Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2015-01-20

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

All times shown according to UTC.

Time Nick Message
00:06 nkinder joined #puppet-openstack
00:37 epx998 joined #puppet-openstack
00:48 rwsu joined #puppet-openstack
00:53 gildub joined #puppet-openstack
01:20 rmoe joined #puppet-openstack
01:48 xarses joined #puppet-openstack
02:12 badiane_ka joined #puppet-openstack
02:12 markvoelker joined #puppet-openstack
02:20 openstackgerrit Merged stackforge/puppet-keystone: service validation: update the default command  https://review.openstack.org/148167
02:33 mfisch michchap_: you here?
02:33 michchap_ I am!
02:33 michchap_ mfisch: you're up late!
02:33 mfisch 7:30 pm here
02:33 michchap_ mfisch: oh. timezones are hard.
02:33 mfisch for sure
02:34 mfisch have you seen situations where mysql operations try to start before /root/.my.cnf is written?
02:34 mfisch we get it on our keystone nodes sometimes, but the fix I CP'd didn't seem to solve it
02:35 mfisch This was what I cherry-picked
02:35 mfisch Puppet::Type.newtype(:mysql_user) do
02:35 mfisch ...
02:35 mfisch autorequire(:file) { '/root/.my.cnf' }
02:35 michchap_ on your keystone nodes?
02:36 mfisch yeah
02:36 michchap_ does the keystone node also include msql?
02:36 mfisch yeah
02:36 mfisch and galera
02:36 michchap_ ah ok.
02:36 mfisch this is why I was trying to move to the latest mysql module
02:37 mfisch the type has the autorequire
02:37 michchap_ Actually wouldn't that make a dependency cycle?
02:37 michchap_ https://github.com/puppetlabs/puppetlabs-mysql/b​lob/master/manifests/server/root_password.pp#L17
02:38 michchap_ .my.cnf requires the user and the user requires .my.cnf?
02:38 mfisch that does in fact seem wrong
02:38 mfisch Error: Could not prefetch mysql_user provider 'mysql': Execution of '/usr/bin/mysql -NBe SELECT CONCAT(User, '@',Host) AS User FROM mysql.user' returned 1: ERROR 1045 (28000): Access denied for user 'root'@'localhost' (using password: NO)
02:38 mfisch thats what I get when the .my.cnf isn't written
02:39 mfisch then everything else explodes of course
02:40 michchap_ I don't think I want to start debugging the mysql_user provider right now. I was having such a good day...
02:40 mfisch me either
02:40 mfisch just wondered if you have any ideas or had seen it
02:40 michchap_ I haven't looked at it since I tried to fix the character count thing
02:41 michchap_ so I don't know. I definitely haven't seen the behaivior you're describing where users are failing, though my test harness is still on mysql and not galera.
02:45 mfisch https://tickets.puppetlabs.com/browse/MODULES-1058
02:45 mfisch thats the issue like you said
02:47 ilbot3 joined #puppet-openstack
02:47 Topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
02:47 michchap_ yeah it will fail to compile!
02:47 mfisch michchap_: unsure how we got here, I just no that sometimes I cannot bring up a clean node
02:47 michchap_ that's why I'm confused. The dep graph will spit out an error telling you there's a cycle and will abort
02:48 michchap_ I'm going to try to get some work time allocated to fixing up puppet-galera and doing a forge release
02:48 michchap_ maybe next week
02:48 michchap_ we're allegedly using it here as well so I don't think I can just do nothing.
02:48 michchap_ so during the course of that there's every chance I'll see your issue.
02:49 mfisch what if sometimes your debian code runs before the root_password code and sometimes not
02:49 mfisch I think root_password is broken
02:50 mfisch is the order in the catalog once I dump it the order its processed in?
02:51 michchap_ the order is arbitrary but consistent afaik
02:52 michchap_ you can do manifest ordering using a cmd line flag
02:52 mfisch consistent per node
02:52 mfisch let me try this hack
02:52 mfisch if it works I'll submit a patch
02:54 mfisch it works of course
02:56 michchap_ yep
02:59 openstackgerrit Merged stackforge/puppet-neutron: fixtures: Unpin puppetlabs-mysql module  https://review.openstack.org/144517
03:08 openstackgerrit Merged stackforge/puppet-nova: Switch to TLSv1 as SSLv3 is considered insecure and is disabled by default  https://review.openstack.org/144809
03:14 markvoelker joined #puppet-openstack
04:03 aimon_ joined #puppet-openstack
04:15 markvoelker joined #puppet-openstack
04:18 berendt1 joined #puppet-openstack
05:16 markvoelker joined #puppet-openstack
05:52 sanjayu joined #puppet-openstack
06:11 saju_m joined #puppet-openstack
06:16 berendt joined #puppet-openstack
06:17 markvoelker joined #puppet-openstack
06:27 sanjayu joined #puppet-openstack
06:50 openstackgerrit Ryan Hefner proposed stackforge/puppet-manila: Change share server config details  https://review.openstack.org/137534
06:51 openstackgerrit sajuptpm proposed stackforge/puppet-heat: added option to configure flavor in heat.conf  https://review.openstack.org/104795
06:51 openstackgerrit Ryan Hefner proposed stackforge/puppet-manila: Change share server config details  https://review.openstack.org/137534
07:19 markvoelker joined #puppet-openstack
07:27 arnaud_orange joined #puppet-openstack
07:27 robklg joined #puppet-openstack
07:28 fandi joined #puppet-openstack
07:33 fandi joined #puppet-openstack
07:33 sileht joined #puppet-openstack
07:37 fandi joined #puppet-openstack
07:44 dgurtner joined #puppet-openstack
07:44 dgurtner joined #puppet-openstack
07:49 stevemar joined #puppet-openstack
08:05 arnaud_orange joined #puppet-openstack
08:20 markvoelker joined #puppet-openstack
08:37 stamak joined #puppet-openstack
08:41 jpena joined #puppet-openstack
09:03 sanjayu joined #puppet-openstack
09:04 robklg joined #puppet-openstack
09:20 markvoelker joined #puppet-openstack
09:22 Madkiss joined #puppet-openstack
09:22 Madkiss hi folks
09:23 Madkiss Just a quick question.with the current puppet-nova implementation, is it possible to run different hypervisor hosts with different hypervisor technologies?
09:35 jistr joined #puppet-openstack
09:59 mmagr joined #puppet-openstack
10:01 kbyrne joined #puppet-openstack
10:21 mattymo Madkiss, if you deploy puppet-nova with different config, yes
10:21 Madkiss hm
10:22 kbyrne joined #puppet-openstack
10:22 Madkiss isn't it true that every nova_config parameter can be set only once, as this creates globally valid resources?
10:22 mattymo you could set per-host config with an ENC, such as heira
10:22 mattymo hiera*
10:22 markvoelker joined #puppet-openstack
10:25 fandi joined #puppet-openstack
10:48 arnaud_orange joined #puppet-openstack
10:50 openstackgerrit Merged stackforge/puppet-cinder: Add an option to not configure RabbitMQ service.  https://review.openstack.org/147420
11:24 markvoelker joined #puppet-openstack
11:47 derekh joined #puppet-openstack
11:59 mmagr joined #puppet-openstack
12:15 tfz joined #puppet-openstack
12:17 pcaruana joined #puppet-openstack
12:24 markvoelker joined #puppet-openstack
12:34 EmilienM|afk good morning!
12:35 tfz joined #puppet-openstack
12:44 dprince joined #puppet-openstack
12:46 openstackgerrit Ryan Hefner proposed stackforge/puppet-manila: Remove enabled_share_backends param to other backends  https://review.openstack.org/139883
12:59 markvoelker joined #puppet-openstack
13:03 morazi joined #puppet-openstack
13:23 aarefiev joined #puppet-openstack
13:27 ddmitriev joined #puppet-openstack
13:39 sanjayu
14:01 dalgaaf joined #puppet-openstack
14:08 arnaud_orange joined #puppet-openstack
14:47 arnaud_orange hi all
14:51 arnaud_orange starting from this commit: 726754fec68b39515a8d17b57665e35c54c42bae
14:51 arnaud_orange i am facing some issues on keystone
14:52 arnaud_orange i can't use keystone_tenant or any keystone_* anymore
14:52 arnaud_orange it seems that moving to openstackclient for keystone_* break the ability to authenticate:
14:52 arnaud_orange Error: /Stage[main]/Keystone::Roles::​Admin/Keystone_tenant[admin]: Could not evaluate: Could not authenticate.
14:54 EmilienM arnaud_orange: do you have the package installed?
14:55 arnaud_orange the package keystone?
14:55 EmilienM arnaud_orange: python-openstackclient
14:56 arnaud_orange yep
14:56 arnaud_orange 0.3.0-1unbutu1
14:58 arnaud_orange i am on ubuntu 14.04 btw
15:04 EmilienM 0.3.0 release is old AFIK
15:04 EmilienM it mays be the reason
15:04 EmilienM arnaud_orange: can you try to use pip to get the last release and check if you still have the problem?
15:04 arnaud_orange ok, this is the one coming from regular repo on ubuntu 14.04
15:04 EmilienM just to find the root cause
15:04 arnaud_orange ok, will try
15:27 rcallawa joined #puppet-openstack
15:36 stevemar joined #puppet-openstack
15:41 stevemar joined #puppet-openstack
15:43 dvorak shouldn't Service[neutron] have a before dependency on all of the neutron_net and neutron_subnet type of resources?
15:44 mdorman joined #puppet-openstack
15:44 dvorak I'm seeing a problem where the package is upgraded, and db sync needs to run, but the service has been restarted by Ubuntu's packaging, but it can't talk to the existing icehouse db
15:45 dvorak hrm, actually, the service has been upgraded by puppet to, and now it's trying to validate it, but the db hasn't been upgraded yet.  This seems broken, let me dig into it
15:45 prad joined #puppet-openstack
15:46 arnaud_orange EmilienM: you were right, it seems better with python-openstackclient from pip source instead of the one from packages
15:47 EmilienM col
15:47 EmilienM cool
15:47 arnaud_orange thank you for the ti
15:47 dvorak also, why doesn't sync_db default to true for neutron, but it does for all other services?
15:47 arnaud_orange tip*
15:51 openstackgerrit Clayton O'Neill proposed stackforge/puppet-neutron: Fixed typo in neutron retry code  https://review.openstack.org/148606
15:53 ericpeterson joined #puppet-openstack
16:01 bitblt joined #puppet-openstack
16:13 dvorak EmilienM: git blame says you added the sync_db neutron code.  What are your thoughts on making it default to on?
16:13 EmilienM don't blame me !
16:13 * EmilienM looking
16:13 EmilienM dvorak: commit id ?
16:13 dvorak it seems like all the other modules have it on by default
16:14 dvorak fea38195
16:15 dvorak it appears that the sync-db needs to notify the service, not use before
16:15 dvorak it appears that if the service is already running, the sync-db won't happen before the restart of the service with just a before
16:15 dvorak but to be honest, this is on the edge of the documentation
16:22 dvorak ah, I see part of my problem.  My question on the sync_db default still stands though :)
16:25 arnaud_orange new question on my side:
16:26 arnaud_orange in this file: https://github.com/stackforge/puppet-keystone/bl​ob/stable/juno/lib/puppet/provider/openstack.rb
16:26 arnaud_orange it seems that it's missing a require 'json'
16:26 arnaud_orange on line 4
16:26 arnaud_orange without it, my keystone catalog throw some errors
16:26 arnaud_orange am i the only one?
16:29 arnaud_orange that kind of error : Error: /Stage[main]/Keystone::Roles:​:Admin/Keystone_user[admin]: Could not evaluate: undefined method `to_json' for #<Hash:0x00000004131790>
16:29 arnaud_orange1 joined #puppet-openstack
16:32 jpena joined #puppet-openstack
16:34 richm joined #puppet-openstack
16:37 off_rhoden joined #puppet-openstack
16:41 * arnaud_orange1 test
16:56 jpena joined #puppet-openstack
17:07 rmoe joined #puppet-openstack
17:08 rwsu joined #puppet-openstack
17:51 aimon joined #puppet-openstack
18:18 EmilienM dprince: I'll produce the patches for puppet-nova & identity URL
18:19 dprince EmilienM: cool. let me know if you change how the logic works with regards to my approach in puppet-neutron
18:20 dprince EmilienM: it would be nice for these things to all align
18:20 EmilienM dprince: I won't
18:20 EmilienM yours is good
18:20 dprince EmilienM: +
18:25 EmilienM dprince: though I have a suggestion for your patch
18:25 EmilienM I would rather construct identity_uri from auth_* deprecated parameters
18:25 EmilienM so people don't have warning in  nova logs
18:26 EmilienM having a "identity_uri_real"
18:26 EmilienM dprince: can you wait this afternoon/later so I can produce a patch to show you how I would do (I can take nova for example)
18:26 EmilienM dprince: ^
18:29 dprince EmilienM: the reason I don't like that approach is because it doesn't align w/ keystoneclient middleware itself
18:29 dprince EmilienM: eventually I suppose I'd like the deprecated parameters to go away entirely
18:31 dprince EmilienM: because that is what keystoneclient auth_token middleware prefers now too: http://git.openstack.org/cgit/opensta​ck/python-keystoneclient/tree/keyston​eclient/middleware/auth_token.py#n202
18:32 dprince EmilienM: the path forward I was trying to take was lets not break people, but also lets move towards having our parameters align better w/ what keystoneclient is going to ultimately use
18:36 EmilienM dprince: if we let the auth_* params in nova::api, with warnings to say it's deprecated but the actual nova_config manages identity_uri (constructed on behalf of what gives auth_params) - would it make sense?
18:37 dprince EmilienM: I think we could do that. But I want to make sure that if someone set auth_uri and identity_uri they don't get the warning messages
18:38 dprince EmilienM: in other words if all you want is an auth_uri_real section (like we do for auth_uri) then fine.
18:38 EmilienM dprince: if auth_uri and identity_uri; then we take identity_uri
18:38 EmilienM since it's a new param, we consider people not stupid enough to configure both
18:38 dprince EmilienM: but that behavior duplicates what keystoneclient already does for you
18:39 dprince EmilienM: the way I see it... why write code in puppet that exists in the middleware already
18:39 EmilienM dprince: all I want is having identity_uri in my nova.conf
18:39 EmilienM and allow people to use old parameters for one release (with a warning for deprecation)
18:39 EmilienM we will drop it in the next release, like we usually do
18:40 dprince EmilienM: okay. Then lets add in an identity_uri_real section to compute it like we do for auth_uri
18:40 dprince EmilienM: if it isn't set.
18:41 EmilienM dprince: do you like the idea?
18:41 EmilienM it avoids having auth_* and identity_uri in nova.conf
18:41 EmilienM since only one of them is enough
18:44 dprince EmilienM: I guess. I feel like we got the parameters backwards in our puppet modules. The auth_port default is 35357... but that is what we end up using for the "identity" uri. auth_uri uses 500
18:44 dprince 0
18:44 dprince EmilienM: I guess that was keystoneclients fault though :)
18:44 dprince EmilienM: do the best we can
18:45 dprince EmilienM: push your patch and I'll review and make mine match if it looks good
18:45 EmilienM dprince: I'll this afternoon
18:46 tfz joined #puppet-openstack
18:54 pradk joined #puppet-openstack
18:54 xarses joined #puppet-openstack
19:11 fvollero joined #puppet-openstack
19:14 openstackgerrit Clayton O'Neill proposed stackforge/puppet-heat: Make package_ensure consistent across classes  https://review.openstack.org/148693
19:31 saju_m joined #puppet-openstack
19:38 imcsk8 joined #puppet-openstack
19:43 arnaud_orange joined #puppet-openstack
19:49 openstackgerrit Clayton O'Neill proposed stackforge/puppet-heat: Run dbsync when engine is upgraded  https://review.openstack.org/148705
19:55 epx998 joined #puppet-openstack
20:00 rcallawa_ joined #puppet-openstack
20:01 gildub joined #puppet-openstack
20:10 mschmitt joined #puppet-openstack
20:33 richm crinkle: ping - have you had a chance to read/digest my last reply to https://review.openstack.org/133601 ?
20:40 openstackgerrit Clayton O'Neill proposed stackforge/puppet-cinder: Run db_sync when upgrading packages  https://review.openstack.org/148716
20:50 crinkle richm: I had not
20:51 crinkle will look and respond in ~30 minutes or so
20:58 richm crinkle: ok - thanks
21:14 rcallawa joined #puppet-openstack
21:15 dprince joined #puppet-openstack
21:20 saju_m joined #puppet-openstack
21:33 pradk joined #puppet-openstack
22:17 crinkle richm: responded
22:17 crinkle thanks for explaining, I wasn't seeing the whole picture
22:36 richm crinkle: Thanks.  I think we are converging on a solution.  If I can assume the use of openstack client 1.0.x, I can get rid of the REST api call.  I'm also going to try to just remove all calls to get or set project/tenant in keystone_user.  From my cursory search, it seems that in almost all cases a keystone_user resource is used in conjunction with a keystone_user_role resource, which should assign the tenant.
22:38 crinkle richm: I don't think you need to assume openstackclient 1.x if we go with my first proposal
22:38 crinkle since 0.3.0 has 'user role list'
22:39 crinkle i was testing it out on ubuntu
22:41 crinkle it would mean `puppet resource keystone_user` wouldn't display default tenants for ldap users, but it would still allow the user to be assigned to a project
22:41 richm ah, ok
22:44 openstackgerrit Clayton O'Neill proposed stackforge/puppet-glance: Glance file backend name is now filesystem  https://review.openstack.org/148740
22:45 openstackgerrit Clayton O'Neill proposed stackforge/puppet-glance: Glance file backend name is renamed to filesystem  https://review.openstack.org/148740
22:46 dvorak apparently the glance team renamed the file backend to filesystem with no backwards compatability code or any mention in the release notes.
23:00 crinkle -_-
23:08 openstackgerrit Clayton O'Neill proposed stackforge/puppet-glance: Glance file backend name is renamed to filesystem  https://review.openstack.org/148740
23:17 xarses joined #puppet-openstack

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