*** Mudpuppy has joined #openstack-ansible | 00:04 | |
*** alop has quit IRC | 00:20 | |
*** bapalm has joined #openstack-ansible | 00:37 | |
*** bapalm has quit IRC | 00:39 | |
*** abitha has quit IRC | 01:02 | |
*** Mudpuppy has quit IRC | 01:05 | |
*** sdake has joined #openstack-ansible | 01:27 | |
*** galstrom_zzz is now known as galstrom | 01:28 | |
*** sdake_ has joined #openstack-ansible | 01:30 | |
*** sdake has quit IRC | 01:34 | |
*** galstrom is now known as galstrom_zzz | 01:46 | |
*** logan2 has quit IRC | 02:08 | |
*** logan2 has joined #openstack-ansible | 02:09 | |
*** sdake_ has quit IRC | 02:09 | |
*** galstrom_zzz is now known as galstrom | 02:10 | |
*** markvoelker has quit IRC | 02:11 | |
*** markvoelker has joined #openstack-ansible | 02:12 | |
*** darrenc is now known as darrenc_afk | 02:12 | |
*** davidself has joined #openstack-ansible | 02:30 | |
*** galstrom is now known as galstrom_zzz | 02:35 | |
*** JRobinson__ is now known as JRobinson__afk | 02:37 | |
*** Mudpuppy has joined #openstack-ansible | 02:57 | |
*** Mudpuppy has quit IRC | 02:58 | |
*** Mudpuppy has joined #openstack-ansible | 02:59 | |
*** markvoelker has quit IRC | 03:04 | |
*** Mudpuppy has quit IRC | 03:09 | |
*** logan2 has quit IRC | 03:14 | |
*** logan2 has joined #openstack-ansible | 03:31 | |
*** darrenc_afk is now known as darrenc | 03:36 | |
*** Mudpuppy has joined #openstack-ansible | 03:48 | |
*** markvoelker has joined #openstack-ansible | 04:04 | |
*** markvoelker has quit IRC | 04:09 | |
*** JRobinson__afk is now known as JRobinson__ | 04:30 | |
*** Mudpuppy has quit IRC | 04:49 | |
*** javeriak has joined #openstack-ansible | 04:56 | |
*** javeriak_ has joined #openstack-ansible | 04:59 | |
*** javeriak has quit IRC | 05:00 | |
*** javeriak has joined #openstack-ansible | 05:01 | |
*** javeriak_ has quit IRC | 05:05 | |
*** javeriak_ has joined #openstack-ansible | 05:05 | |
*** javeriak has quit IRC | 05:06 | |
*** markvoelker has joined #openstack-ansible | 05:51 | |
*** markvoelker has quit IRC | 05:56 | |
*** javeriak_ has quit IRC | 06:11 | |
*** javeriak has joined #openstack-ansible | 07:04 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Verify lxc cache file after download https://review.openstack.org/209788 | 07:16 |
---|---|---|
*** JRobinson__ has quit IRC | 07:28 | |
*** gparaskevas_ has joined #openstack-ansible | 07:31 | |
*** javeriak has quit IRC | 07:38 | |
*** javeriak has joined #openstack-ansible | 07:39 | |
*** javeriak_ has joined #openstack-ansible | 07:42 | |
*** javeriak has quit IRC | 07:43 | |
*** javeriak_ has quit IRC | 07:52 | |
*** markvoelker has joined #openstack-ansible | 07:52 | |
*** sdake has joined #openstack-ansible | 07:54 | |
*** markvoelker has quit IRC | 07:56 | |
evrardjp | good morning | 07:58 |
evrardjp | update branch kilo from .4, cherry picked ceph and the reduced neutron configuration, and I got an issue with dynamic inventory | 07:59 |
evrardjp | http://paste.openstack.org/show/411079/ | 07:59 |
evrardjp | I thought it was worth mentionning | 07:59 |
*** javeriak has joined #openstack-ansible | 08:06 | |
evrardjp | (I've added static_routes also) | 08:17 |
*** gparaskevas_ has quit IRC | 08:20 | |
*** subscope has quit IRC | 08:33 | |
svg | evrardjp: I had something similar yesterday, wa a misconfig in openstack_user_cnfig.yml | 09:02 |
evrardjp | hey svg, I thought you were on holiday | 09:02 |
evrardjp | I've read your email | 09:02 |
svg | I am, prepping for the trip :)\ | 09:02 |
svg | My email? | 09:03 |
evrardjp | sorry, wc | 09:09 |
evrardjp | w/c* | 09:10 |
svg | w/c? | 09:11 |
* svg confused | 09:11 | |
evrardjp | I've seen you had a similar problem, although it was not for the same thing | 09:11 |
evrardjp | wrong channel | 09:11 |
svg | ah | 09:11 |
evrardjp | wrong conversation | 09:11 |
evrardjp | not WC as where ppl go when they need to flush ;) | 09:11 |
evrardjp | anyway, enjoy your holidays | 09:12 |
evrardjp | svg: I don't want to destroy my openstack_inventory.json >< | 09:18 |
evrardjp | you found a way to fix it? | 09:18 |
odyssey4me | mancdaz https://bugs.launchpad.net/openstack-ansible/+bug/1482141 :) | 09:27 |
openstack | Launchpad bug 1482141 in openstack-ansible "inventory-manage should show groups" [Wishlist,Confirmed] | 09:27 |
odyssey4me | evrardjp morning - sorry, was busy elsewhere | 09:28 |
evrardjp | no need, I'll fix it myself ;) | 09:28 |
mancdaz | thanks odyssey4me | 09:28 |
evrardjp | I think I found some kind of anoying thing in the dynamic inventory generation | 09:29 |
evrardjp | I'll try on another environment where I could delete at will | 09:29 |
odyssey4me | evrardjp I'm sure that there are many annoying things in the dynamic inventory :p | 09:29 |
evrardjp | I put 2 times the same host, hoping the dynamic inventory would create 2 times a container on the same host | 09:30 |
evrardjp | I was wrong, it messed up the json file | 09:30 |
odyssey4me | evrardjp quick question - does your git log show this patch? https://github.com/stackforge/os-ansible-deployment/commit/7ccca5ed957ac90749b67f4d5c2854cc3b19f6f7 | 09:31 |
odyssey4me | evrardjp you may find that the inventory-manage script will help you resolve that, otherwise manual intervention :/ | 09:31 |
evrardjp | it does | 09:31 |
odyssey4me | if you want two containers on the same host, the way to do it is to set the affinity | 09:31 |
evrardjp | I was doing manual intervention | 09:32 |
odyssey4me | ok, of you have that patch, then you should remove the user_group_vars file from /etc/openstack_deploy/ | 09:32 |
odyssey4me | if you've customised it, then review what you've customised and add the changes to user_variables | 09:32 |
* evrardjp trusts odyssey4me blindly and rm the file | 09:33 | |
odyssey4me | evrardjp lol, perhaps mv it and name it .bak or something :p | 09:33 |
evrardjp | no, complete trust! :D | 09:34 |
odyssey4me | but yeah, that was an experiment that went wrong - we fixed it, but not before svg and evrardjp started using the stuff in production | 09:34 |
evrardjp | I'll check on your inventory-manage script | 09:34 |
evrardjp | it's not really production | 09:34 |
evrardjp | it's pre-prod :D | 09:34 |
evrardjp | sadly the inventory-manage seems fine... so it proves I have to do manual editing | 09:36 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add irqbalance to apt package list https://review.openstack.org/209840 | 09:38 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Re-order setup-openstack play to match the gate check order https://review.openstack.org/209071 | 09:42 |
evrardjp | odyssey4me: is it normal that a container named tl10_cinder_volumes_container-2b93cf7f in my inventory.json gets in properties is_metal: true, although being a lxc container? | 09:52 |
odyssey4me | is_metal: true would be if it's a host and not a container | 09:52 |
*** markvoelker has joined #openstack-ansible | 09:53 | |
evrardjp | so that's not normal | 09:53 |
odyssey4me | if you're not using lvm storage and are using ceph/nfs/whatever then you should edit the cinder env.d entry to make it include is_metal: false | 09:53 |
odyssey4me | that way your cinder-volume service will be in a container | 09:53 |
evrardjp | found the culprint | 09:54 |
evrardjp | oh you said it | 09:54 |
evrardjp | sorry for your time lost | 09:55 |
*** gparaskevas has joined #openstack-ansible | 09:56 | |
evrardjp | I think it's worth documenting, what do you think Sam-I-Am (and others) ? | 09:56 |
*** markvoelker has quit IRC | 09:57 | |
odyssey4me | evrardjp we actually do have that documented for rpc, but haven't done it for osad so it should be a relatively simple conversion | 09:57 |
evrardjp | I don't see it there http://docs.rackspace.com/rpc/api/v10/bk-rpc-installation/content/sec_configure-cinder.html it's maybe on v11 documentation? | 09:59 |
odyssey4me | evrardjp yeah, the upcoming v11 documentation | 09:59 |
evrardjp | ok | 09:59 |
evrardjp | is that already out somewhere or it's purely internal? This way I can import if you want | 10:00 |
odyssey4me | I'd suggest perhaps registering a bug for it and if you can submit a patch then that'd be good - when Sam-I-Am's got some time he can check the v11 docs and revise your patch :) | 10:00 |
evrardjp | I'll first deploy for Sam-I-Am, this way he gets the results of what he asked me :p | 10:01 |
odyssey4me | yeah, they're still busy doing tweaks and will only release it when v11 releases - but once the docs are done then they typically will take a look at which parts should be converted into doc updates for osad too | 10:01 |
odyssey4me | evrardjp good plan :) | 10:01 |
odyssey4me | quid pro quo | 10:02 |
evrardjp | I never released a bug... Should I add this on launchpad, or directly submit a commit for gerrit? | 10:06 |
odyssey4me | evrardjp ah, so bugs are registered on launchpad: https://bugs.launchpad.net/openstack-ansible | 10:08 |
evrardjp | yeah I'm filling one right now, but I wondered if it was really needed to be declared as bug | 10:08 |
odyssey4me | we have docs and code in the same repo, and we do have other doc bugs too - essentially this is the only place where we can track patches across multiple branches | 10:09 |
odyssey4me | a bug is always good as a reference so that we can remember to backport fixes | 10:09 |
odyssey4me | in this case the backport will need to be from master to kilo | 10:10 |
odyssey4me | we aren't overly strict with doc bugs though - often you can just submit a patch for review, whereas for code we insist that they're either associated with a bug or a blueprint | 10:10 |
evrardjp | Should I add tags, like DocImpact? | 10:11 |
evrardjp | or these tags are only for gerrit? | 10:11 |
odyssey4me | evrardjp if it's a doc bug, not - if you're doing a commit then docimpact will automatically create a bug for the patch :) | 10:11 |
odyssey4me | heh, I forgot about that | 10:11 |
odyssey4me | so yeah, if you do a docimpact flag in the commit message then a bug will automatically be generated for the patch - no need to register the bug yourself :p | 10:12 |
evrardjp | so I shouldn't create a bug in launchpad ;) | 10:12 |
evrardjp | ok | 10:12 |
odyssey4me | sorry - that's a new thing in the project so I'm still getting used to it | 10:12 |
evrardjp | Nobody can know everything ;) | 10:14 |
evrardjp | Although if you do, call me | 10:16 |
odyssey4me | you're not gonna get that call | 10:17 |
*** heww has quit IRC | 10:24 | |
openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Allow cinder-backup to use ceph https://review.openstack.org/209537 | 10:32 |
openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Cinder_volumes_containers are automatically tagged with is_metal https://review.openstack.org/209861 | 10:34 |
openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Cinder_volumes_containers are automatically tagged with is_metal https://review.openstack.org/209861 | 10:35 |
*** markvoelker has joined #openstack-ansible | 10:54 | |
openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Allow cinder-backup to use ceph https://review.openstack.org/209537 | 10:57 |
*** markvoelker has quit IRC | 10:58 | |
*** sdake has quit IRC | 11:07 | |
*** javeriak has joined #openstack-ansible | 11:35 | |
*** javeriak_ has joined #openstack-ansible | 11:41 | |
*** javeria__ has joined #openstack-ansible | 11:42 | |
*** javeriak has quit IRC | 11:45 | |
*** javeriak_ has quit IRC | 11:45 | |
evrardjp | hey odyssey4me: I'm not sure if a bug was created... in launchpad for the 209861 | 11:53 |
odyssey4me | evrardjp it'll only get created when it merges | 11:53 |
evrardjp | ok | 11:53 |
odyssey4me | no worries, I've added Sam-I-Am to the reviewers list, so he'll get an email notifying him about the review | 11:54 |
evrardjp | :D | 11:54 |
*** markvoelker has joined #openstack-ansible | 12:16 | |
*** javeria__ has quit IRC | 12:18 | |
*** javeriak has joined #openstack-ansible | 12:18 | |
*** woodard has joined #openstack-ansible | 12:21 | |
*** javeriak has quit IRC | 12:29 | |
*** bapalm has joined #openstack-ansible | 12:44 | |
*** tlian has joined #openstack-ansible | 13:11 | |
Sam-I-Am | hmm @ docs patches with docimpact? | 13:12 |
Sam-I-Am | thats going to be messy | 13:12 |
*** javeriak has joined #openstack-ansible | 13:22 | |
*** TheIntern has joined #openstack-ansible | 13:29 | |
*** jmckind has joined #openstack-ansible | 13:34 | |
*** KLevenstein has joined #openstack-ansible | 13:44 | |
*** javeriak has quit IRC | 13:48 | |
*** jaypipes has quit IRC | 13:53 | |
*** sigmavirus24_awa is now known as sigmavirus24 | 13:55 | |
*** sdake has joined #openstack-ansible | 13:55 | |
evrardjp | want me to remove the docimpact? | 14:00 |
*** TheIntern is now known as TheIntern_awa | 14:00 | |
*** Mudpuppy has joined #openstack-ansible | 14:00 | |
*** Mudpuppy has quit IRC | 14:01 | |
*** Mudpuppy has joined #openstack-ansible | 14:01 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [unblock gate] Update SHA, Keystone config and Neutron plays https://review.openstack.org/209047 | 14:03 |
*** Mudpuppy_ has joined #openstack-ansible | 14:04 | |
*** TheIntern_awa has quit IRC | 14:05 | |
*** jmckind has quit IRC | 14:05 | |
mattt | svg: howdy | 14:07 |
*** Mudpuppy has quit IRC | 14:07 | |
mattt | svg: thanks for the cinder-backup comment! | 14:07 |
*** jmckind has joined #openstack-ansible | 14:08 | |
mattt | svg: in cpeh_auth.yml, could i not just do this? https://gist.github.com/mattt416/fccb43fe30f7dab621cd | 14:08 |
*** Mudpuppy_ has quit IRC | 14:08 | |
*** Mudpuppy has joined #openstack-ansible | 14:09 | |
odyssey4me | evrardjp nope, leave it as-is | 14:09 |
*** wmlynch has joined #openstack-ansible | 14:09 | |
evrardjp | k | 14:09 |
evrardjp | if I write in user_variables.yml nova_ceph_client: openstack it should overwrite the role's defaults, right? | 14:11 |
mattt | evrardjp: yep | 14:12 |
evrardjp | then I wonder why I have this: http://paste.openstack.org/show/411179/ | 14:13 |
*** sdake has quit IRC | 14:14 | |
evrardjp | (I have same for cinder_ceph_client as nova_ceph_client: openstack) | 14:14 |
evrardjp | I'll debug and tell you more | 14:15 |
mattt | evrardjp: best to just override cinder_ceph_client instead of nova_ceph_client | 14:22 |
mattt | evrardjp: because nova_ceph_client points to cinder_ceph_client | 14:22 |
evrardjp | I've overriden both... | 14:23 |
*** Ti-mo- has quit IRC | 14:23 | |
mattt | evrardjp: ah failed to read :P | 14:24 |
*** Ti-mo has joined #openstack-ansible | 14:24 | |
mattt | git-harry: do these ceph-related vars need to be set in group_vars? https://github.com/stackforge/os-ansible-deployment/blob/master/playbooks/inventory/group_vars/hosts.yml#L191 | 14:25 |
evrardjp | we have a winner :) | 14:36 |
evrardjp | the inventory/group_vars/hosts.yml seems to be read instead of my user_variables | 14:37 |
mattt | evrardjp: that is strange, i just tested this here and my user_variables.yml version still won | 14:37 |
*** jasondot_ is now known as jasondotstar | 14:38 | |
evrardjp | will check | 14:39 |
evrardjp | it's well set in my user_variables... but it took my group_vars | 14:41 |
*** phalmos has joined #openstack-ansible | 14:41 | |
evrardjp | did you try with different users in group_vars and in user_variables ? (just to be sure it's me) | 14:42 |
mattt | evrardjp: this is in /etc/openstack_deploy/user_variables.yml right ? | 14:42 |
mattt | evrardjp: i did yeah | 14:42 |
mattt | i set nova_ceph_client and cinder_ceph_client to test | 14:42 |
evrardjp | and default value to the inventory/group_vars? | 14:43 |
evrardjp | I doubt the is_metal changed anything | 14:44 |
evrardjp | (because it happened on cinder_volumes, which makes me wonder) | 14:44 |
git-harry | evrardjp: are you using the openstack-ansible command? | 14:44 |
evrardjp | yup | 14:44 |
git-harry | are either of those vars in an other vars file in /etc/openstack_deploy or repeated in the same file? | 14:46 |
evrardjp | let me grep that for you | 14:46 |
evrardjp | in /etc/openstack_deploy/* only one occurence, in user_variables.yml | 14:48 |
*** yaya has joined #openstack-ansible | 14:50 | |
evrardjp | the rest of the grep show the default variables (which are supposed to be overridden) and the inventory/group_vars/hosts.yml and all_containers.yml (which is a symlink) | 14:50 |
*** TheIntern has joined #openstack-ansible | 14:53 | |
*** yaya has quit IRC | 14:59 | |
*** bapalm_ has joined #openstack-ansible | 15:00 | |
*** bapalm has quit IRC | 15:01 | |
*** phalmos has quit IRC | 15:01 | |
*** gparaskevas has quit IRC | 15:03 | |
*** jaypipes has joined #openstack-ansible | 15:05 | |
odyssey4me | evrardjp that makes no sense - what version of ansible are you using? | 15:08 |
mattt | evrardjp: git-harry and i have confirmed what you are seeing | 15:09 |
*** phalmos has joined #openstack-ansible | 15:10 | |
evrardjp | 1.9.0.1 | 15:10 |
evrardjp | did you try with a is_metal: True uncommented in /etc/openstack_deploy/env.d/cinder.yml ? | 15:11 |
evrardjp | or better | 15:12 |
mattt | evrardjp: if you comment out the entries in group_vars it seems to work | 15:12 |
evrardjp | can one of you drop me its openstack_inventory.json for the cinder_volume part? | 15:12 |
mattt | evrardjp: oddly when i ran my dummy playbook just dumping out the variables it was printing what i had overriden in user_variables.yml | 15:12 |
svg | mattt: where does cinder_service_backup_program_enabled come from? | 15:16 |
svg | keep in mind that conditional runs also on non cinder hosts, amd should only check the combination where cinder-backup is enabled and for the item where iut handles the cinder-backup user | 15:17 |
mattt | svg: yeah that conditional is bad, i realised after i updated the review :P | 15:17 |
svg | (the example I gave wa not complete, as the cinder-backup-driver variable might not be known or be undefined to the ceph role) | 15:17 |
evrardjp | mattt I've added a dummy playbook | 15:19 |
evrardjp | it indeed dumps the correct variable when I'm running my debug task on a controller node, I'll check with the cinder_volume node | 15:20 |
*** yaya has joined #openstack-ansible | 15:23 | |
mattt | evrardjp maybe svg can answer this issue | 15:23 |
evrardjp | that's really weird | 15:25 |
*** abitha has joined #openstack-ansible | 15:29 | |
mattt | evrardjp: can you log a bug? something is up but i'm not familiar enough w/ ansible to know what the cuase is | 15:29 |
*** abitha has quit IRC | 15:30 | |
*** sdake has joined #openstack-ansible | 15:31 | |
svg | What's the issue exactly? | 15:31 |
*** daneyon has quit IRC | 15:32 | |
*** abitha has joined #openstack-ansible | 15:32 | |
evrardjp | I'll show you with a paset | 15:32 |
evrardjp | paste* | 15:32 |
evrardjp | http://paste.openstack.org/show/411198/ | 15:33 |
*** daneyon has joined #openstack-ansible | 15:33 | |
evrardjp | when using openstack-ansible os-glance/cinder-install.yml, the playbook uses openstack-groupvars as value of the variable glance_ceph_user | 15:34 |
evrardjp | (set like that in inventory/group_vars/hosts.yml) | 15:34 |
evrardjp | while the glance_ceph_user is set in /etc/openstack_deploy/user_variables.yml | 15:35 |
evrardjp | (to openstack in my case) | 15:36 |
evrardjp | note that in the paste I sent you, I've used a dummy playbook that drops the variable value, and the value is correct | 15:36 |
*** galstrom_zzz is now known as galstrom | 15:36 | |
svg | so "glance_ceph_client": "openstack" is what you set in /etc/openstack_deploy/user_variables.yml? | 15:36 |
evrardjp | yup | 15:37 |
evrardjp | glance_ceph_client: openstack-groupvars when in inventory/group_vars/hosts.yml | 15:38 |
evrardjp | is there some kind of cache somewhere ? | 15:39 |
svg | I don;t see any glance_ceph_client set in inventory/group_vars/hosts.yml | 15:40 |
svg | so you have an alternative value set there? not sure i follow | 15:41 |
evrardjp | there wasn't at first | 15:41 |
evrardjp | there was cinder_ceph_client | 15:41 |
evrardjp | so I added the others | 15:41 |
svg | 1/ what others, and 2/ why add something in inventory/group_vars/hosts.yml? | 15:42 |
svg | the way osad works, inventory group vars are part of the 'code', better avoid changing that | 15:42 |
*** jaypipes has quit IRC | 15:43 | |
evrardjp | I noticed that cinder_ceph_client was there... It had cinder as value | 15:43 |
evrardjp | I had in my user_variables the cinder_ceph_client set to openstack | 15:43 |
evrardjp | I wondered first why cinder_ceph_client had the value "cinder" while running my playbook | 15:44 |
evrardjp | I changed the value to openstack-groupvars, to see | 15:44 |
evrardjp | because os-glance-install runs before os-cinder-install in the openstack-install playbook, I tried the same with glance | 15:44 |
evrardjp | that's just it | 15:44 |
evrardjp | if I remove all these variables from group_vars/hosts, I guess it will run fine.... so it's a bug to have cinder_ceph_client in inventory/group_vars/hosts.yml | 15:45 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Remove unused python clients from repo & tempest https://review.openstack.org/209998 | 15:45 |
evrardjp | that's my conclusion | 15:46 |
evrardjp | but still, I wonder why my user_variables are overriden by inventory/group_vars/hosts.yml | 15:47 |
mattt | why are those vars set in inventory/group_vars/hosts.yml anyway? | 15:47 |
evrardjp | and why it isn't like that when I run my dummy playbook | 15:47 |
evrardjp | I didn't set them | 15:47 |
evrardjp | it's like that in master | 15:47 |
mattt | evrardjp: when you run the dummy playbook i guess it's not loading in the role defaults because it's not loading roles | 15:47 |
mattt | evrardjp: and it's also not loading the roles/ceph_client/vars file | 15:48 |
mattt | evrardjp: no i know, but i'm wondering why they were committed to that file | 15:48 |
evrardjp | that's for sure | 15:48 |
evrardjp | but if they are in vars, they won't be overriden by anything | 15:48 |
evrardjp | hopefully these are not in vars | 15:49 |
evrardjp | just in default | 15:49 |
evrardjp | so they are overriden | 15:49 |
evrardjp | by your user_variables and the inventory/group_vars/hosts.yml | 15:49 |
evrardjp | so that's perfectly correct | 15:49 |
evrardjp | the role is not to blame ;) | 15:49 |
evrardjp | my install maybe | 15:50 |
evrardjp | wait | 15:50 |
evrardjp | you're right | 15:50 |
svg | baically | 15:50 |
evrardjp | there is a glance_ceph_client in vars/main.yml | 15:50 |
svg | basically: defaults are set in role defaults, unless they are needed cross-roles, and then they live in those group vars | 15:51 |
svg | then everything you want to customize go in user_variables | 15:51 |
evrardjp | I agree with you | 15:51 |
*** TheIntern has quit IRC | 15:51 | |
evrardjp | so cinder_ceph_user shouldn't be defined in group_vars | 15:51 |
evrardjp | that's the first point | 15:52 |
svg | the default is defined there. why? because that var is the default value for nova_ceph_client | 15:52 |
svg | those are the defaults, override in user_varsibales | 15:52 |
mattt | svg: but those defaults are also in the roles, so it's a bit confusing | 15:53 |
mattt | (also i don't really know ansible :P) | 15:53 |
evrardjp | defaults should be in the role only, except if they are the default for a group of hosts | 15:53 |
*** phalmos has quit IRC | 15:53 | |
svg | all defaults are set in role defaults, and some variables are also needed cross-roles, and then they live also in those group vars | 15:54 |
evrardjp | here we don't override the default variable from the playbook in a grouped way | 15:54 |
mattt | svg: ah cross role defaults | 15:54 |
svg | evrardjp: the way osad does it inventory, is a bit non-ansibly actually | 15:54 |
evrardjp | I agree | 15:54 |
evrardjp | for cross role, I see the point | 15:54 |
svg | role defaults are the way to make sure the code ha defaults | 15:55 |
evrardjp | although, for cross role you can only check in user_variables | 15:55 |
evrardjp | you can afford to only check in user_variables | 15:55 |
svg | yes and not: user_variables are what you customize: that is the 'data' to osad | 15:55 |
svg | group_vars are here in osad inherently part of the code | 15:56 |
*** jaypipes has joined #openstack-ansible | 15:56 | |
svg | at least that's how I see it | 15:56 |
evrardjp | let me rephrase my question differently. If cinder_ceph_user is needed as part of group_variable, why isn't glance or nova ? I find that arbitrary | 15:56 |
svg | One wants to avoid to change things in group_vars, as then you change what you get from upstream | 15:56 |
*** alop has joined #openstack-ansible | 15:56 | |
evrardjp | how I see it: when I removed the line from group_vars it worked, when I let it, it doesn't | 15:57 |
svg | and that makes it a customization to the upstream project. avoid that | 15:57 |
svg | So the question is why did you put it there in the first place :) | 15:57 |
evrardjp | I didn't! | 15:57 |
evrardjp | that's my message! | 15:57 |
evrardjp | or I don't remember | 15:58 |
svg | to answer your rephrase question ^^ : cinder_ceph_user holds the default value for nova_ceph_user | 15:58 |
svg | inventory/group_vars/hosts.yml:nova_ceph_client: '{{ cinder_ceph_client }}' | 15:58 |
svg | Why? because nova needs access to both cinder backend and to nova backend | 15:58 |
svg | those are two different ceph pools | 15:59 |
svg | and nova is the only one that accesses both | 15:59 |
svg | which is why those vars needs to be defaulted cross-roles | 15:59 |
*** TheIntern has joined #openstack-ansible | 15:59 | |
svg | evrardjp: must be it, if you dont remember, you sureley can see what changed in git compared to the upstream code? | 16:00 |
evrardjp | https://github.com/stackforge/os-ansible-deployment/blob/master/playbooks/inventory/group_vars/hosts.yml | 16:00 |
evrardjp | cinder_ceph_client: cinder | 16:00 |
evrardjp | I'm not to blame there... | 16:00 |
odyssey4me | community meeting in #openstack-meeting-4 cloudnull, mattt, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, Sam-I-Am, odyssey4me, serverascode, rromans, mancdaz, dolphm, _shaps_, BjoernT, claco, echiu, dstanek, jwagner, ayoung | 16:00 |
*** yaya has quit IRC | 16:00 | |
evrardjp | sorry I get heated because I feel misunderstood... >< | 16:01 |
svg | yes, cinder_ceph_client: cinder is there inventory/group_vars/hosts.yml | 16:02 |
svg | again, because it's a value used by the nova role | 16:02 |
prometheanfire | odyssey4me: oh thanks for the warning | 16:02 |
evrardjp | my issue mainly comes because it overwrites my user_variables | 16:03 |
svg | Oh | 16:03 |
svg | so inventory/group_vars/hosts.yml > user_variables? | 16:03 |
svg | That should not happen! | 16:03 |
evrardjp | yes | 16:03 |
evrardjp | no | 16:03 |
odyssey4me | prometheanfire add your name to the wiki for a warning :) | 16:04 |
evrardjp | and it only happened in os-*-install, because my dummy-playbook works fine | 16:04 |
svg | tbh, I use these default values, and i guess most testers did, so it's possible you found some bug | 16:04 |
prometheanfire | odyssey4me: we have a wiki? what type of shame is this? | 16:04 |
evrardjp | yes It's definitely that, but I didn't like being blamed :p | 16:04 |
Apsu | Bugs? In our code?! What sorcery is this | 16:04 |
Apsu | Surely it must be your fault :P | 16:04 |
evrardjp | I take the blame entirely! | 16:04 |
*** phalmos has joined #openstack-ansible | 16:04 | |
evrardjp | damn team spirit. | 16:05 |
prometheanfire | perfect, we need a fall guy | 16:05 |
Apsu | Oh ok, excellent. Feeling better already | 16:05 |
odyssey4me | prometheanfire https://wiki.openstack.org/wiki/Meetings/openstack-ansible#Regular_attendees :) | 16:05 |
prometheanfire | I mean patsy | 16:05 |
Apsu | prometheanfire: Sir Patsy, III, thank you | 16:05 |
svg | k, evrardjp looking back at your paste, you say the error happens with glance | 16:05 |
evrardjp | glance cinder whatever | 16:05 |
evrardjp | same combat | 16:06 |
svg | euhm, not really | 16:06 |
svg | o that task that fails is TASK: [ceph_client | Retrieve keyrings for openstack clients from ceph cluster] *** | 16:07 |
svg | running on glance hosts | 16:07 |
evrardjp | yes ? | 16:07 |
evrardjp | I had the failure on both glance and cinder, that's what I meant | 16:07 |
svg | Which means there it uses glance_ceph_client for that task | 16:08 |
evrardjp | glance was just a tool for me to discover the issue: the overwriting of my user_variables | 16:08 |
svg | so somehow glance_ceph_client gets set to openstack-groupvars in that example | 16:08 |
evrardjp | let's forget about glance_ceph_client | 16:08 |
evrardjp | I'll dump another paste with cinder if you prefer to | 16:09 |
evrardjp | I've reset the groups vars to default value (so cinder_ceph_client: cinder) | 16:09 |
evrardjp | and the task will fail because there is no client.cinder in my ceph env | 16:09 |
svg | ok | 16:10 |
svg | but hard to debug if I don't see the full monty | 16:13 |
svg | can you push your exact osad branch to github perhaps? that and user_variables? that should not hold anything sensitive I guess | 16:14 |
*** yaya has joined #openstack-ansible | 16:14 | |
*** sdake has quit IRC | 16:14 | |
*** TheIntern has quit IRC | 16:15 | |
evrardjp | I'll do what I can tomorrow if you don't mind | 16:23 |
evrardjp | you're in holidays starting on monday, or starting tomorrow? | 16:23 |
*** phalmos has quit IRC | 16:27 | |
svg | tomorrow and saturday I'll be on the road mostly | 16:28 |
svg | feel free to mail me the info, might help to not get bored at the road motel tomorrow :) | 16:29 |
evrardjp | lol | 16:31 |
*** jmckind has quit IRC | 16:39 | |
*** phalmos has joined #openstack-ansible | 16:43 | |
*** javeriak has joined #openstack-ansible | 16:53 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Enable tempest testing of ceilometer https://review.openstack.org/209568 | 16:54 |
*** spotz_zzz is now known as spotz | 16:56 | |
*** phalmos has quit IRC | 16:56 | |
*** bapalm_ has quit IRC | 17:01 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Re-order setup-openstack play to match the gate check order https://review.openstack.org/209071 | 17:12 |
*** sdake has joined #openstack-ansible | 17:33 | |
openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Set the correct variable in run-upgrade.sh https://review.openstack.org/210054 | 17:35 |
*** markvoelker has quit IRC | 17:44 | |
*** yaya has quit IRC | 17:44 | |
*** TheIntern has joined #openstack-ansible | 17:49 | |
*** javeriak_ has joined #openstack-ansible | 17:50 | |
*** javeriak has quit IRC | 17:50 | |
*** javeriak has joined #openstack-ansible | 17:54 | |
*** javeriak_ has quit IRC | 17:54 | |
*** javeriak_ has joined #openstack-ansible | 17:59 | |
*** spotz is now known as spotz_zzz | 18:02 | |
*** yaya has joined #openstack-ansible | 18:02 | |
*** javeriak has quit IRC | 18:02 | |
*** bapalm has joined #openstack-ansible | 18:04 | |
*** markvoelker has joined #openstack-ansible | 18:05 | |
*** phalmos has joined #openstack-ansible | 18:06 | |
*** metral is now known as metral_zzz | 18:09 | |
*** metral_zzz is now known as metral | 18:12 | |
*** phalmos has quit IRC | 18:16 | |
*** phalmos has joined #openstack-ansible | 18:17 | |
*** daneyon has quit IRC | 18:20 | |
*** daneyon has joined #openstack-ansible | 18:20 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Remove unused python clients from repo & tempest https://review.openstack.org/209998 | 18:43 |
*** javeriak has joined #openstack-ansible | 18:44 | |
odyssey4me | sigmavirus24 palendae I had removed swift by mistake, hence the failed run. The above patch set restores that var. | 18:44 |
*** javeriak_ has quit IRC | 18:45 | |
*** javeriak has quit IRC | 18:53 | |
*** phalmos has quit IRC | 18:53 | |
*** jmckind has joined #openstack-ansible | 18:53 | |
*** harlowja has quit IRC | 18:56 | |
*** alextricity has quit IRC | 18:56 | |
*** phalmos has joined #openstack-ansible | 19:02 | |
*** yaya has quit IRC | 19:05 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Improve Keystone Apache configuration https://review.openstack.org/209047 | 19:08 |
*** javeriak has joined #openstack-ansible | 19:09 | |
*** javeriak_ has joined #openstack-ansible | 19:10 | |
*** javeriak has quit IRC | 19:13 | |
*** javeriak has joined #openstack-ansible | 19:21 | |
*** javeriak_ has quit IRC | 19:22 | |
*** javeriak_ has joined #openstack-ansible | 19:25 | |
*** Mudpuppy has quit IRC | 19:25 | |
*** Mudpuppy has joined #openstack-ansible | 19:26 | |
*** javeriak has quit IRC | 19:29 | |
*** javeriak has joined #openstack-ansible | 19:30 | |
*** yaya has joined #openstack-ansible | 19:31 | |
*** javeriak_ has quit IRC | 19:33 | |
*** javeriak_ has joined #openstack-ansible | 19:38 | |
*** yaya has quit IRC | 19:38 | |
*** javeriak has quit IRC | 19:42 | |
*** javeriak has joined #openstack-ansible | 19:47 | |
*** javeriak_ has quit IRC | 19:48 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated master for new dev work - Liberty-2 https://review.openstack.org/199126 | 19:55 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated master for new dev work - Liberty-2 https://review.openstack.org/199126 | 19:56 |
*** TheIntern has quit IRC | 20:00 | |
Sam-I-Am | odyssey4me: is gating undead yet? | 20:08 |
odyssey4me | Sam-I-Am not yet, working through a gate check as we speak | 20:09 |
*** logan2 has quit IRC | 20:12 | |
*** logan2 has joined #openstack-ansible | 20:12 | |
richoid | hi all - known issue with repo container new setup from master trunk? > One or more undefined variables: 'dict object' has no attribute 'repo_pubkey' | 20:15 |
sigmavirus24 | lol | 20:16 |
sigmavirus24 | richoid: interesting | 20:17 |
sigmavirus24 | I was seeing something similar on kilo ... are you using --limit? | 20:17 |
sigmavirus24 | i.e., have you run something, it failed, and you're re-running with --limit? | 20:17 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Update tempest configuration https://review.openstack.org/210107 | 20:20 |
odyssey4me | sigmavirus24 it's entirely possible that the ssh key distribution would require the whole play to be run, rather than something limited via --limit or via --tags | 20:21 |
sigmavirus24 | odyssey4me: that's why I'm asking | 20:21 |
*** KLevenstein has quit IRC | 20:21 | |
sigmavirus24 | that's what happened to me | 20:21 |
odyssey4me | richoid ^ | 20:21 |
sigmavirus24 | when I removed --limit from the equation it worked again | 20:21 |
odyssey4me | sounds like a good thing to know - similar to some notes from bgmccollum in last night's scrollback | 20:22 |
odyssey4me | I think I've managed to split the megapatch into its component parts - now to see what passes the gate :) | 20:22 |
*** yaya has joined #openstack-ansible | 20:23 | |
bgmccollum | yeah --limit no bueno | 20:23 |
bgmccollum | unless you're intimately aware of the playbooks, and surgically slicing things | 20:23 |
sigmavirus24 | yeah | 20:24 |
odyssey4me | bgmccollum if we could tag the tasks better, we're open to suggestion/reviews :) | 20:25 |
odyssey4me | *tags (not tasks) | 20:25 |
odyssey4me | but I suppose tasks too :) | 20:25 |
*** jmckind_ has joined #openstack-ansible | 20:28 | |
bgmccollum | odyssey4me: not sure best way to handle, since key distribution success requires always running a set of tasks on compute[0] (the population step). is delegate_to is affected by --limit? | 20:29 |
bgmccollum | i know key distribution was refactored a bit...so maybe a moot point. but that particular problem keeps creeping up, since the docs show using --limit when adding a compute node... | 20:30 |
*** jmckind has quit IRC | 20:32 | |
odyssey4me | bgmccollum that issue is being dealt with, but perhaps we could be better at making the roles idempotent and using tags to isolate things properly | 20:32 |
odyssey4me | the idempotence would make the play run through existing compute nodes quickly enough | 20:33 |
bgmccollum | indeed | 20:33 |
odyssey4me | personally I think that perhaps the compute node tasks should be in their own role | 20:33 |
odyssey4me | but that's up for debate | 20:33 |
*** yaya has quit IRC | 20:33 | |
*** jmckind_ has quit IRC | 20:33 | |
*** jmckind has joined #openstack-ansible | 20:38 | |
*** javeriak_ has joined #openstack-ansible | 20:44 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add swift service region to group vars https://review.openstack.org/209326 | 20:47 |
*** javeriak has quit IRC | 20:48 | |
*** logan2 has quit IRC | 20:48 | |
*** bapalm has quit IRC | 20:50 | |
*** KLevenstein has joined #openstack-ansible | 20:57 | |
odyssey4me | stevelle ^ sorry, I couldn't restrain myself from resolving a spelling error :/ | 20:58 |
odyssey4me | in restrospect it was a pointless additional patch set | 20:58 |
stevelle | odyssey4me: more arm curls, less coding :P | 20:59 |
odyssey4me | mwahahaha | 20:59 |
*** britthou_ has quit IRC | 21:05 | |
*** woodard_ has joined #openstack-ansible | 21:07 | |
*** woodard has quit IRC | 21:10 | |
*** woodard_ has quit IRC | 21:11 | |
*** jmckind has quit IRC | 21:19 | |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Enable tempest testing of ceilometer https://review.openstack.org/209568 | 21:21 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Re-order setup-openstack play to match the gate check order https://review.openstack.org/209071 | 21:21 |
*** markvoelker has quit IRC | 21:22 | |
*** Mudpuppy_ has joined #openstack-ansible | 21:26 | |
*** Mudpuppy has quit IRC | 21:29 | |
*** Mudpuppy_ has quit IRC | 21:31 | |
*** phalmos has quit IRC | 21:34 | |
*** harlowja has joined #openstack-ansible | 21:34 | |
*** yaya has joined #openstack-ansible | 21:36 | |
*** KLevenstein has quit IRC | 21:36 | |
*** galstrom is now known as galstrom_zzz | 21:38 | |
*** harlowja has quit IRC | 21:39 | |
*** harlowja has joined #openstack-ansible | 21:39 | |
*** KLevenstein has joined #openstack-ansible | 21:39 | |
*** yaya has quit IRC | 21:59 | |
richoid | sigmavirus24, yes it is with --limit | 22:02 |
richoid | sigmavirus24, It's a new aio setup, the first run of setup-everything had a couple failures, so i re-ran with --limit | 22:02 |
sigmavirus24 | richoid: --limit doesn't work great | 22:03 |
openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated master for new dev work - Liberty-2 https://review.openstack.org/199126 | 22:03 |
richoid | ill re-run without --limit, thx | 22:04 |
*** javeriak_ has quit IRC | 22:11 | |
*** KLevenstein has quit IRC | 22:16 | |
*** abitha has quit IRC | 22:23 | |
openstackgerrit | Merged stackforge/os-ansible-deployment: Remove unused python clients from repo & tempest https://review.openstack.org/209998 | 22:23 |
*** abitha has joined #openstack-ansible | 22:26 | |
*** abitha has quit IRC | 22:27 | |
*** abitha has joined #openstack-ansible | 22:28 | |
*** britthouser has joined #openstack-ansible | 22:38 | |
*** markvoelker has joined #openstack-ansible | 22:50 | |
openstackgerrit | Nolan Brubaker proposed stackforge/os-ansible-deployment: Add an inventory Juno -> Kilo migrate script https://review.openstack.org/210152 | 23:00 |
*** JRobinson__ has joined #openstack-ansible | 23:03 | |
*** logan2 has joined #openstack-ansible | 23:06 | |
*** abitha has quit IRC | 23:07 | |
*** sdake has quit IRC | 23:38 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!