runtimes close to one hour ;)
15:51 trash And on the production servers we could not even finish one puppet run after one hour...
15:52 trash In fact we had to cancel the puppetrun because it virtually never ended.
15:52 Robe could be related to https://bugs.launchpad.net/puppet-keystone/+bug/1224179
15:52 richm Robe: trash: crinkle has a patch out for review that may fix this problem - https://review.openstack.org/150200
15:52 Robe richm: testing
16:04 Robe richm: that makes it a bit better, still noticeable worse than 5.0.0: http://nopaste.narf.at/show/16407/
16:04 Robe can't tell how much of this is static overhead and how much of this will explode with higher object counts
16:06 mdorman joined #puppet-openstack
16:11 richm Robe: ok - can you please update the review with that information?
16:12 richm I think crinkle and mfisch would be interested in that information
16:12 mfisch yeah thats scary
16:13 Robe richm: yep, will do
16:13 Robe richm: we currently run the patch against production
16:13 mfisch we have 148*83*12
16:13 mfisch we're not up to top of tree on puppet-keystone right now
16:14 richm mfisch: so you haven't been testing with the latest puppet-keystone that uses openstackclient?
16:15 mfisch so I think we are using it, but I wont guarntee that, I'll check for sure in a bit
16:15 mfisch in the middle of an upgrade this morning
16:17 Robe no hurries
16:17 Robe but judging from the runtime I guess 150200 needs some additional love :)
16:18 mfisch does this scale by how many roles/users you have or by how many roles/users you have managed by pupepet?
16:19 Robe I think first
16:19 Robe we hardly manage resources with puppet
16:19 mfisch we've got maybe 5/6 users and 2/3 roles in puppet
16:20 mdorman joined this convo late, what was the test runtime using 150200 ?
16:20 mfisch we're running against commit 8ac2f4f05a6 - Jan 19
16:20 mfisch have not seen exponential
16:20 mdorman i’ve never been a big fan of caching all the user/role tuples.  i thought cirnkle refactored so it loaded most of them on demand?
16:21 trash It doesn't matter how many roles/users you manage with puppet, the provider will get all the users from keystone directly and check those
16:22 mdorman right.
16:41 Robe ok, we're running into https://review.openstack.org/#/c/151180/ as well, we can only test against in production on monday
16:43 Robe updated 150200 with our dev numbers
16:43 richm Robe: thanks
16:47 cdent joined #puppet-openstack
16:52 flaviof joined #puppet-openstack
17:01 dachary EmilienM: thanks to dgurtner & dmsimard your contributions to puppet-ceph will benefit  from integration tests :-)
17:03 EmilienM dachary: expect me on-board - I'm working on tripleo puppet elements that are using stackforge/puppet-ceph
17:04 dachary speaking of which, dgurtner updated the use cases
17:06 dachary https://github.com/stackforge/puppet-ceph/blob/master/USECASES.md changed today (the review went in after the CI was back)
17:06 EmilienM dachary: that's a great news, so I can start from here
17:06 dgurtner dmsimard: dachary thanks for all the reviews. feels good to finally get all those outstanding reviews merged :)
17:06 * dachary heading to https://sfconservancy.org/supporter/event.html
17:07 dmsimard dgurtner, dachary : Yeah, I'll hit the review queue pretty hard today
17:07 EmilienM dachary: I'm not surprised you go there ;-)
17:07 dachary :-D
17:09 rmoe joined #puppet-openstack
17:10 ericpeterson joined #puppet-openstack
17:21 dgurtner dachary: dmsimard I just noticed that the CI made a mistake on https://review.openstack.org/#/c/98730/
17:22 dgurtner it set the following run to success: http://paste.ubuntu.com/9958061/
17:22 dgurtner will need to 1) figure out why that happened. and 2) fix this asap
17:22 dmsimard Indeed, the test does mention 8 failures
17:23 dmsimard The vote part is easy
17:23 dmsimard It's based on the return code
17:24 dgurtner in any case, just rechecked all the other runs and they seemed all fine
17:24 dgurtner i.e. 0 failures out of 250 tests
17:29 rcallawa joined #puppet-openstack
17:33 rmoe joined #puppet-openstack
17:34 xarses joined #puppet-openstack
17:35 xarses dgurtner: yay
17:37 dgurtner joined #puppet-openstack
17:37 dgurtner joined #puppet-openstack
17:55 aimon joined #puppet-openstack
17:56 aimon joined #puppet-openstack
18:10 xavpaice joined #puppet-openstack
18:10 jamespd joined #puppet-openstack
18:11 ebogdanov joined #puppet-openstack
18:13 _nick joined #puppet-openstack
18:14 rcallawa joined #puppet-openstack
18:17 higgins joined #puppet-openstack
18:24 tchaypo joined #puppet-openstack
18:56 badiane_ka joined #puppet-openstack
19:07 arnaud_orange joined #puppet-openstack
19:36 dprince EmilienM: hey. Back to this SELinux issue you with w/ the ceilometer swift module the other day. Is the easiest thing to do just do add those SELinux permission commands to the ceilometer swift module?
19:39 dprince EmilienM: also, Eoghan left me some feedback for the ceilometer-expirer we are using: https://review.openstack.org/#/c/149376/4/puppet/overcloud_controller.pp,cm
19:39 openstackgerrit Mike Dorman proposed stackforge/puppet-openstacklib: Properly handle policy values containing spaces  https://review.openstack.org/151746
19:40 dprince EmilienM: I got that from puppet-openstack-cloud....
19:40 EmilienM dprince: I would rater fix selinux policies
19:40 dprince EmilienM: right, so I expect the issue is where does this go?
19:40 EmilienM and ack for https://review.openstack.org/#/c/149376/4/puppet/overcloud_controller.pp,cm
19:40 EmilienM thx
19:41 dprince EmilienM: I'll start a thread with the relevant packages on this and copy you
19:45 EmilienM dprince: thx
20:07 openstackgerrit joined #puppet-openstack
20:11 vinsh joined #puppet-openstack
20:17 vinsh xarses: Hi.  I wanted to reach out and check with you on an old topic... from Paris.  Something about a "better" swift puppet-module you could get your hands on and release.  Has that gone anywhere recently?    Was maybe a module that did not use exported resources?
20:20 nosleep77 joined #puppet-openstack
20:20 xarses vinsh: we use https://github.com/stackforge/fuel-library/tree/master/deployment/puppet/swift in fuel for swift
20:21 vinsh Very cool.  Thanks for sharing.   I'm working on moving a cluster away from "swiftstack"... and into puppet.
20:21 xarses if you look at the commit history https://github.com/stackforge/fuel-library/commits/master/deployment/puppet/swift
20:21 vinsh need to get my self to an inplace swap of config mgmt here.
20:22 xarses only the last few commits are not from upstream, upstream was synced at https://github.com/stackforge/puppet-swift/commit/b2fca8ed0af48fa7d68ce711813b54a4ecd9c086
20:22 vinsh I do see that
20:23 vinsh xarses: I'll be focusing most of my cycles on adopting this or atleast trying to for the next few weeks.
20:23 vinsh See how it pans out here
20:24 vinsh I'm excited to look through this alternative.
20:24 xarses we enter swift from https://github.com/stackforge/fuel-library/blob/master/deployment/puppet/osnailyfacter/manifests/cluster_ha.pp#L543-598
20:24 xarses and then to https://github.com/stackforge/fuel-library/tree/master/deployment/puppet/openstack/manifests/swift
20:25 vinsh what do you mean by "enter swift" ?
20:25 xarses where we start entering/ calling the swift module from
20:26 tfz joined #puppet-openstack
20:26 xarses so thats the highest level where swift will receive any portion of configuration from our manifest run
20:27 xarses (in-case the calling options become relevant to why we don't need to use stored configs)
20:27 vinsh I'm going to have to study this more to truly understand that I think.
20:28 vinsh The good/bad news is I come from a large chef managed puppet deployment.
20:28 vinsh so.. am aware of some of the challenges over all.
20:28 vinsh Must speak better puppet :)
20:28 xarses there are a number of the mirantis folks stalking in the channel too, most of them are GMT+1 though
20:29 vinsh Cool. I am MDT.
20:30 isd joined #puppet-openstack
20:30 jbell joined #puppet-openstack
20:32 openstackgerrit Matt Fischer proposed stackforge/puppet-neutron: Notify the ovs-agent service if the config changes  https://review.openstack.org/151534
20:33 vinsh that Fischer guy ^^^ up to no good.
20:38 isd Hey, we're trying to get basic cinder/swift services working, and want to use the easiest thing we can as a backend -- my intuition is there's got to be a local disk option somewhere (though for cinder at least I'm only finding nfs). Any opinions on what we should use?
20:39 vinsh How many cinder nodes?
20:39 isd one?
20:39 vinsh Can you back to a "file" store?  on local disk?
20:40 isd I was hoping?
20:41 isd that was kindof what I was trying to find out.
20:42 vinsh I think the option is just called "file"
20:44 pradk joined #puppet-openstack
20:50 xarses if the cinder role is remote, you can just do lvm, it uses iscsi to attach
20:51 vinsh https://github.com/openstack-dev/devstack/blob/master/lib/cinder#L80  kinda touches on that.
20:55 richm joined #puppet-openstack
20:57 hogepodge crinkle, PR 651 on mysql has broken the module for me.
20:58 hogepodge https://gist.github.com/hogepodge/2c2ad3fe51838cd4f2c4
20:58 hogepodge On both CentOS 7 and on Ubuntu 14.04
21:23 mmagr joined #puppet-openstack
21:27 dgurtner joined #puppet-openstack
21:27 dgurtner joined #puppet-openstack
21:30 jbell joined #puppet-openstack
21:55 rcallawa joined #puppet-openstack
22:00 aimon joined #puppet-openstack
22:01 jbell joined #puppet-openstack
22:03 aimon joined #puppet-openstack
22:32 crinkle hogepodge: I wasn't part of the discussion on that patch, you could ping igalic in #puppet-dev or comment on the pr or submit a ticket
22:32 crinkle Robe: i responded to your comment, i think i need a little more to go on
22:33 crinkle I'm in Belgium this week though so I have limited time and am in a different time zone from usual
23:02 Hunner I always think of the whole countryside of belgium looking like a giant waffle
23:02 Hunner I haven't been there before
23:02 nigelk joined #puppet-openstack
23:12 Robe crinkle: say hi to fosdem/cfgmgtcamp from me ;)
23:13 Robe crinkle: we had some other issues in prod so I couldn't get proper timings there - will follow up on monday
23:14 Robe and sorry, didn't want to disrespect your work with that -1 - it's just that from what we've seen it's not the end of the story I'm afraid :(
23:21 imcsk8 joined #puppet-openstack
23:25 crinkle Robe: I'm open to suggestion for improvements, but I'm not sure how to make it better than ~2 requests per resource
23:26 crinkle it's naturally going to be much slower than the other resources because unlike other resources we can't get a full list of instances with one call
23:27 mdorman is there a reason that we _must_ preload/cache all of them?  vs. only make calls to verify those actually referenced in the catalog?
23:27 crinkle mdorman: in my review we're not preloading them, except in the case where someone runs `puppet resource keystone_user_role`
23:27 mdorman i am not totally up to speed with custom providers, so that more a question of ignorance than anything
23:27 crinkle otherwise it's just looking them up one at a time
23:27 mdorman ah, ok.
23:28 mdorman so is that the difference between the ‘class’ method and the ‘instance’ method (or whatever the right terminology is)?
23:28 crinkle yeah sort of
23:28 crinkle puppet resource runs self.instances and gets the whole list
23:28 mdorman k
23:28 mdorman (no need to give me a full explanation)
23:28 mdorman (if you don’t want)
23:29 crinkle for other resources it's efficient to run self.instances and then use that for the exists and the getters, but it's not efficient here
23:29 mdorman kk, yup
23:29 mdorman well yeah then in that case i don’t really see how this could be made any more efficient
23:29 mdorman so i like the proposal to merge this as-is, and then we can iterate on it later if there are other specific improvements identified
23:30 crinkle +1
23:38 jbell joined #puppet-openstack
23:47 michchap joined #puppet-openstack
23:59 cwolferh joined #puppet-openstack

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