Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2013-11-22

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

All times shown according to UTC.

Time Nick Message
00:01 michchap_ joined #puppet-openstack
00:17 hogepodge joined #puppet-openstack
00:56 dachary michchap_: hi, is there a reason for qemu to be the default hypervisor ( instead of kvm for instance ) ? https://github.com/CiscoSystems/openstack-instal​ler/blob/master/data/hiera_data/common.yaml#L26
00:59 michchap joined #puppet-openstack
01:22 badiane_ka joined #puppet-openstack
01:23 rongze joined #puppet-openstack
01:48 digicloseup joined #puppet-openstack
01:48 digicloseup joined #puppet-openstack
02:00 jkyle joined #puppet-openstack
02:00 jkyle evening
02:02 jkyle I'm hacking on the havana branch of the openstack modules and noticed there's some pretty opinionated url generation for keystone service endpoints. Makes me wonder if folks would be open to a less opinionated, more declarative approach and if there's a forum to discuss that
02:03 rongze joined #puppet-openstack
02:04 jkyle for example, you can't set the admin or internal endpoints to https schemes or customize service endpoints for custom uri's (e.g. compute.foo.com, auth.foo.com, etc)
02:07 xingchao joined #puppet-openstack
02:08 jkyle which is recommended SoP by the openstack docs
02:08 michchap jkyle: I don't think anyone would be against a set of patched to support https keystone endpoints. This will be across almost all the modules won't it?
02:08 michchap patches*
02:09 jkyle michchap: yeah, I just implemented it for the nova compute/ec2 services
02:09 michchap What you're proposing doesn't sound controversial at all, but if you want to generate discussion the mailing list would be the best place to start.
02:09 jkyle so that you could still use the openstack abstraction layers
02:09 jkyle openstack module*
02:10 jkyle I'll give that a shot. is there a preference for backwards compat? That's how I did this so all existing users would see the same behavior after upgrading without changing any parameter inputs. But it creates some pretty nasty code due to branching. It'd be lot cleaner to refactor totally
02:11 jkyle do I need to sign CLA's for upstream commits to the stackforge modules (like with openstack)?
02:12 michchap I think you do, but I don't know for sure. Is that a problem for you?
02:13 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Module stackforge/puppet-horizon uses puppet templating to create the /etc/openstack-horizon/local_settings.py configuration file. That module also allows the end user manifests of the module to specify a different template via the parameter local_setting  https://review.openstack.org/57810
02:13 jkyle probably not, though I have to go through a pita process for upstream code submissions. but that's true regardless of a CLA
02:16 michchap does auth_protocol not work for your case?
02:16 michchap https://github.com/stackforge/puppet-​nova/blob/master/manifests/api.pp#L33
02:18 jkyle I never hit that file to do what I needed, which was in nova::keystone::auth
02:19 jkyle to map the advertised service endpoint to: compute.cloud.me.com/v2/tenant_id
02:19 michchap oh the endpoint configuration isn't configurable
02:19 michchap I see
02:19 jkyle that particular case would take care of the scheme, but if you're running through a proxy that doesn't do one to one maps to the 5000...then nova::api::auth_protocol wouldn't do the trick either
02:20 jkyle we serve folks behind a corporate firewall with aggressive port blocking, so it's all mapped to 443 by virtual domain
02:20 jkyle for publics
02:21 michchap yep, so the nova:api case you would pass in auth_uri and just specify the whole thing, but you're still left with the wrong endpoints
02:23 jkyle right
02:23 jkyle I was goign to rewrite them all so you could just pass in uri
02:24 jkyle public_compute_uri = foo and public_ec2_uri = bar and so on
02:24 jkyle I couldn't see int he module where the decomposed uri parameters was needed
02:24 xingchao joined #puppet-openstack
02:25 michchap in api-paste.ini maybe?
02:25 jkyle the other idea I had was to create some helper custom functions if the individual components were needed in other modules so they could all consistently take "endpoint_service_uri"
02:25 jkyle maybe, didn't see any template calls. at least in teh nova::keystone::auth and friends
02:26 michchap it's in nova:api
02:26 jkyle right, so the others don't need it as far as I can see.
02:27 jkyle anyway, I'll take a good look at them and see what I can do
02:27 michchap actually api-paste-ini was deprecated, now it's in nova.conf
02:27 jkyle correct
02:27 michchap https://github.com/stackforge/puppet-n​ova/blob/master/manifests/api.pp#L104
02:27 jkyle depending on version iirc
02:27 michchap yep
02:27 jkyle liek I think grizzly might still use it, definitely folsom
02:28 jkyle ok, beer time!
02:28 michchap enjoy
02:40 thumpba joined #puppet-openstack
02:49 thumpba joined #puppet-openstack
03:07 xarses joined #puppet-openstack
03:17 rharrison10 joined #puppet-openstack
03:48 djschaap joined #puppet-openstack
03:52 rmoe joined #puppet-openstack
03:56 thumpba joined #puppet-openstack
04:26 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-keystone: Create memcache_servers option  https://review.openstack.org/57361
04:46 openstackgerrit A change was merged to stackforge/puppet-neutron: Add missing vpnaas class in example manifest  https://review.openstack.org/57792
05:03 thumpba joined #puppet-openstack
05:24 rongze joined #puppet-openstack
05:40 michchap joined #puppet-openstack
06:26 michchap joined #puppet-openstack
06:29 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Module stackforge/puppet-horizon uses puppet templating to create the /etc/openstack-horizon/local_settings.py configuration file. That module also allows the end user manifests of the module to specify a different template via the parameter local_setting  https://review.openstack.org/57810
06:37 rongze joined #puppet-openstack
06:40 tnoor joined #puppet-openstack
07:08 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Module stackforge/puppet-horizon uses puppet templating to create the /etc/openstack-horizon/local_settings.py configuration file. That module also allows the end user manifests of the module to specify a different template via the parameter local_setting  https://review.openstack.org/57810
07:59 EmilienM good morning here
08:20 EmilienM michchap: thanks for review :)
08:57 mmagr joined #puppet-openstack
09:12 mattymo morning all
09:12 mattymo mmagr, hi
09:15 derekh joined #puppet-openstack
09:15 mmagr mattymo, hi
09:16 marun joined #puppet-openstack
09:16 mattymo mmagr, derekh told me you're doing work on packstack. Has packstack started doing any ceph backend support? I know it's a bad q because RH prefers glusterfs
09:16 mattymo mmagr, I'm trying to find implementations of puppet-openstack that make use of cinder + ceph and maybe you can tell me
09:18 mmagr mattymo, nope, there is no work done on ceph on packstack side ... but modifying cinder plugin to support it shouldn't be too hard, patches are welcome :)
09:19 mattymo actually cinder for packstack has ceph, but it's implemented completely differently in Fuel
09:19 mattymo oops
09:19 mattymo cinder in puppet-openstack
09:20 mattymo there needs to be an abbreviation
09:20 EmilienM mattymo: not a big deal, puppet-cinder supports RBD
09:21 mattymo EmilienM, does your company maintain any toolset that implements it that's publicly available?
09:22 EmilienM mattymo: a toolset for what ?
09:22 mattymo deployment
09:22 EmilienM like Fuel ?
09:23 EmilienM mattymo: for system deployment, we use https://github.com/enovance/edeploy
09:23 mattymo not necessarily so involved
09:23 marun joined #puppet-openstack
09:23 EmilienM mattymo: for openstack configuration we have our puppet manifests, which are not public
09:24 mattymo that sounds not so nice
09:24 EmilienM mattymo: fuel is good, but is not elastic when specific architectures
09:25 EmilienM mattymo: it's really not a big deal, since we commit every day on puppet modules. I think we could publish our manifests, nobody will use them since they are specific to each of our deployments
09:26 EmilienM mattymo: i also contribute on packstack (for red hat), which is good but again, not enough for customer needs
09:27 EmilienM imho, manifests are not a big deal to do, comparing to modules ;-)
09:27 mattymo okay
09:27 EmilienM mattymo: we are waiting contributions on modules
09:27 EmilienM a lot of people use them, fork them, etc
09:28 EmilienM but not a lot of guys make some patchs
09:28 EmilienM mattymo: if you want to have a look at our manifests, i would love to share it with you, if that helps to see how we could improve puppet openstack modules.
09:30 mattymo yeah that would be cool
09:30 mattymo I'm just coming from the perspective that when I joined Fuel, we were already quite forked from puppet-openstack
09:31 mattymo I know how our components interact, but maybe a 3rd variant might show a good middle ground
09:31 EmilienM mattymo: my method is to create class for roles
09:32 EmilienM i.e. heat_server
09:32 EmilienM compute_node
09:32 EmilienM etc
09:32 EmilienM and in those class, i use puppet openstack modules
09:32 EmilienM like nova::compute, neutron::agent::ovs
09:32 EmilienM etc
09:32 EmilienM mattymo: it helps me to have the elasticity for architectures
09:33 EmilienM you are free to install  a service on the node you want, and in HA or not
10:17 openstackgerrit Benedikt Trefzer proposed a change to stackforge/puppet-heat: make auth_encryption_key a parameter  https://review.openstack.org/57881
10:20 openstackgerrit Benedikt Trefzer proposed a change to stackforge/puppet-heat: make auth_encryption_key a parameter  https://review.openstack.org/57881
10:47 marun joined #puppet-openstack
10:56 marun joined #puppet-openstack
12:10 xingchao joined #puppet-openstack
12:18 badiane_ka joined #puppet-openstack
12:23 michchap joined #puppet-openstack
12:46 michchap joined #puppet-openstack
13:00 marun joined #puppet-openstack
13:02 e1mer joined #puppet-openstack
13:03 e1mer joined #puppet-openstack
13:18 bauzas1 joined #puppet-openstack
13:20 openstackgerrit Martin Mágr proposed a change to stackforge/puppet-neutron: VXLAN support  https://review.openstack.org/55578
13:23 EmilienM mmagr: nice patch!
13:23 mmagr EmilienM, thanks
13:23 EmilienM mmagr: i'll review it asap
13:24 mmagr EmilienM, thanks for that too :)
13:39 openstackgerrit Martin Mágr proposed a change to stackforge/puppet-neutron: VXLAN support  https://review.openstack.org/55578
13:50 mattymo EmilienM, mmagr how do you guys deal with nova and cinder limits?
13:50 mattymo do you define a really long string outside of the puppet class? or are default limits just fine?
13:50 EmilienM mattymo: limits of what ?
13:50 mattymo api limits
13:51 mattymo http://docs.openstack.org/havana/config-referenc​e/content//configuring-compute-API.html#d6e1428
13:51 EmilienM rate limits
13:51 mattymo sorry, yes rate limits
13:51 EmilienM mattymo: we don't deal in the puppet modules, should be in a patch
13:51 EmilienM please submit if you can
13:52 mattymo of course
13:52 mattymo any cluster that's more than 5 nodes and has more than 2 users will exceed default rate limits easily
13:52 EmilienM mattymo: actually, it's a great question, i don't know any module dealing with that
13:52 EmilienM mattymo: for sure
13:53 mattymo we have it in fuel, but we implement it differently for each swift, cinder, and nova
13:55 EmilienM mattymo: ok, i'm curious how you're going to make it for our modules :)
13:55 dprince joined #puppet-openstack
13:55 mattymo actually it's not so different
13:55 mattymo https://github.com/Mirantis/fuel/blob/master​/deployment/puppet/nova/manifests/limits.pp
13:58 marun joined #puppet-openstack
14:00 EmilienM mattymo: i see, i think a copy paste + unit tests would be great
14:01 michchap joined #puppet-openstack
14:07 morazi joined #puppet-openstack
14:09 mattymo EmilienM, I'm still analyzing our differences. It'll be Monday when I am ready to start preparing patches and running tests
14:09 EmilienM mattymo: that's awesome
14:14 michchap joined #puppet-openstack
15:00 dmsimard joined #puppet-openstack
15:02 mjblack joined #puppet-openstack
15:09 marun joined #puppet-openstack
15:18 andi- joined #puppet-openstack
15:18 andi- joined #puppet-openstack
15:32 mjblack joined #puppet-openstack
16:04 rongze joined #puppet-openstack
16:16 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Add parameter for custom Horizon local settings template  https://review.openstack.org/57810
16:32 rongze_ joined #puppet-openstack
16:34 badiane_ka joined #puppet-openstack
16:35 thumpba joined #puppet-openstack
16:41 kgerman_ joined #puppet-openstack
16:41 thumpba joined #puppet-openstack
17:01 thumpba joined #puppet-openstack
17:05 openstackgerrit Jeffrey Cox proposed a change to stackforge/puppet-openstack: Add parameter for custom Horizon local settings template  https://review.openstack.org/57810
17:25 hogepodge joined #puppet-openstack
17:44 rmoe joined #puppet-openstack
17:57 xarses joined #puppet-openstack
17:57 djschaap joined #puppet-openstack
17:57 rongze joined #puppet-openstack
18:00 hogepodge joined #puppet-openstack
18:02 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Add metering agent support  https://review.openstack.org/57790
18:08 EmilienM not sure about this one ^, would need a review
18:09 rmoe_ joined #puppet-openstack
18:11 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-keystone: Create memcache_servers option  https://review.openstack.org/57361
18:12 brdude joined #puppet-openstack
18:29 dmsimard joined #puppet-openstack
18:38 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-cinder: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58003
18:58 hogepodge joined #puppet-openstack
19:06 hogepodge joined #puppet-openstack
19:20 openstackgerrit A change was merged to stackforge/puppet-vswitch: Update repo for stackforge migration  https://review.openstack.org/53828
19:27 pabelanger joined #puppet-openstack
19:43 rongze joined #puppet-openstack
19:44 kgerman_ left #puppet-openstack
19:49 djschaap joined #puppet-openstack
20:22 mjblack joined #puppet-openstack
20:32 hogepodge joined #puppet-openstack
20:46 djschaap joined #puppet-openstack
21:38 xarses dachary: ping
21:42 dachary xarses: late pong
21:43 xarses dachary: is this my commit causing this or is this missing otherwise ? http://logs.openstack.org/63/57063/3/check/gat​e-puppet-ceph-puppet-lint/02cf6cc/console.html ref https://review.openstack.org/#/c/57063/
21:44 xarses it looks like the lint is failing because of missing rspec-system-puppet, and the version lints are missing yum
21:44 xarses which i though was a builtin
21:46 dachary https://review.openstack.org/#/c/56427/ passed, therefore I tend to think it's something else
21:47 dachary http://logs.openstack.org/27/56427/1/che​ck/gate-puppet-ceph-puppet-lint/050d2f1/​console.html.gz#_2013-11-14_17_21_36_300
21:48 dachary something may have changed in the environment in which jenkins runs though
21:49 xarses lets see if it will re-test an abandoned change
21:55 openstackgerrit Andrew Woodward proposed a change to stackforge/puppet-ceph: Jenkins Test, do not merge  https://review.openstack.org/58037
22:02 xarses ahh, found the one lint error
22:02 xarses in mine, mostly burried
22:03 xarses still is missing rspec-system-puppet though
22:14 JonnyNomad joined #puppet-openstack
22:20 xarses joined #puppet-openstack
22:23 angdraug joined #puppet-openstack
22:24 openstackgerrit Doug Schaapveld proposed a change to stackforge/puppet-nova: Creating qpid_sasl_mechanisms parameter  https://review.openstack.org/58045
22:25 tnoor1 joined #puppet-openstack
22:28 tnoor2 joined #puppet-openstack
22:31 dmsimard joined #puppet-openstack
23:01 xarses joined #puppet-openstack
23:02 tnoor1 joined #puppet-openstack
23:03 dmsimard joined #puppet-openstack

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