Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-08-14

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

All times shown according to UTC.

Time Nick Message
00:04 openstackgerrit Daneyon Hansen proposed a change to stackforge/puppet-heat: Updated Module File Dependencies  https://review.openstack.org/114085
00:21 britthouser joined #puppet-openstack
00:30 richm left #puppet-openstack
00:55 rwsu joined #puppet-openstack
00:57 rwsu joined #puppet-openstack
01:05 sputnik13 joined #puppet-openstack
01:06 michchap hogepodge: ok, so I might leave the uca option in there but disable it by default, then.
01:09 sputnik13 joined #puppet-openstack
01:17 Krast joined #puppet-openstack
01:20 rmoe joined #puppet-openstack
01:26 sputnik13 joined #puppet-openstack
01:28 Dafna joined #puppet-openstack
01:29 Dafna joined #puppet-openstack
01:37 sputnik13 joined #puppet-openstack
01:41 xarses joined #puppet-openstack
01:45 sputnik13 joined #puppet-openstack
01:45 imcsk8 joined #puppet-openstack
01:47 gildub joined #puppet-openstack
01:51 xingchao joined #puppet-openstack
02:19 sputnik13 joined #puppet-openstack
02:25 sputnik13 joined #puppet-openstack
02:31 sputnik13 joined #puppet-openstack
02:50 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
03:08 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
03:16 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
03:19 openstackgerrit Mark T. Voelker proposed a change to stackforge/puppet_openstack_builder: Add nova::scheduler::filter to nova_controller  https://review.openstack.org/114109
03:40 aimon joined #puppet-openstack
03:51 chlunde_ joined #puppet-openstack
03:51 imcsk8_ joined #puppet-openstack
03:51 Zordrak_ joined #puppet-openstack
03:51 Zordrak_ joined #puppet-openstack
03:52 Quack_ joined #puppet-openstack
03:52 toabctl_ joined #puppet-openstack
03:52 ken_barber1 joined #puppet-openstack
03:53 mgagne joined #puppet-openstack
03:53 Dafna joined #puppet-openstack
03:53 marcel joined #puppet-openstack
03:53 peterstac joined #puppet-openstack
03:53 dmsimard_away joined #puppet-openstack
03:53 uberj_ joined #puppet-openstack
03:55 ilbot3 joined #puppet-openstack
03:55 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
03:56 comptona joined #puppet-openstack
04:10 ilbot3 joined #puppet-openstack
04:10 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
04:26 sputnik13 joined #puppet-openstack
04:29 blkperl joined #puppet-openstack
04:48 xarses joined #puppet-openstack
04:58 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
05:06 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
05:09 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
05:09 exed_- joined #puppet-openstack
05:10 gaudenz_ joined #puppet-openstack
05:13 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
05:18 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
05:29 sanjayu joined #puppet-openstack
05:34 openstackgerrit A change was merged to stackforge/puppet_openstack_builder: Add nova::scheduler::filter to nova_controller  https://review.openstack.org/114109
05:51 saneax_-_ joined #puppet-openstack
06:16 sanjayu joined #puppet-openstack
06:20 xingchao_ joined #puppet-openstack
06:26 sanjayu_ joined #puppet-openstack
06:28 sanjayu joined #puppet-openstack
06:30 sputnik13 joined #puppet-openstack
06:49 sputnik13 joined #puppet-openstack
06:53 xingchao joined #puppet-openstack
07:06 daneyon joined #puppet-openstack
07:08 EmilienM good morning
07:13 sputnik13 joined #puppet-openstack
07:24 daneyon joined #puppet-openstack
07:24 daneyon joined #puppet-openstack
07:27 sputnik13 joined #puppet-openstack
07:35 tfz joined #puppet-openstack
07:39 sanjayu joined #puppet-openstack
07:40 sputnik13 joined #puppet-openstack
07:57 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack-specs: Add hash based repo management in extras  https://review.openstack.org/114155
08:00 michchap EmilienM: morning
08:00 mmagr joined #puppet-openstack
08:10 qba73 joined #puppet-openstack
08:42 derekh joined #puppet-openstack
08:59 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
09:16 EmilienM michchap: I may lag in review over the last days, I had to finish my these. I'm done tonight. Next week, I'll do more reviews than before
09:16 michchap EmilienM: no problem, good luck
09:52 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
10:09 openstackgerrit A change was merged to stackforge/puppet-cinder: volume/iscsi: Fix typo and use real default iscsi_helper in tests  https://review.openstack.org/114021
10:20 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
10:24 sanjayu joined #puppet-openstack
10:25 sanjayu joined #puppet-openstack
10:27 sanjayu joined #puppet-openstack
10:34 sanjayu joined #puppet-openstack
10:35 michchap EmilienM: ping
10:39 michchap EmilienM: I am reviewing the openstacklib database refactor across modules. In keystone we notify (restart) the keystone service when we create a DB, but for glance we tie it to the db_sync exec - which do you prefer?
10:40 michchap EmilienM: keystone: https://github.com/stackforge/puppet-keystone/blob/master/manifests/db/mysql.pp#L50 glance: https://github.com/stackforge/puppet-glance/blob/master/manifests/registry.pp#L324
10:41 michchap I think in keystone we could do the db_sync after we restart keystone, which is not really correct
11:05 tfz joined #puppet-openstack
11:12 openstackgerrit Johannes Grassler proposed a change to stackforge/puppet-ceilometer: Add extended logging options.  https://review.openstack.org/113560
11:14 jgrassler So much for yesterday afternoon (An ill advised 'I'll quickly whip up a heat template for $other_thing.' got in the way. Apparently there is no such thing as quickly whipping up a heat template...).
11:16 jgrassler Good thing I added the platform specific contexts: while there's nothing platform specific in the class they did cause some namespacing issues that forced me to refactor the spec test.
11:16 michchap Need an automatic (idea -> heat template) converter :)
11:17 jgrassler Also, a Heat specification that is less broken :-)
11:18 jgrassler Anyway, the refactored spec turned out to be less broken than it was before and uncovered a copy & paste related error in the class (which has been fixed now).
11:18 michchap It has its moments. I quite like heat for simple resource declarations, but once I start needing any logic the syntax is pretty bad.
11:19 jgrassler I didn't even get to the point of really needing logic.
11:19 jgrassler I just tried to build ResourceGroups as documented - which turned out not to work.
11:20 jgrassler There's something shady going on with the '%' in the '%index%' this class uses for interpolation in the resource definition proper.
11:21 jgrassler I'm fairly sure it's related to the YAML incarnation of Heat choking on the '%'.
11:22 michchap is the % double quoted? I have a similar issue in hiera sometimes.
11:22 michchap % isn't a legal first character in yaml or something. I have to quote it even though yaml allegedly just stringifies everything it isn't sure about.
11:22 jgrassler In the docs it's not quoted at all - they just use it liberally everywhere, whithout any quoting whatsoever.
11:23 jgrassler Exactly, yes.
11:23 jgrassler Hmm. I guess I could try quoting it...
11:23 jgrassler That might just work...
11:25 openstackgerrit A change was merged to stackforge/puppet-neutron: Makes ca_file parameter optional when use_ssl => true  https://review.openstack.org/113673
11:27 jgrassler michchap: Indeed. Quoting did it. Thanks!
11:40 michchap jgrassler: my pleasure
11:45 openstackgerrit Martin Mágr proposed a change to stackforge/puppet-cinder: Add path to exec cinder type-create  https://review.openstack.org/114209
11:46 morazi joined #puppet-openstack
11:50 EmilienM michchap: I like the notify and you?
11:51 michchap EmilienM: I think the service should restart if something in the db changes, because it could be something like the allowed hosts is updated and now the node allowed to connect.
11:53 EmilienM I do agree.
11:59 michchap EmilienM: some of the modules have it. I will file a bug against the ones that are missing it.
12:00 EmilienM michchap: for the allowed hosts stuff, you don't have to restart the service
12:00 michchap EmilienM: how long will, say, neutron server keep trying to connect to a db that disallows it, for example?
12:01 EmilienM michchap: from my experience, I see the process running and trying to connect to DB every x seconds (x is a parameter btw)
12:01 michchap EmilienM: if neutron-server will just sit there retrying and failing forever until it can connect, we don't need to refresh it. If it will stop eventually then we should notify it when there are db changes.
12:02 michchap EmilienM: although something else I thought is that in some cases it shouldn't restart, it should reload, but I don't think openstack daemons support that.
12:02 EmilienM michchap: I don't mind to refresh but still I think it's not useful for MySQL permissions. However it is for db_sync stuff
12:02 EmilienM I mean, restart is not a big deal when you scale out services
12:02 EmilienM though you can face race conditions sometimes
12:03 EmilienM at scale maybe
12:03 michchap EmilienM: it's not, but it still drops connections in progress
12:03 EmilienM michchap: it annoys monitoring :(
12:03 michchap EmilienM: so at scale maybe that's 50 customers whose API call will fail.
12:04 michchap EmilienM: and if you're using heat it might rollback a whole stack because of one API call failure
12:04 michchap (or does heat-engine retry?)
12:06 EmilienM michchap: interesting topic indeed
12:10 EmilienM michchap: using ceph?
12:11 EmilienM michchap: because the release in fedora/centos is broken FYI. Prefer using ceph.com pkg
12:11 michchap EmilienM: not in prod but in test yes
12:11 EmilienM michchap: don't expect something from 0.81
12:11 EmilienM use 0.8
12:11 michchap EmilienM: :( thanks. What happened?
12:12 EmilienM michchap: http://tracker.ceph.com/issues/9075
12:12 EmilienM michchap: http://tracker.ceph.com/issues/9076
12:13 michchap EmilienM: ouch.
12:14 EmilienM yup
12:14 EmilienM michchap: for RH I deploy ICE: http://www.inktank.com/enterprise/
12:19 britthouser joined #puppet-openstack
12:21 dprince joined #puppet-openstack
12:24 cjellick joined #puppet-openstack
12:27 saneax_-_ joined #puppet-openstack
12:29 cjellick joined #puppet-openstack
13:03 [1]Mr_D joined #puppet-openstack
13:07 tfz joined #puppet-openstack
14:03 prad joined #puppet-openstack
14:19 [1]Mr_D joined #puppet-openstack
14:23 rmoe joined #puppet-openstack
14:51 dgollub joined #puppet-openstack
14:56 daneyon joined #puppet-openstack
15:00 marun joined #puppet-openstack
15:04 marun joined #puppet-openstack
15:07 [1]Mr_D joined #puppet-openstack
15:07 marun joined #puppet-openstack
15:11 sputnik13 joined #puppet-openstack
15:12 marun joined #puppet-openstack
15:17 marun joined #puppet-openstack
15:19 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
15:22 richm joined #puppet-openstack
15:25 meylor joined #puppet-openstack
15:33 pradk joined #puppet-openstack
15:33 richm michchap: ping - Do you have a minute to chat about https://review.openstack.org/109676?
15:34 michchap richm: I'm barely awake but I'll do my best :)
15:34 mmaglana joined #puppet-openstack
15:34 richm michchap: is the main objection that a class should not call another class?
15:35 richm I'm concerned that users of puppet-keystone are accustomed to calling class {'keystone'} and having it both configure the server _and_ make the server run
15:35 richm That is the semantic of using class {'keystone'}
15:36 michchap richm: It's not a definite rule - as you can see above the  keystone::service class is called. The difference is that all the reasonably required parameters can be exposed without increasing the parameter count of the calling class too much
15:36 richm With the introduction of wsgi, I don't want to break that semantic unnecessarily
15:36 michchap richm: The problem is that it will turn out like older classes, like openstack::compute, where we had parameters like 'enable_xen' and 'enable_libvirt' which then went on to create classes
15:37 michchap so the parameter list for openstack::compute grows and grows
15:37 michchap so we would have to add 12 parameters to the keystone class just to handle all the wsgi parameters
15:37 richm Ok, I see.  Is it ok if keystone has 12 duplicated parameters that must exactly match the same parameters to keystone::wsgi::apache?
15:38 michchap That's what we want to avoid
15:38 michchap that's why we break up the classes and make them modular
15:38 michchap because otherwise the class is responsible for too many things, and it has  a huge parameter count and is hard to work with.
15:39 michchap we want many small classes, and let people choose which ones to include
15:39 richm What I mean is - if I do make it two separate classes, then there will be about a dozen parameters that must be passed into keystone, and those same parameters must also be passed into keystone::wsgi::apache
15:39 richm because they must match exactly
15:39 richm e.g. keystone admin host, keystone admin port, keystone main host, keystone main port, keystone admin location, keystone main location, etc. etc.
15:40 michchap richm: if you add class {'keystone::wsgi::apache': ... params .. } to the keystone class, then yes, you'll have to add all the parameters for keystone::wsgi::apache to keystone.
15:40 richm Yes, I understand that.  I'm talking about the case where I _don't_ add class keystone::wsgi::apache to keystone.  I will have many duplicated parameters
15:40 michchap so servername, public_port, admin_port ... ssl_certs_dir would all need to be added to keystone so that they are configurable by the user
15:41 richm and they would also need to be added to keystone::wsgi::apache
15:41 michchap the parameters are already there for keystone::wsgi::apache
15:41 richm right
15:43 michchap this situation is not uncommon - for example we enable a plugin in neutron and expect the user to include the correct plugin and agent classes
15:43 michchap all the neutron class has is one parameter to control core_plugin, and everything else is handled by the other classes
15:44 michchap this keeps the central init classes from becoming monolithic
15:46 rmoe joined #puppet-openstack
15:46 openstackgerrit Colleen Murphy proposed a change to stackforge/puppet-glance: Migrate mysql backend to use openstacklib::db::mysql  https://review.openstack.org/111072
15:46 daneyon joined #puppet-openstack
15:46 michchap I see the duplication you're talking about, but honestly that's also everywhere - we set the same value to the same parameters for many classes all over the place. Think of the number of places where the IP address of the control node has to be set.
15:48 michchap So duplicating a few SSL parameters for apache is honestly not something I'm too worried about. I'm more worried about making the same mistakes as ~18 months ago when we tried to make monolithic classes.
15:49 richm michchap: http://paste.openstack.org/show/95089/
15:49 richm ok
15:50 richm michchap: Then perhaps should I also take this opportunity to remove all of the service code from class keystone and put it in a keystone::eventlet_service class or something like that?
15:51 michchap richm: #2 can get sticky with parse order but you can try it.
15:51 richm or keystone::standalone_service?
15:52 richm michchap: I guess it doesn't matter - if these are going to be replaceable parameters then they will be provided via macros from the installer framework e.g. in packstack %(CONFIG_KEYSTONE_PUBLIC_PORT)s
15:52 michchap richm: it's the upstream default and would be hard to keep backwards compat so I would leave it as is. The service is already in its own class.
15:52 richm michchap: ok
15:52 michchap richm: yeah exactly - everyone uses composition and mappings anyway, which is why having many classes with the same params is easy to handle
15:53 richm michchap: Thanks.
15:53 michchap richm: no worries. Sorry for making it hard.
15:53 openstackgerrit Colleen Murphy proposed a change to stackforge/puppet-keystone: Migrate mysql backend to use openstacklib::db::mysql  https://review.openstack.org/111075
15:53 richm michchap: no worries - I'm new to this and I want to do it the right way
15:54 michchap richm: here's the monolithic mistake we made: https://github.com/stackforge/puppet-openstack/blob/master/manifests/controller.pp
15:55 michchap richm: so every time we change a parameter for a subclass, we have to make a change in that class. It was horrible.
15:55 openstackgerrit Colleen Murphy proposed a change to stackforge/puppet-keystone: Migrate mysql backend to use openstacklib::db::mysql  https://review.openstack.org/111075
15:56 michchap richm: I think packstack was written the way it was because the author saw how bad it was with everything in together.
15:56 blentz joined #puppet-openstack
15:56 richm yeah - that's a large ugly parameter list
16:00 openstackgerrit Michael Chapman proposed a change to stackforge/puppet-openstack_extras: Add hash based repository management  https://review.openstack.org/113922
16:01 michchap crinkle: hogepodge: repo mgmt spec at https://review.openstack.org/#/c/114155/ WIP implementation at: https://review.openstack.org/#/c/113922/
16:21 openstackgerrit Mark T. Voelker proposed a change to stackforge/puppet_openstack_builder: Add nova::scheduler::filter to nova_controller  https://review.openstack.org/114288
16:30 mgagne joined #puppet-openstack
16:32 dprince joined #puppet-openstack
16:45 aimon joined #puppet-openstack
16:53 sputnik13 joined #puppet-openstack
17:27 mmaglana joined #puppet-openstack
17:40 daneyon joined #puppet-openstack
17:45 openstackgerrit Kiran Chunduri proposed a change to stackforge/puppet-neutron: New manifest file to deploy Cisco N1KV-VEM  https://review.openstack.org/103442
17:56 xarses joined #puppet-openstack
18:08 Kiran joined #puppet-openstack
18:09 openstackgerrit Daneyon Hansen proposed a change to stackforge/puppet-heat: Require Class['mysql::server'], not Service['mysqld'] when creating DB.  https://review.openstack.org/114308
18:17 Mr_D joined #puppet-openstack
18:24 daneyon joined #puppet-openstack
18:45 [1]Mr_D joined #puppet-openstack
18:58 dprince joined #puppet-openstack
19:37 openstackgerrit Daneyon Hansen proposed a change to stackforge/puppet-heat: Updates Example Manifest  https://review.openstack.org/114321
19:46 snk joined #puppet-openstack
21:07 openstackgerrit Richard Megginson proposed a change to stackforge/puppet-keystone: setup keystone using apache mod_wsgi  https://review.openstack.org/109676
21:24 cjellick_ joined #puppet-openstack
21:44 britthouser joined #puppet-openstack
21:48 openstackgerrit joined #puppet-openstack
21:50 mmaglana joined #puppet-openstack
22:04 cjellick joined #puppet-openstack
23:06 hogepodge joined #puppet-openstack
23:20 thumpba joined #puppet-openstack
23:25 hogepodge joined #puppet-openstack
23:29 gildub joined #puppet-openstack

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