Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-12-11

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

All times shown according to UTC.

Time Nick Message
23:04 badiane_ka joined #puppet-openstack
23:06 rmoe joined #puppet-openstack
23:10 badiane_ka left #puppet-openstack
23:53 michchap_ joined #puppet-openstack
00:41 xingchao joined #puppet-openstack
00:56 sputnik13net joined #puppet-openstack
01:01 tnoor joined #puppet-openstack
01:24 xarses joined #puppet-openstack
01:31 thumpba joined #puppet-openstack
01:32 otherwiseguy joined #puppet-openstack
01:45 thumpba_ joined #puppet-openstack
01:46 thumpba_ joined #puppet-openstack
01:47 michchap joined #puppet-openstack
02:01 rongze joined #puppet-openstack
02:46 rongze joined #puppet-openstack
02:49 thumpba joined #puppet-openstack
03:00 marun joined #puppet-openstack
03:52 markvoelker_ joined #puppet-openstack
04:28 badiane_ka joined #puppet-openstack
04:29 comptona joined #puppet-openstack
05:18 tnoor joined #puppet-openstack
05:20 rongze joined #puppet-openstack
05:24 openstackgerrit A change was merged to stackforge/puppet-ceilometer: TTL mechanism support with expirer service  https://review.openstack.org/60029
05:28 openstackgerrit A change was merged to stackforge/puppet-neutron: Add metering agent support  https://review.openstack.org/60656
05:31 openstackgerrit A change was merged to stackforge/puppet-nova: Clean-up files use  https://review.openstack.org/60792
05:47 tnoor1 joined #puppet-openstack
06:03 rongze joined #puppet-openstack
06:11 brdude joined #puppet-openstack
06:15 rongze joined #puppet-openstack
07:03 tnoor joined #puppet-openstack
08:29 bauzas joined #puppet-openstack
08:29 EmilienM hello
08:35 tnoor joined #puppet-openstack
08:35 tnoor joined #puppet-openstack
08:43 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Add multi-worker support  https://review.openstack.org/61345
08:47 rongze joined #puppet-openstack
08:49 EmilienM michchap: thank you
08:49 tnoor joined #puppet-openstack
08:51 michchap EmilienM: hey, do you do any doc building at eNovance, using puppetdoc?
08:51 EmilienM michchap: no
08:51 EmilienM michchap: and you ?
08:52 tnoor joined #puppet-openstack
08:52 michchap EmilienM: I tried it out yesterday. Some modules work fine, but others don't have the correct syntax for RDoc
08:52 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-ceilometer: TTL mechanism support with expirer service  https://review.openstack.org/60660
08:52 michchap EmilienM: I'm going to try to fix the rest this week, and see if we can host the doc somewhere.
08:53 michchap EmilienM: and then add a gating job for it.
08:53 EmilienM michchap: please explain me what could i do, i would like to help
08:53 brdude joined #puppet-openstack
08:53 EmilienM michchap: that's a VERY good idea
08:53 michchap EmilienM: all your patches are already following the correct syntax
08:53 EmilienM michchap: cool, good news. Thanks to good reviewers :)
08:54 EmilienM michchap: maybe could we coordinate the work on the ML
08:54 EmilienM michchap: let's say, I take Neutron, you take foo, someone takes bar
08:54 EmilienM etc etc
08:54 michchap EmilienM: https://review.openstack.or​g/#/c/61296/1/doc/build_doc.sh
08:55 EmilienM michchap: any example of online result ?
08:55 michchap you can build the doc for all the modules that are used by builder using that script, then take a look at what is missing. I think Nova is probably the best place to start
08:55 EmilienM ok got it
08:55 michchap EmilienM: I haven't hosted it anywhere yet
08:55 EmilienM michchap: ok, i'll try
08:55 EmilienM that would be awesome
08:55 EmilienM to auto generate doc for upstream openstack manuals
08:56 EmilienM i know some folks would be happy to help out that stuff
08:56 EmilienM i would suggest an email to puppet ML first
08:56 michchap I was going to try to catch chris hoge on here and see if there's anything PL can do to help
08:57 tnoor joined #puppet-openstack
08:57 michchap because ideally the doc is hosted on the forge, or upstream somewhere, rather than by Cisco
08:57 tnoor joined #puppet-openstack
08:59 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-ceilometer: TTL mechanism support with expirer service  https://review.openstack.org/60660
08:59 EmilienM michchap: i think he's sleeping
09:00 EmilienM afik, he is on west coast
09:00 EmilienM michchap: i love your idea, please let me know
09:00 michchap EmilienM: yeah I will try in my morning, tomorrow
09:00 EmilienM i'll have a look later
09:03 EmilienM michchap: regarding to https://review.openstack.o​rg/#/c/60799/6//COMMIT_MSG
09:04 EmilienM you are right... it's just sometimes hard to have reviews, so
09:04 michchap EmilienM: I say that because when I was reviewing, I looked at the files and was a bit confused at first, and thought it might be a bit cleaner as a separate patch
09:04 EmilienM i can split in 2 patches
09:04 EmilienM it's not a big deal
09:05 EmilienM michchap: can I rename at least the test where I change something inside ?
09:06 michchap EmilienM: the problem I had was the file that was just renamed but not modified
09:06 EmilienM ok
09:06 EmilienM so I just cancel the rename of one test.
09:06 michchap R
09:06 michchap spec/classes/nova_network_neutron_spec.rb
09:06 michchap renamed from spec/classes/nova_network_quantum_spec.rb
09:07 michchap ^ That should be it's own patch, with a commit message saying tests renamed from Quantum to Neutron
09:09 michchap Ah I see your issue - the compute_ one is also renamed and then modified
09:10 michchap If you could make the rename patch one commit, then depend on that commit for the second patch that changes the default driver, that would be ideal.
09:10 tnoor joined #puppet-openstack
09:15 tnoor joined #puppet-openstack
09:16 tnoor joined #puppet-openstack
09:17 tnoor joined #puppet-openstack
09:18 tnoor joined #puppet-openstack
09:19 tnoor joined #puppet-openstack
09:19 piliman974 joined #puppet-openstack
09:20 tnoor joined #puppet-openstack
09:20 EmilienM michchap: ok, i'll do like this
09:21 tnoor joined #puppet-openstack
09:22 tnoor joined #puppet-openstack
09:23 derekh joined #puppet-openstack
09:24 tnoor joined #puppet-openstack
09:24 tnoor joined #puppet-openstack
09:26 tnoor joined #puppet-openstack
09:28 tnoor joined #puppet-openstack
09:28 tnoor joined #puppet-openstack
09:29 tnoor joined #puppet-openstack
09:31 tnoor joined #puppet-openstack
09:32 tnoor joined #puppet-openstack
09:33 tnoor joined #puppet-openstack
09:36 tnoor joined #puppet-openstack
09:39 michchap EmilienM: https://github.com/stackforge/puppet-​nova/blob/master/manifests/api.pp#L28
09:39 michchap EmilienM: Why do we default to false?
09:39 tnoor joined #puppet-openstack
09:40 EmilienM michchap: lol, indeed
09:40 michchap EmilienM: This is the kind of thing I find when I try to write documentation
09:40 EmilienM michchap: https://github.com/stackforge/puppet-c​inder/blob/master/manifests/api.pp#L78
09:41 EmilienM i think it deserves a patch
09:41 michchap EmilienM: yeah I checked the Neutron ones and they are enabled too
09:41 EmilienM i do it now.
09:41 michchap EmilienM: is there a chance someone is depending on that behavior?
09:41 tnoor joined #puppet-openstack
09:41 EmilienM really, nova-api MUST be enabled when we call the class.
09:42 EmilienM nova-api is just... important :)
09:42 michchap yeah it's a bit strange.
09:43 EmilienM michchap: it's the same for all manifests
09:43 tnoor joined #puppet-openstack
09:44 tnoor joined #puppet-openstack
09:46 tnoor joined #puppet-openstack
09:47 EmilienM michchap: https://bugs.launchpad.net​/puppet-nova/+bug/1220473
09:47 EmilienM already know issue
09:47 tnoor joined #puppet-openstack
09:47 EmilienM https://bugs.launchpad.net/pupp​et-nova/+bug/1220473/comments/3
09:47 EmilienM he's right ...
09:48 tnoor joined #puppet-openstack
09:48 michchap Oh that's annoying
09:48 mmagr joined #puppet-openstack
09:50 EmilienM michchap: indeed.
09:50 tnoor joined #puppet-openstack
09:51 michchap I feel like I could send a message to the list and change it in he next release
09:51 tnoor joined #puppet-openstack
09:52 michchap I'm going away a bit, I'll be back in a few hours.
09:52 tnoor joined #puppet-openstack
09:53 tnoor joined #puppet-openstack
09:54 tnoor joined #puppet-openstack
09:56 tnoor joined #puppet-openstack
09:57 tnoor joined #puppet-openstack
09:58 tnoor joined #puppet-openstack
10:00 tnoor joined #puppet-openstack
10:01 tnoor joined #puppet-openstack
10:02 tnoor joined #puppet-openstack
10:18 EmilienM michchap: ack
11:06 mgagne joined #puppet-openstack
11:06 mgagne joined #puppet-openstack
11:27 LAnthony joined #puppet-openstack
11:33 bogdando joined #puppet-openstack
11:34 bauzas joined #puppet-openstack
11:44 rongze joined #puppet-openstack
12:09 hbroyon joined #puppet-openstack
12:31 michchap EmilienM: back
12:32 markvoelker1 joined #puppet-openstack
12:40 LAnthony left #puppet-openstack
12:42 hbroyon left #puppet-openstack
12:46 EmilienM michchap: wow, you never sleep
12:46 michchap it's only midnight
13:07 morazi joined #puppet-openstack
13:09 mattymo like a boss
13:11 dprince joined #puppet-openstack
13:13 EmilienM michchap: thank you for your reviews by the way
13:13 michchap EmilienM: no problem
13:18 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-nova: Change nova::api doc to conform to style guide  https://review.openstack.org/61414
13:18 EmilienM michchap: nice one ^
13:19 michchap EmilienM: thanks. It's going to take a while to fix all of it
13:19 EmilienM yeah
13:21 mattymo michchap, do you work with bodepd in Cisco?
13:21 michchap mattymo: I was for the previous cycle, but bodepd is not at Cisco at the moment.
13:21 mattymo oh!!
13:22 mattymo back to PuppetLabs? or did he slip through the cracks?
13:22 michchap I think he is doing short contracts at the moment
13:22 mattymo cool
13:22 mattymo good for us) we want to steal him for a short run
13:23 mattymo michchap, were at you at summit in HK?
13:23 michchap Yeah I know :)
13:23 michchap yep
13:23 prad joined #puppet-openstack
13:23 mattymo were you at the meetup for puppet-openstack?
13:23 michchap yep
13:23 mattymo I didn't get a chance to meet everyone..
13:23 mattymo a big room
13:23 michchap yeah. I feel like we should organise it a bit better next time
13:23 EmilienM +1
13:24 michchap One option is if someone does the work to integrate with tripleo, we can tack on the end of their summit track
13:24 mattymo I agree. I'm not sure if I'll make it to Atlanta in May, but I hope so
13:25 mattymo I sat on 2/3 of the tripleo sessions there and it seems leaning 70% toward python/bash and avoiding puppet, with a lot of pressure from dell crowbar guys to look at chef
13:26 mattymo for a reference model
13:27 michchap they will reinvent everything. That's fine. I also suspect their use case is somewhat different from ours
13:27 mattymo there is a lot of reinvention going on in this project's scope
13:28 michchap but the point remains that puppet-openstack should be deployable by heat on top of ironic, and that could be via a tripleo element.
13:28 mattymo and our team at Mirantis is trying to reinvent ironic to scale a lot faster than glance
13:28 michchap by that do you mean glance is the limiting factor in an ironic deployment?
13:29 mattymo getting images out to 1,000 or 10,000 nodes still relies on your "undercloud" size
13:29 mattymo there's a middle ground where you could use p2p (like bittorrent) to exchange that data and distribute and verify it many times faster
13:30 mattymo right now the undercloud is some fixed permanent role that really limits your deployment speed
13:30 mattymo and then it's costly to dedicate so many systems for what in many cases is a temporary role
13:34 michchap That makes a lot of sense
13:34 michchap sounds like a pretty fun problem tbh!
13:35 michchap by the way...does Mirantis use 0MQ in deployments at all?
13:35 EmilienM 0MQ \o/
13:35 xingchao joined #puppet-openstack
13:36 michchap EmilienM: are you guys using it?
13:36 EmilienM michchap: oh no !
13:36 EmilienM we use RabbitMQ
13:36 michchap ah...rabbit for debian and qpid for RHEL?
13:36 EmilienM yeah
13:36 EmilienM i wonder who use 0MQ
13:37 mattymo still rabbitmq
13:37 mattymo we use qpid for RHEL but HA queues still doesn't work
13:37 mattymo and rabbitmq for ubuntu and centos
13:37 michchap I'd like to try it out but I don't have enough kit to really test the scale
13:38 mattymo michchap, the biggest problem is everyone is a critic of installation methodology
13:38 EmilienM imo, rabbitmq is the best way today when installing openstack
13:38 mattymo except when it fails it's a very bad day
13:39 mattymo running out of disk space is a good way to kill rabbitmq
13:39 michchap mattymo: yep. Especially the bare metal part - there are how many deployment tools around now?
13:39 michchap mattymo: MAAS, Razor, Cobbler, Foreman, Crowbar
13:39 mattymo crowbar does bare metal on its own?
13:40 michchap mattymo: I was talking to bodepd about this, there is a fundamental divide between people operating at massive scale, who must use discovery methods to deploy, and smaller clusters which prefer specificity because it's the more natural way to deploy.
13:41 michchap mattymo: I actually think tripleo is going to go with the former, and the latter may remain the domain of puppet/chef/ansible
13:41 mattymo Fuel has its own discovery agent (not too different from Foreman's) and then cobbler for actual deployment
13:42 mattymo very few people have homogenous hardware
13:43 michchap mattymo: true.
13:43 michchap mattymo: that doesn't prevent discovery though
13:44 mattymo discovery is crucial
13:44 mattymo and asking an admin to plug in mac addresses is oldschool
13:45 michchap mattymo: we still do that. It's crap.
13:45 mattymo in our more recent release we now try to group hosts based on similar components
13:46 michchap so you're matching roles to flavors in a kind of fuzzy way?
13:49 EmilienM michchap: mattymo: with eDeploy, we don't (mac @ stuff)
13:49 EmilienM we make discovery from hardware
13:51 michchap EmilienM: then what process do you use to decide what hardware takes which role?
13:52 EmilienM when a server boot up by PXE, eDeploy is able to say "it's a server with 8 disks, it's a swift servers"
13:52 EmilienM that's an example ^
13:52 EmilienM of course, it's very modular
13:52 michchap so you set some filter?
13:52 EmilienM maybe I'm not the best person to talk about that but we are close to release a complete documentation
13:53 EmilienM somes rules yeah
13:53 EmilienM those rules could be everything
13:53 michchap ok
13:53 EmilienM michchap: and we manage upgrades :)
13:54 michchap upgrades?
13:54 michchap of openstack? kernel? ceph?
13:55 EmilienM michchap: every software
13:55 EmilienM michchap: as far the doc will be ready, i'll share with you. It will be public of course
13:56 michchap EmilienM: cool
13:56 EmilienM michchap: sorry for spoiling
13:56 EmilienM michchap: i don't like that
13:56 EmilienM but i'll show you the stuff
13:56 EmilienM maybe you will be interested in
14:00 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-nova: Make parameter doc RDoc compatible  https://review.openstack.org/61426
14:01 michchap ^ I did 4 classes. It takes so long :(
14:02 mattymo roles to flavors?
14:02 mattymo sorry was helping a colleague
14:02 mattymo we only have 4 roles right now - controller, compute, cinder, and ceph
14:02 mattymo and can be combined
14:03 michchap mattymo: no problem. I guess...how does a user define which hardware they want to have which services on them
14:03 michchap mattymo: what's the cinder role for?
14:03 michchap mattymo: we have one for neutron but barely use it.
14:04 mattymo hmm I don't have any demos of our current system but I can show you a screenshot
14:07 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-nova: Make parameter doc RDoc compatible  https://review.openstack.org/61426
14:12 prad joined #puppet-openstack
14:14 mattymo michchap, like this http://imgur.com/tTxy15p
14:15 mattymo these are vms
14:15 michchap mattymo: so you discover first, then match roles to what has been discovered
14:15 michchap mattymo: that makes sense. looks neat :)
14:18 mattymo yeah it's hard to check 50+ boxes
14:18 mattymo cinder role is a storage node
14:18 mattymo controllers automatically run cinder and create a cinder lv
14:18 mattymo unless you explicitly don't create one
14:19 mattymo neutron needs to be HA so we run it on all controllers
14:19 mattymo we had a separate neutron node role before but non-HA neutron is a bad idea
14:19 michchap yep
14:23 morazi joined #puppet-openstack
14:31 xingchao joined #puppet-openstack
15:20 otherwiseguy joined #puppet-openstack
15:40 xingchao_ joined #puppet-openstack
15:46 badams joined #puppet-openstack
15:49 dtalton joined #puppet-openstack
15:54 rmoe joined #puppet-openstack
15:54 dmsimard joined #puppet-openstack
16:08 tnoor joined #puppet-openstack
16:12 mgagne fun... https://bugs.launchpad.net/bugs/1259911
16:13 EmilienM damn
16:20 otherwiseguy joined #puppet-openstack
16:23 mgagne we have the same problem, puppetlabs mirror is unreliable
16:25 mgagne even the mirroring process suggested by puppetlabs isn't working all the time
16:31 prad_ joined #puppet-openstack
16:33 mattymo we mirror it for my org
16:34 mattymo and then rebuild)
16:34 mattymo just in case
16:42 dmsimard yeah, puppetlabs' apt repo is definitely not working properly right now, blowing my things up
16:46 xingchao joined #puppet-openstack
16:46 mattymo dmsimard, this hasn't updated yet today if it's any help http://mirror.its.sfu.ca/mirror/PuppetLabs/
16:47 mgagne mattymo: we do have a local mirror =)
16:47 mgagne mattymo: gonna move our apt source to it now
16:50 xingchao_ joined #puppet-openstack
16:51 rongze_ joined #puppet-openstack
16:52 rongze__ joined #puppet-openstack
16:55 thumpba joined #puppet-openstack
17:08 EmilienM bodepd: thanks
17:13 bodepd joined #puppet-openstack
17:17 rongze joined #puppet-openstack
17:23 angdraug joined #puppet-openstack
17:27 rmoe joined #puppet-openstack
17:38 brdude joined #puppet-openstack
17:44 xarses joined #puppet-openstack
18:12 terryw joined #puppet-openstack
18:33 mmagr joined #puppet-openstack
18:43 badiane_ka joined #puppet-openstack
19:05 tnoor joined #puppet-openstack
19:06 prad joined #puppet-openstack
19:07 tnoor1 joined #puppet-openstack
19:15 rongze joined #puppet-openstack
19:34 openstackgerrit joined #puppet-openstack
19:43 prad joined #puppet-openstack
19:55 seiflotfy_ joined #puppet-openstack
20:18 terryw joined #puppet-openstack
20:34 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Expose fqdn parameter of stackforge/horizon module  https://review.openstack.org/58453
20:39 Jean-Roger joined #puppet-openstack
21:34 thumpba joined #puppet-openstack
21:48 thumpba joined #puppet-openstack
22:10 occupant joined #puppet-openstack
22:31 michchap joined #puppet-openstack

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