| *** vp has joined #openstack-ansible | 00:00 | |
| *** vp is now known as Guest28353 | 00:00 | |
| *** Guest28353 has quit IRC | 00:09 | |
| *** markvoelker has joined #openstack-ansible | 00:12 | |
| *** dave-mccowan has joined #openstack-ansible | 00:15 | |
| *** germs has joined #openstack-ansible | 00:18 | |
| *** germs has quit IRC | 00:33 | |
| *** woodard has joined #openstack-ansible | 00:44 | |
| *** markvoelker has quit IRC | 00:45 | |
| *** dave-mccowan has quit IRC | 01:30 | |
| *** dave-mccowan has joined #openstack-ansible | 01:32 | |
| *** markvoelker has joined #openstack-ansible | 01:42 | |
| *** dave-mcc_ has joined #openstack-ansible | 01:49 | |
| *** dave-mccowan has quit IRC | 01:51 | |
| *** dave-mccowan has joined #openstack-ansible | 01:57 | |
| *** dave-mcc_ has quit IRC | 01:59 | |
| *** markvoelker has quit IRC | 02:16 | |
| *** dave-mcc_ has joined #openstack-ansible | 02:17 | |
| *** dave-mccowan has quit IRC | 02:18 | |
| *** esberglu has joined #openstack-ansible | 02:26 | |
| *** esberglu has quit IRC | 02:31 | |
| *** captaindave has quit IRC | 02:54 | |
| *** captaindave has joined #openstack-ansible | 02:55 | |
| *** dave-mcc_ has quit IRC | 02:55 | |
| *** markvoelker has joined #openstack-ansible | 03:13 | |
| *** markvoelker has quit IRC | 03:46 | |
| *** gouthamr has joined #openstack-ansible | 03:55 | |
| *** esberglu has joined #openstack-ansible | 04:14 | |
| *** esberglu has quit IRC | 04:19 | |
| *** gouthamr has quit IRC | 04:30 | |
| *** markvoelker has joined #openstack-ansible | 04:43 | |
| *** ecelik has joined #openstack-ansible | 05:15 | |
| *** markvoelker has quit IRC | 05:17 | |
| *** logan- has quit IRC | 05:18 | |
| *** logan- has joined #openstack-ansible | 05:21 | |
| *** thorst has joined #openstack-ansible | 05:41 | |
| *** logan- has quit IRC | 05:48 | |
| *** thorst has quit IRC | 05:51 | |
| *** thorst has joined #openstack-ansible | 05:51 | |
| *** thorst has quit IRC | 05:52 | |
| *** logan- has joined #openstack-ansible | 05:53 | |
| *** cshen has joined #openstack-ansible | 06:02 | |
| *** esberglu has joined #openstack-ansible | 06:02 | |
| *** ivve has quit IRC | 06:05 | |
| *** esberglu has quit IRC | 06:07 | |
| *** jvidal has joined #openstack-ansible | 06:10 | |
| *** Oku_OS-away is now known as Oku_OS | 06:10 | |
| *** markvoelker has joined #openstack-ansible | 06:14 | |
| *** cshen has quit IRC | 06:20 | |
| *** cshen has joined #openstack-ansible | 06:21 | |
| *** eumel8 has joined #openstack-ansible | 06:24 | |
| *** ivve has joined #openstack-ansible | 06:29 | |
| *** ansibler has quit IRC | 06:30 | |
| *** shardy has joined #openstack-ansible | 06:37 | |
| *** drifterza has joined #openstack-ansible | 06:46 | |
| *** markvoelker has quit IRC | 06:47 | |
| hwoarang | good morning | 06:52 |
|---|---|---|
| *** chas has joined #openstack-ansible | 06:54 | |
| *** ecelik has left #openstack-ansible | 06:58 | |
| *** pcaruana has joined #openstack-ansible | 07:08 | |
| *** mbuil has joined #openstack-ansible | 07:10 | |
| evrardjp | good morning | 07:11 |
| *** Taseer has quit IRC | 07:12 | |
| *** taseer2 has joined #openstack-ansible | 07:13 | |
| *** electrofelix has joined #openstack-ansible | 07:18 | |
| *** mrch has joined #openstack-ansible | 07:29 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron stable/pike: Update paste, policy and rootwrap configurations 2017-09-28 https://review.openstack.org/508139 | 07:37 |
| *** markvoelker has joined #openstack-ansible | 07:44 | |
| *** esberglu has joined #openstack-ansible | 07:51 | |
| *** pbandark has joined #openstack-ansible | 07:51 | |
| *** esberglu has quit IRC | 07:56 | |
| *** taseer2 is now known as Taseer | 08:14 | |
| *** markvoelker has quit IRC | 08:18 | |
| hwoarang | just to confirm, can values in user_secrets be overriden by openstack_user_config? | 08:32 |
| *** gkadam has joined #openstack-ansible | 08:33 | |
| andymccr | hwoarang: i dont thinkso - user_secrets would be passed using -e i think, so that would take precedence. | 08:34 |
| hwoarang | i can't find exactly how user_secrets is used :/ | 08:36 |
| odyssey4me | ivve use the container destroy playbook to destroy the containers, then remove the contains from the inventory using inventory-manage, then remove the host group/host mapping in your openstack_user_config/conf.d | 08:47 |
| evrardjp | andymccr: welcome back! | 08:47 |
| odyssey4me | hwoarang user_* files are used as extra-vars, so they have highest precedence | 08:47 |
| odyssey4me | openstack_user_config is in inventory, so it's almost the lowest precedence | 08:47 |
| hwoarang | odyssey4me: ok thank you. but where exactly in the code is the user_secrets used? | 08:48 |
| hwoarang | just looking at the details because i am curious :) | 08:48 |
| evrardjp | hwoarang: -e | 08:55 |
| *** TomMc has joined #openstack-ansible | 08:56 | |
| evrardjp | hwoarang: let me point you to the code | 08:56 |
| evrardjp | https://github.com/openstack/openstack-ansible/blob/master/scripts/bootstrap-ansible.sh#L245 | 08:57 |
| evrardjp | the wrapper gets that | 08:57 |
| *** armaan has joined #openstack-ansible | 08:58 | |
| hwoarang | hmm i wonder if it's possible to add passwords in user_variables.yml instead | 08:58 |
| hwoarang | it should be possible since both user_* files are treated the same | 09:01 |
| hwoarang | thanks evrardjp | 09:01 |
| evrardjp | hwoarang: yes it is | 09:01 |
| evrardjp | but because the way they are listed, don't forget about the precedence ;) | 09:02 |
| evrardjp | but it's technically possible to have user_blabla.yml or user_zzz.yml | 09:02 |
| hwoarang | yep | 09:03 |
| hwoarang | sneaky | 09:03 |
| evrardjp | not really :p | 09:09 |
| *** markvoelker has joined #openstack-ansible | 09:15 | |
| neith | hello, Is it possible to safely skip repo related tasks in set-infrastructure.yml using tags? | 09:29 |
| neith | openstack-ansible setup-infrastructure.yml --skip-tags repo-build,repo-server | 09:30 |
| neith | ? | 09:30 |
| *** epalper has joined #openstack-ansible | 09:31 | |
| epalper | hi, i am trying to run nova role with openstack-ansible command: openstack-ansible -i /tmp/.xci-deploy-env/openstack-ansible/playbooks/inventory /tmp/.xci-deploy-env/openstack-ansible/playbooks/os-nova-install.yml -vvv. But it fails with following error: https://paste.ubuntu.com/25659673/ | 09:35 |
| epalper | can you please look into it ? | 09:35 |
| *** esberglu has joined #openstack-ansible | 09:39 | |
| neith | evrardjp: odyssey4me any idea on my question about skipping repo related tasks ? | 09:40 |
| evrardjp | neith: why not using the playbooks instead? | 09:41 |
| odyssey4me | neith they will auto-skip if requirements haven't changed in any recent version | 09:41 |
| neith | evrardjp: cause repo tasks are soooooo long | 09:41 |
| odyssey4me | but yes, look at the playbooks - they should have tags | 09:41 |
| neith | ok | 09:41 |
| evrardjp | neith: I meant running the playbooks, and skipping playbooks you don't need | 09:42 |
| odyssey4me | neith all recent versions of newton onwards will not run any of the build tasks if requirements haven't changed | 09:42 |
| evrardjp | neith: a necessary evil sometimes :) | 09:42 |
| evrardjp | and odyssey4me has done a tough job on improving this, and cracking to small pieces | 09:42 |
| evrardjp | :D | 09:42 |
| neith | I completely crashed my rabbit cluster and I think my last option is to rebuild all the containers | 09:42 |
| evrardjp | just rebuild rabbit | 09:43 |
| neith | all rabbit containers of course | 09:43 |
| evrardjp | if you haven't change credentials, why change the rest? | 09:43 |
| evrardjp | yeah, faster than recovery :p | 09:43 |
| evrardjp | ask jmccrory :p | 09:43 |
| *** esberglu has quit IRC | 09:43 | |
| neith | odyssey4me: including 14.2.8 ? | 09:45 |
| odyssey4me | neith yes | 09:45 |
| neith | Is it possible to run openstack-ansible setup-infrastructure.yml --limit rabbitmq_all ? | 09:46 |
| *** markvoelker has quit IRC | 09:48 | |
| *** overskylab has joined #openstack-ansible | 10:03 | |
| overskylab | do you have roadmap to deploy panko on this project | 10:04 |
| *** ejb1123 has quit IRC | 10:04 | |
| admin0 | \o | 10:13 |
| admin0 | question on aio .. if my server has 2 interfaces, how to tell aio to use eth1 instead of eth0 | 10:13 |
| *** overskylab has quit IRC | 10:14 | |
| evrardjp | neith: use the playbooks instead | 10:15 |
| evrardjp | easier | 10:15 |
| evrardjp | no extra tasks for repo and all that jazz | 10:15 |
| evrardjp | admin0: I think that would require a patch | 10:17 |
| *** ricardoas has quit IRC | 10:23 | |
| admin0 | evrardjp, ok .. thanks | 10:23 |
| admin0 | in pike, with the cells, is it possible to move an existing running cluster to the new cells (pod) ? | 10:24 |
| admin0 | from documentation, the default is treated as pod1, and from the config, can add 2nd pod etc ? | 10:25 |
| epalper | hi, have anyone had a chance to look at my issue ? | 10:33 |
| *** markvoelker has joined #openstack-ansible | 10:45 | |
| eumel8 | epalper: line 106. you skipped resolvconf_enabled so all inventory variables are empty and the task to install resolv.conf will fail | 10:52 |
| *** DanyC has joined #openstack-ansible | 11:04 | |
| *** pbandark has quit IRC | 11:06 | |
| *** pbandark1 has joined #openstack-ansible | 11:06 | |
| *** dave-mccowan has joined #openstack-ansible | 11:07 | |
| *** pbandark1 is now known as pbandark | 11:08 | |
| *** markvoelker has quit IRC | 11:19 | |
| admin0 | evrardjp, i upgraded from ocata -> pike .. can you tell me the line to override one cinder server to have is_metal: true from the config file | 11:33 |
| odyssey4me | admin0 I believe that evrardjp gave you the configs last week. I'd suggest going through the eavesdrop logs. | 11:36 |
| evrardjp | admin0: you can now use hostvars directly | 11:36 |
| evrardjp | if you upgraded you have new features that allow us simplicity, but you're not forced to migrate to hostvars | 11:36 |
| evrardjp | you can keep it as it. | 11:36 |
| admin0 | "you have new features that allow simplicity" -- what are these ? | 11:37 |
| *** jamesdenton has quit IRC | 11:39 | |
| admin0 | so in pike, is the big config being broken down to individual service.yml inside conf.d ? | 11:44 |
| *** smatzek has joined #openstack-ansible | 11:44 | |
| admin0 | in the example, some yml start with --- , some don't .. is it documentation mistake or it works both way .. and if it works both ways, include --- or remove --- ? | 11:46 |
| epalper | eumel8: what do you recommend to run resolvconf_enabled ? | 11:52 |
| epalper | line 106. while file you are talking about ? | 11:52 |
| eumel8 | epalper: your paste file | 11:56 |
| *** jamesdenton has joined #openstack-ansible | 11:58 | |
| *** jamesdenton has quit IRC | 12:02 | |
| *** jamesdenton has joined #openstack-ansible | 12:06 | |
| *** markvoelker has joined #openstack-ansible | 12:06 | |
| epalper | eumel8: ok, do you want to have resolvconf_enabled: "{{ groups['unbound'] is defined and groups['unbound'] | length > 0 }}" to be run before executing a role ? | 12:06 |
| eumel8 | epalper: I think you want definitly use resolv.conf | 12:09 |
| neith | evrardjp: I really need to dig to understand which playbooks are ruuned in setup host/infra/openstack . | 12:11 |
| evrardjp | neith: that should be very simple | 12:12 |
| evrardjp | they are very explicit :p open your setup-infrastructure.yml or setup-openstack.yml you'll understand :) | 12:13 |
| evrardjp | neith: don't forget you can run openstack-ansible playbook1 playbook2 playbook3 (thanks odyssey4me for the tip that I completely forgot!) | 12:13 |
| neith | evrardjp: It probably is, but my ass is between openstack and Kubernetes currently... I need to fight to get some extra time on OS | 12:13 |
| epalper | eumel8: yes, it works after resolvconf_enabled to true | 12:16 |
| epalper | thanks | 12:16 |
| eumel8 | :) | 12:17 |
| *** captaindave has quit IRC | 12:19 | |
| evrardjp | neith: ) | 12:33 |
| *** woodard has quit IRC | 12:37 | |
| *** woodard has joined #openstack-ansible | 12:38 | |
| *** smatzek has quit IRC | 12:43 | |
| *** smatzek has joined #openstack-ansible | 12:44 | |
| odyssey4me | epalper eumel8 resolvconf_enabled only applies if you're using unbound containers in the environment | 12:45 |
| odyssey4me | it will automatically enable/disable based on the presence of unbound_hosts in your environment | 12:46 |
| odyssey4me | admin0 for the interface to use for the public interface (eth0 or eth1) just make sure your default route uses eth1 | 12:48 |
| *** smatzek has quit IRC | 12:48 | |
| odyssey4me | the public interface is determined by the default route, or can be set: https://github.com/openstack/openstack-ansible/blob/master/tests/roles/bootstrap-host/defaults/main.yml#L235-L238 | 12:48 |
| admin0 | odyssey4me: thanks .. got it now | 12:49 |
| *** thorst has joined #openstack-ansible | 13:02 | |
| *** lbragstad has joined #openstack-ansible | 13:07 | |
| *** esberglu has joined #openstack-ansible | 13:09 | |
| *** smatzek has joined #openstack-ansible | 13:10 | |
| cloudnull | mornings | 13:10 |
| neith | cloudnull: mornings | 13:12 |
| jamesdenton | morning cloudnull | 13:13 |
| cloudnull | yo yo yo | 13:13 |
| *** thorst has quit IRC | 13:17 | |
| admin0 | evrardjp, so from the paste and conversation, how do i make use of the host_vars folder | 13:20 |
| admin0 | that seems more clean method . you just list the container/host and say its its on metal or not on metal ? | 13:21 |
| *** mrtenio has joined #openstack-ansible | 13:21 | |
| evrardjp | the paste was complete IIRC | 13:21 |
| evrardjp | cloudnull: hello! | 13:22 |
| admin0 | going thorigh it again: http://eavesdrop.openstack.org/irclogs/%23openstack-ansible/%23openstack-ansible.2017-09-29.log.html :) | 13:22 |
| admin0 | i am not hardcore dev as you guys :) | 13:22 |
| *** smatzek has quit IRC | 13:25 | |
| *** smatzek has joined #openstack-ansible | 13:25 | |
| mrtenio | Good Morning osa team. I have been having some problems when using osa master. I am using it as my development environment for Ironic. I think I am not supposed to do that. What do you advise if one is to use osa as the development environment? My intention is to have the master branches for other services and Ironic. Is there a way of using master for the services, but stable version for osa? Thank you. | 13:27 |
| cloudnull | master is the openstack master (head of queens). | 13:28 |
| *** thorst has joined #openstack-ansible | 13:29 | |
| cloudnull | so its not super stable, but is the dev environment I normally work from | 13:29 |
| cloudnull | if you need something more stable I'd recommend the stable/pike branch | 13:29 |
| *** smatzek has quit IRC | 13:29 | |
| cloudnull | gives a moving target in terms of backports and fixes getting into the various projects, but a more stable environment. | 13:29 |
| *** mrch has quit IRC | 13:31 | |
| evrardjp | mrtenio: that sounds great! an ironic developer using our stuff! | 13:31 |
| evrardjp | mrtenio: why not supposed? Other developers are doing it. It's just that ironic is maybe less tested | 13:32 |
| *** mrch has joined #openstack-ansible | 13:32 | |
| evrardjp | mrtenio: our deploy methods seem to be very much linked to the branches we deploy, so if you are deploying master, you should probably use master | 13:33 |
| evrardjp | but it should be working | 13:33 |
| *** mrch has quit IRC | 13:33 | |
| *** lucasxu has joined #openstack-ansible | 13:34 | |
| *** kylek3h has joined #openstack-ansible | 13:34 | |
| *** mrch has joined #openstack-ansible | 13:34 | |
| mrtenio | evrardjp, osa is life. | 13:35 |
| evrardjp | :D | 13:35 |
| evrardjp | tell me about it! | 13:35 |
| mrtenio | cloudnull and evrardjp, I don't have too much contact with OpenStack Ansible other than using it to deploy OpenStack, so when there is a problem I kind of get stuck for a long time. | 13:35 |
| evrardjp | mmm but there hsouldn't have a problem :D | 13:36 |
| evrardjp | what are you stuck on ? | 13:36 |
| cloudnull | mrtenio: you're always welcome to ping out. lots of folks happy to help. | 13:36 |
| mrtenio | cloudnull, thanks. | 13:36 |
| admin0 | went thorough the post https://evrard.me/group-and-host-variables-overriding-in-openstack-ansible.html (pike) went through all the 3 files .. still do not get it :( | 13:37 |
| evrardjp | admin0: so | 13:37 |
| evrardjp | you are not forced to use pike method, moreover if you have updated | 13:37 |
| evrardjp | admin0: ^ | 13:37 |
| admin0 | i have updated | 13:37 |
| evrardjp | you can keep using your old method | 13:37 |
| mrtenio | evrardjp, I am having a problem with haproxy, it is getting nova_console_port undefined. | 13:37 |
| admin0 | to pike | 13:37 |
| admin0 | if the pike method is the way things in future will go, better to understand it now and maybe write better docs on it | 13:38 |
| evrardjp | mrtenio: do you have a log and a commit sha ? | 13:38 |
| evrardjp | openstack ansible sha* | 13:38 |
| evrardjp | admin0: well that's simple for pike: | 13:38 |
| evrardjp | you create a hostvars folder | 13:39 |
| evrardjp | into your /etc/openstack_deploy/ | 13:39 |
| evrardjp | host_vars* | 13:39 |
| admin0 | created :) | 13:39 |
| evrardjp | then you're using your ansible host as filename inside that | 13:39 |
| evrardjp | that should be all | 13:39 |
| evrardjp | very standard ansible. | 13:39 |
| evrardjp | cloudnull: did you see cinder errors recently? | 13:40 |
| cloudnull | no? | 13:40 |
| evrardjp | or anyone, not just you | 13:40 |
| evrardjp | http://paste.openstack.org/show/622470/ | 13:40 |
| admin0 | my hostname that will be the isci server will be c3.. so i create c3.yml .. but, c3 might serve iscsi on_metal, and maybe logs which is a container | 13:41 |
| evrardjp | what's its ansible hostname | 13:41 |
| evrardjp | ? | 13:41 |
| evrardjp | that's what you should use as filename | 13:41 |
| evrardjp | like any ansible host_vars :p | 13:42 |
| odyssey4me | admin0 using host+vars or group_vars is explained here: http://docs.ansible.com/ansible/latest/intro_inventory.html#splitting-out-host-and-group-specific-data | 13:42 |
| odyssey4me | admin0 for OSA those can be implemented in /etc/openstack_deploy/{host_vars,group_vars} | 13:42 |
| evrardjp | mrtenio: I am concerned it should work out of the box | 13:42 |
| evrardjp | mrtenio: however we recently shuffled around group vars to reduce their scope | 13:43 |
| evrardjp | so maybe there is a group that's using it and has it undefined | 13:43 |
| cloudnull | evrardjp: I've not seen that stacktrace, ever | 13:43 |
| odyssey4me | evrardjp I think admin0 is getting stuck on exactly which var to set to enable/disable on_metal when using host_vars | 13:44 |
| mrtenio | I do some hardcode modifications before running it. When I install Ironic I need to install isntall it baremetal. The error is: http://paste.openstack.org/show/622471/ | 13:44 |
| admin0 | odyssey4me, my cinder.yml has is_metal: false due to me using nfs/ceph .. i need the override for just 1 storage host .. which is what i am trying to do ( and looking for examples ) | 13:44 |
| evrardjp | mrtenio: https://github.com/openstack/openstack-ansible/commit/85501cbf263029cfd5be44b9274113e4e5e4b6f8#diff-0cf0ed34488dfe491c3f98f22bc6e391 | 13:45 |
| odyssey4me | admin0 have you tried just setting is_metal: true using the host_vars for that one host? | 13:45 |
| *** thorst has quit IRC | 13:45 | |
| evrardjp | odyssey4me: the problem is that our usage of is_metal is very much inconsistent, sometimes it's a property sometimes it's a direct variable | 13:46 |
| admin0 | odyssey4me, that brings me to that question .. this host will serve cinder on metal, and possibly rsyslog which will be a container .. .. will setting that on a host level say everything that goes on this host is on metal ? | 13:46 |
| evrardjp | admin0: I'd say, configure with the proper ansible_hostname file, then use is_metal: True there | 13:46 |
| odyssey4me | is_metal is a group_var in master, then the dynamic inventory sets it as a host_var I think - anyway, if it's inconsistent we should change that | 13:47 |
| evrardjp | in the meantime, like I said, you can still use the env.d | 13:47 |
| odyssey4me | admin0 good question - no idea | 13:47 |
| odyssey4me | this was all designed to be uniform, not to pick and choose | 13:47 |
| *** thorst has joined #openstack-ansible | 13:47 | |
| evrardjp | admin0: that's why the name you are using is important | 13:47 |
| *** thorst has quit IRC | 13:48 | |
| odyssey4me | oh yeah, the name has to be the 'container' name right? | 13:48 |
| evrardjp | if you are using in your host_vars/myphysicalhostname.yml it would apply to everything that targets that | 13:48 |
| *** thorst has joined #openstack-ansible | 13:48 | |
| evrardjp | so if you are using myphysicalhostname_cinder_12341632 or something like that you'd only apply to the proper thing | 13:48 |
| mrtenio | evrardjp and odyssey4me I set the property is_metal in the ironic.yml inside env.d. I also disable the nova-compute. | 13:49 |
| evrardjp | mrtenio: that is what's documented | 13:49 |
| evrardjp | and for now, the is_metal is a special case, and should probably be done that way | 13:49 |
| evrardjp | because the inconsistencies of its usage right now | 13:49 |
| evrardjp | mrtenio: I think you have a variable scoping bug | 13:50 |
| *** DanyC has quit IRC | 13:50 | |
| mrtenio | evrardjp, what may I do to check that? | 13:51 |
| evrardjp | I'd add a few variables in group_vars/all/nova.yml : nova_console_port: "{% if nova_console_type == 'spice' %}{{ nova_spice_html5proxy_base_port }}{% elif nova_console_type == 'novnc' %}{{ nova_novncproxy_port }}{% else %}{{ nova_serialconsoleproxy_port }} {% endif %}" | 13:51 |
| evrardjp | well | 13:51 |
| evrardjp | https://github.com/openstack/openstack-ansible/commit/85501cbf263029cfd5be44b9274113e4e5e4b6f8#diff-0cf0ed34488dfe491c3f98f22bc6e391 | 13:52 |
| evrardjp | on the left side for group_vars/all/nova.yml L30 to L34. | 13:52 |
| evrardjp | if that fixes it for you, then we're good for a bug! | 13:52 |
| evrardjp | logan-: maybe we should discuss about this group vars change | 13:53 |
| mrtenio | evrardjp, I will try it, thanks. | 13:55 |
| evrardjp | well, thanks for triaging a bug :) | 13:55 |
| evrardjp | and filing | 13:55 |
| evrardjp | sorry for you to be in that place then, and we should probably document better | 13:55 |
| evrardjp | or test better | 13:55 |
| admin0 | evrardjp, odyssey4me : need help there: https://gist.github.com/anonymous/c58976fa202f0b62b31d034186481bf6 -- what to put in the storage-1 line and where | 13:55 |
| evrardjp | is storage-1 the only dedicated to that? | 13:57 |
| *** drifterza has quit IRC | 13:57 | |
| evrardjp | that different cinder backends I mean | 13:57 |
| admin0 | for now yes, but if resources are not used much, i plan to move rsyslog container there | 13:57 |
| admin0 | the other are qnap, so cinder.yml has is_metal: false so that its globally applied .. | 13:57 |
| mrtenio | evrardjp, no problem. Thank you again! | 13:58 |
| admin0 | i want to use this server as lvm host in metal .. but want to keep the possibilites of moving workload here in future | 13:58 |
| evrardjp | sl | 13:59 |
| evrardjp | so* | 13:59 |
| *** thorst has quit IRC | 13:59 | |
| evrardjp | in container_vars you could apply is_metal: True | 13:59 |
| *** smatzek has joined #openstack-ansible | 14:00 | |
| evrardjp | it would apply to everything that runs on that host, and because it's storage you should be fine | 14:00 |
| *** smatzek has quit IRC | 14:01 | |
| evrardjp | but if you want to adapt in the future you should do the following | 14:01 |
| *** thorst has joined #openstack-ansible | 14:01 | |
| evrardjp | edit your env.d/cinder.yml to have: | 14:01 |
| *** smatzek has joined #openstack-ansible | 14:01 | |
| evrardjp | (see previous paste) | 14:02 |
| evrardjp | this way you can scope to only cinder | 14:02 |
| *** hachi has joined #openstack-ansible | 14:02 | |
| evrardjp | in the future, when is_metal will be uniform, you will be able to use host_vars easily for is_metal. | 14:02 |
| evrardjp | right now host_vars should be used for the rest. | 14:02 |
| *** admin0 has left #openstack-ansible | 14:02 | |
| *** admin0 has joined #openstack-ansible | 14:02 | |
| *** smatzek_ has joined #openstack-ansible | 14:03 | |
| admin0 | so 1. edit cinder.yml and change the line to is_metal: "{{ is_metal_per_host.get(inventory_hostname) | default(False) }}" | 14:03 |
| mrtenio | evrardjp, getting the topic about is_metal, how do I set the var is_metal? | 14:03 |
| admin0 | and then, from where do I override the is_metal: true ? from the container_var ? | 14:03 |
| mrtenio | besides doing it in env.d | 14:04 |
| *** thorst has quit IRC | 14:04 | |
| admin0 | evrardjp, possible to write/blog a mockup sample file ? those help a lot | 14:04 |
| evrardjp | admin0: after step 1, you must define in your user_variables a variable named | 14:04 |
| evrardjp | is_metal_per_host, which is a dict | 14:04 |
| *** thorst has joined #openstack-ansible | 14:04 | |
| *** dxiri_ has joined #openstack-ansible | 14:05 | |
| evrardjp | a dict in which a key could the inventory_hostname of the node you want to change the behavior | 14:05 |
| evrardjp | mrtenio: env.d for now | 14:05 |
| evrardjp | it's the best | 14:05 |
| evrardjp | in the future we can adapt this to be more clear | 14:05 |
| evrardjp | but for now, let's count on that. | 14:05 |
| admin0 | is_metal_per_host = ['c3','c33'] : True | 14:05 |
| evrardjp | and it will probably be for a while, because we take time | 14:06 |
| evrardjp | is_metal_per_host: | 14:06 |
| *** smatzek has quit IRC | 14:06 | |
| evrardjp | c3: | 14:06 |
| evrardjp | c3: True I meant | 14:06 |
| evrardjp | I think that should work | 14:06 |
| *** smatzek_ has quit IRC | 14:07 | |
| admin0 | ok .. .. but this will not allow me to move the rsyslog container here in future right ? | 14:07 |
| evrardjp | mrtenio: generally ppl don't override for one host, they override for many :p | 14:07 |
| evrardjp | admin0: it's unlinked | 14:07 |
| mrtenio | evrardjp, that makes sense. | 14:07 |
| evrardjp | we are not changing anything to rsyslog | 14:07 |
| admin0 | ok | 14:08 |
| evrardjp | admin0: don't forget: dict, not list. | 14:08 |
| evrardjp | {} not [] | 14:08 |
| *** dxiri has quit IRC | 14:08 | |
| evrardjp | else .get( won't work. | 14:08 |
| *** armaan has quit IRC | 14:09 | |
| *** armaan has joined #openstack-ansible | 14:09 | |
| *** smatzek has joined #openstack-ansible | 14:11 | |
| admin0 | done .. | 14:15 |
| admin0 | is there a way to run ansible to validae this ? | 14:15 |
| *** smatzek has quit IRC | 14:16 | |
| *** smatzek_ has joined #openstack-ansible | 14:16 | |
| *** smatzek_ has quit IRC | 14:16 | |
| *** mrch has quit IRC | 14:19 | |
| *** mrch has joined #openstack-ansible | 14:20 | |
| *** dxiri_ has quit IRC | 14:23 | |
| *** dxiri has joined #openstack-ansible | 14:25 | |
| *** smatzek has joined #openstack-ansible | 14:29 | |
| *** smatzek_ has joined #openstack-ansible | 14:33 | |
| *** smatzek has quit IRC | 14:34 | |
| evrardjp | admin0: write your own playbook that checks the is_metal value | 14:37 |
| evrardjp | backup your inventory first | 14:37 |
| evrardjp | and everything | 14:37 |
| *** smatzek_ has quit IRC | 14:37 | |
| *** smatzek has joined #openstack-ansible | 14:37 | |
| admin0 | err... i am already running the playbooks :D | 14:37 |
| evrardjp | k | 14:38 |
| admin0 | nothing failed, but it did not even touched c3 :) | 14:38 |
| admin0 | play recap: only infra hosts | 14:38 |
| admin0 | i will try to look into more details here | 14:38 |
| admin0 | with limit, i get no hosts matched | 14:39 |
| *** smatzek has quit IRC | 14:42 | |
| *** chyka has joined #openstack-ansible | 14:44 | |
| *** smatzek has joined #openstack-ansible | 14:44 | |
| *** smatzek_ has joined #openstack-ansible | 14:46 | |
| idlemind | does openstack-ansible have a preset configuration file that is similar in behavior to devstack already? seems like the "test" environment one still wants multiple nodes. | 14:46 |
| *** armaan has quit IRC | 14:48 | |
| *** armaan has joined #openstack-ansible | 14:48 | |
| *** smatzek has quit IRC | 14:49 | |
| eumel8 | evrardjp, cloudnull: you want to repair the legacy zuul-v3 jobs or switch immediately to .zuul.yaml in osa repo? | 14:50 |
| *** smatzek has joined #openstack-ansible | 14:51 | |
| *** smatzek_ has quit IRC | 14:51 | |
| *** dxiri_ has joined #openstack-ansible | 14:57 | |
| *** jvidal has quit IRC | 14:57 | |
| *** chyka has quit IRC | 14:57 | |
| *** chyka has joined #openstack-ansible | 14:58 | |
| *** dxiri has quit IRC | 15:01 | |
| *** captaindave has joined #openstack-ansible | 15:01 | |
| *** mrch has quit IRC | 15:02 | |
| *** Oku_OS is now known as Oku_OS-away | 15:05 | |
| evrardjp | eumel8: that would be great if we switch in repo just after the release. | 15:09 |
| evrardjp | odyssey4me: and logan- also worked on this | 15:10 |
| evrardjp | odyssey4me: did you see that bug: https://bugs.launchpad.net/openstack-ansible/+bug/1716663 ? | 15:10 |
| openstack | Launchpad bug 1716663 in openstack-ansible "repo_build : Initialize local facts fails when upgrading" [Undecided,New] | 15:10 |
| cloudnull | evrardjp: eumel8: we have https://review.openstack.org/#/c/508509 waiting to be merged. | 15:11 |
| cloudnull | once that goes in I think our normal gates will work again. | 15:11 |
| cloudnull | but It be great to get that converted to the new setup. | 15:12 |
| odyssey4me | evrardjp looks to me lie the jmespath solution would be cleanest... although I'd cage "ansible_local | default({})" in () to make sure the filter application is dine right | 15:13 |
| odyssey4me | *done | 15:13 |
| evrardjp | yeah latest have to be tested | 15:13 |
| evrardjp | but it was just to know if you have seen or heard that too | 15:13 |
| evrardjp | I had that with my on metal test | 15:13 |
| odyssey4me | it's very weird though because ansible_local should exist at that point - a fact is dropped in the lxc-containers-create playbook to register the container variant used | 15:14 |
| evrardjp | I'd rather not do the jmespath with stable branches | 15:14 |
| evrardjp | I don't run that :p | 15:14 |
| evrardjp | that's why ;) | 15:14 |
| odyssey4me | I've seen that discussed in the channel - the last person had not updated their roles properly after changing to the new version ;) | 15:14 |
| evrardjp | that could be a cause too | 15:14 |
| evrardjp | I'll do a dual implementation I think | 15:15 |
| odyssey4me | but yeah, it'd be nice to make those conditionals easier to read | 15:15 |
| evrardjp | one with jmespath in master, the simple in stable/ | 15:15 |
| evrardjp | well the jmespath is kinda ugly too | 15:16 |
| evrardjp | because it will take the value | 15:16 |
| evrardjp | but let me evaluate | 15:16 |
| odyssey4me | IIRC nolan did a review to add a filter which worked too - but I'd rather do jmespath than add more code we carry | 15:16 |
| admin0 | i see the pods example in pike ..and it has 3 infra nodes in 3 pods .. so each infra node now has only data of its own set of nodes only ? its a single point of failure if that node goes down ? | 15:18 |
| admin0 | or that example is very generic to just give example | 15:18 |
| *** TomMc has quit IRC | 15:18 | |
| eumel8 | cloudnull: thx. Just to hear how you think about the zuul migration. | 15:19 |
| *** pcaruana has quit IRC | 15:27 | |
| *** cshen has quit IRC | 15:31 | |
| *** galstrom_zzz is now known as galstrom | 15:31 | |
| *** dave-mccowan has quit IRC | 15:32 | |
| *** dave-mcc_ has joined #openstack-ansible | 15:32 | |
| *** hachi has quit IRC | 15:35 | |
| *** epalper has quit IRC | 15:38 | |
| logan- | o/ | 15:40 |
| *** acormier has joined #openstack-ansible | 15:41 | |
| *** armaan has quit IRC | 15:42 | |
| *** armaan has joined #openstack-ansible | 15:43 | |
| idlemind | is openstack-ansible able to configure ceph along with cinder or does ceph have to built manually w/custom playbooks outside of starting openstack-ansible? | 15:44 |
| admin0 | logan-, \o | 15:45 |
| admin0 | idlemind, work is ongoing/done for having an integrated build | 15:46 |
| idlemind | thx! k looks like i have a day full of playing ahead of me | 15:47 |
| *** thorst has quit IRC | 15:49 | |
| *** weezS has joined #openstack-ansible | 15:52 | |
| *** galstrom is now known as galstrom_zzz | 15:53 | |
| *** gouthamr has joined #openstack-ansible | 15:54 | |
| *** rpittau_ has joined #openstack-ansible | 15:58 | |
| *** rpittau has quit IRC | 16:01 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-repo_build master: Fix typo on conditional https://review.openstack.org/508928 | 16:07 |
| idlemind | am i able to run everything on br-mgmt? (not even create the other networks) for lab spin-ups | 16:08 |
| *** thorst has joined #openstack-ansible | 16:10 | |
| *** armaan has quit IRC | 16:10 | |
| *** gmonteiro has quit IRC | 16:12 | |
| *** thorst has quit IRC | 16:13 | |
| admin0 | idlemind, you can use the aio | 16:14 |
| admin0 | else other networks are listed as a requirements | 16:14 |
| idlemind | aio? | 16:14 |
| idlemind | all-in-one? | 16:14 |
| admin0 | all-in-one | 16:14 |
| dmsimard | evrardjp, cloudnull, odyssey4me: fyi boxrick discovered an issue with ara in 2.4, it'll be fixed by https://github.com/ansible/ansible/pull/31200/commits/865d57aaf2fe47bc95d7becaac3a0cb182f71aa8 | 16:14 |
| *** smatzek has quit IRC | 16:14 | |
| dmsimard | I'll work on a patch to work around that issue until it's backported to 2.4 | 16:14 |
| dmsimard | it causes ara to use default configuration parameters all the time | 16:15 |
| idlemind | admin0 what is the sample playbook for that one / command to kick it off | 16:15 |
| dmsimard | if using an ansible.cfg file. | 16:15 |
| evrardjp | dmsimard: oh | 16:15 |
| *** smatzek has joined #openstack-ansible | 16:16 | |
| admin0 | idlemind, https://docs.openstack.org/openstack-ansible/pike/contributor/quickstart-aio.html#building-an-aio | 16:16 |
| *** smatzek has quit IRC | 16:16 | |
| evrardjp | dmsimard: you think this will merge? | 16:16 |
| *** chhavi has joined #openstack-ansible | 16:16 | |
| evrardjp | well thanks for the patch | 16:17 |
| dmsimard | evrardjp: I discussed the issue with bcoca so it should hopefully merge yes | 16:18 |
| dmsimard | I mentioned it because I think you guys are using get_config too | 16:19 |
| *** smatzek has joined #openstack-ansible | 16:19 | |
| evrardjp | let me double check but I don't think we do | 16:23 |
| evrardjp | darn you're right | 16:24 |
| evrardjp | good catch! | 16:24 |
| idlemind | thx admin0 reading it now | 16:24 |
| logan- | https://review.openstack.org/#/c/508931/ cc evrardjp odyssey4me cloudnull | 16:25 |
| logan- | our regular aio check job is non-voting atm https://review.openstack.org/#/c/508509/ heh | 16:26 |
| odyssey4me | heh, ouch | 16:28 |
| *** thorst has joined #openstack-ansible | 16:29 | |
| *** thorst has quit IRC | 16:30 | |
| *** thorst has joined #openstack-ansible | 16:33 | |
| *** thorst has quit IRC | 16:33 | |
| evrardjp | odyssey4me: vote vote vote :) | 16:34 |
| evrardjp | thanks! | 16:34 |
| *** thorst has joined #openstack-ansible | 16:40 | |
| *** thorst has quit IRC | 16:44 | |
| *** dave-mcc_ has quit IRC | 16:46 | |
| *** woodard has quit IRC | 16:47 | |
| *** drifterza has joined #openstack-ansible | 16:50 | |
| *** weezS has quit IRC | 16:56 | |
| *** woodard has joined #openstack-ansible | 16:56 | |
| *** thorst has joined #openstack-ansible | 16:57 | |
| *** idlemind has quit IRC | 16:59 | |
| mrtenio | evrardjp, I reported a bug, I don't know if I did the right way https://bugs.launchpad.net/openstack-ansible/+bug/1720847 | 17:00 |
| openstack | Launchpad bug 1720847 in openstack-ansible "OpenStack Ansible fails on haproxy setup with nova_console_port undefined" [Undecided,New] | 17:00 |
| *** thorst has quit IRC | 17:01 | |
| *** gkadam has quit IRC | 17:01 | |
| mrtenio | The test you asked me to do solved the problem with nova_console_port undefined, but other errors happened later. | 17:01 |
| evrardjp | logan-: ^ the issue appearing there is due to the group var migration | 17:02 |
| evrardjp | what do you think of reverting? Or we go fixing one per one... | 17:02 |
| evrardjp | mrtenio: could you tell more about the next issue? | 17:03 |
| logan- | interesting | 17:03 |
| evrardjp | sorry for using you as a lab rat, if you allow me the expression (I don't mean to be rude or anything, just trying to be funny with my limited vocabulary!) | 17:03 |
| logan- | i wonder why that wouldnt be showing up in the gate | 17:04 |
| evrardjp | yeah | 17:04 |
| evrardjp | maybe it appears in roles? | 17:05 |
| mrtenio | evrardjp, it is no offense. I am happy to help (and be helped :P) | 17:05 |
| evrardjp | :D | 17:06 |
| *** acormier has quit IRC | 17:07 | |
| *** armaan has joined #openstack-ansible | 17:08 | |
| *** acormier has joined #openstack-ansible | 17:08 | |
| mrtenio | The other errors were related to nova as well. http://paste.openstack.org/show/622486/ | 17:09 |
| evrardjp | that is more worrying | 17:10 |
| mrtenio | I can't say too much about the errors, sorry. | 17:11 |
| *** chyka_ has joined #openstack-ansible | 17:12 | |
| *** chyka has quit IRC | 17:14 | |
| *** chyka has joined #openstack-ansible | 17:16 | |
| *** chyka_ has quit IRC | 17:18 | |
| *** gouthamr has quit IRC | 17:21 | |
| *** chhavi has quit IRC | 17:21 | |
| *** chyka_ has joined #openstack-ansible | 17:21 | |
| *** chyka has quit IRC | 17:22 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Load nova_console_port from a nova_console inventory item https://review.openstack.org/508949 | 17:22 |
| logan- | mrtenio could you test ^? | 17:22 |
| mrtenio | Yes, will take some time. I will tell when it finishes. | 17:23 |
| *** thorst has joined #openstack-ansible | 17:23 | |
| *** idlemind has joined #openstack-ansible | 17:23 | |
| *** chyka has joined #openstack-ansible | 17:25 | |
| *** chyka_ has quit IRC | 17:29 | |
| *** mbuil has quit IRC | 17:33 | |
| *** dxiri has joined #openstack-ansible | 17:33 | |
| *** dxiri_ has quit IRC | 17:36 | |
| *** weezS has joined #openstack-ansible | 17:39 | |
| *** shardy has quit IRC | 17:52 | |
| *** electrofelix has quit IRC | 17:56 | |
| *** poopcat has joined #openstack-ansible | 17:56 | |
| *** gouthamr has joined #openstack-ansible | 17:57 | |
| *** thorst_ has joined #openstack-ansible | 18:01 | |
| *** thorst has quit IRC | 18:03 | |
| captaindave | Hey all. I'm trying to set up an openstack lab on a sufficiently beefy single physical box where the hosts are simple kvm centos vms. I'm passing through the networks from the NICs without touching them, save to put them on the necessary bridges for the VMs. AFAICT, I have all the requisite connectivity to the VMs (as confirmed with tcpdump). I've run setup-hosts.yml with 0 errors, but I'm not | 18:04 |
| captaindave | getting any values for ansible_host on the containers (I am, however, on the hosts). Here's the output from 'ip addr' on all the hosts: http://sprunge.us/eYKb and here's my openstack_user_config.yml (the prod example with different addressing and commented out host entries): http://sprunge.us/DdeP Anyone have any input on what I might be doing wrong? | 18:04 |
| admin0 | captaindave: have you created the necessary bridges on the vm itself ? | 18:06 |
| captaindave | Yes. | 18:07 |
| captaindave | The hosts in http://sprunge.us/eYKb are the vms. | 18:07 |
| admin0 | i have it documented here: http://www.openstackfaq.com/openstack-dev-server-setup-centos/ | 18:07 |
| admin0 | checking! | 18:07 |
| admin0 | how does your config file looks like ? | 18:07 |
| captaindave | the openstack_user_config.yml? | 18:08 |
| admin0 | yes | 18:08 |
| captaindave | http://sprunge.us/DdeP | 18:08 |
| admin0 | :) why 2lines for used_ips ? 240.1, 240.63 can go in 1 line :) | 18:09 |
| admin0 | ignore me .. | 18:09 |
| captaindave | Different reasons for exclusion. | 18:09 |
| *** hachi has joined #openstack-ansible | 18:10 | |
| captaindave | One's for preexisting infrastructure and the other's for openstack's requirements in particular. | 18:10 |
| captaindave | I'll eventually be modifying the second one for the actual (non-proof of concept) deployment. | 18:11 |
| captaindave | The repo container builds itself fine, but the haproxy install craps out when it doesn't have all the ansible_host values it needs. | 18:12 |
| admin0 | are there the vars in user_variables as well for haproxy ? | 18:13 |
| captaindave | aproxy_keepalived_external_vip_cidr: "{{external_lb_vip_address}}/32" | 18:13 |
| captaindave | haproxy_keepalived_internal_vip_cidr: "{{internal_lb_vip_address}}/32" | 18:13 |
| captaindave | haproxy_keepalived_external_interface: eth0 | 18:13 |
| captaindave | haproxy_keepalived_internal_interface: br-mgmt | 18:13 |
| captaindave | there's an h on that first one; don't worry. | 18:13 |
| admin0 | maybe paste output from openstack-ansible haproxy-install.yml run | 18:14 |
| captaindave | Sure. Give me a few. | 18:14 |
| captaindave | I'll have to rebuild the repo first. | 18:15 |
| captaindave | and unbound | 18:15 |
| admin0 | no need | 18:15 |
| captaindave | Oh, OK. | 18:15 |
| *** thorst_ has quit IRC | 18:15 | |
| admin0 | in fact, i run haproxy first before even running setup-infra or setup-openstack | 18:15 |
| captaindave | Interesting! | 18:16 |
| captaindave | How verbose do you want it? | 18:16 |
| admin0 | becuase if haproxy has errors, it fails on repo build .. so i ensure haproxy and the vips are there before i go to setup-infra | 18:16 |
| admin0 | some | 18:16 |
| captaindave | two v's it is. | 18:16 |
| captaindave | http://sprunge.us/hKRX | 18:19 |
| *** chas has quit IRC | 18:23 | |
| *** jbadiapa_ has joined #openstack-ansible | 18:23 | |
| admin0 | i have never used an IP on an address field.. always used address …. and always used more than 1 controllers .. so i am not sure why its borking out .. “set _ = entry.append(\"backu | 18:25 |
| admin0 | p\") %}\n {{ entry | join(' ') }}\n{% endfor %}\n): unsupported operand type(s) for +: 'NoneType' and 'str’”}” — maybe it needs a backup infra node as well // stay here and someone else can comment on that | 18:25 |
| *** jbadiapa has quit IRC | 18:25 | |
| admin0 | but to try out, also specify infra1 , some mockup address and see if that is the case | 18:26 |
| captaindave | an IP on an address field? | 18:27 |
| captaindave | you mean external_lb_vip_address? | 18:27 |
| admin0 | internal_lb_vip_address: api-internal.mydomain.cloud type | 18:27 |
| admin0 | this may not be the case | 18:27 |
| admin0 | remove the # from balancer1 and rerun it once more | 18:28 |
| captaindave | From both infra1 and balancer1? | 18:28 |
| admin0 | only the haproxy-install.yml .. see how far it goes and if it borks at this same error | 18:28 |
| admin0 | no no | 18:28 |
| admin0 | just for haproxy_hosts | 18:28 |
| captaindave | OK. | 18:28 |
| captaindave | I need to spin up that VM again. Give me a few. | 18:29 |
| admin0 | well, no need actually .. it might fail saying that vm is dead | 18:29 |
| admin0 | but we are trying to check if the config needs a backup host | 18:29 |
| captaindave | Gotcha. | 18:29 |
| admin0 | if its the backup host thats needed, then the config will pass for balancer0 | 18:29 |
| *** drifterza has quit IRC | 18:33 | |
| *** goldenfri has joined #openstack-ansible | 18:34 | |
| captaindave | Looks like the same thing: http://sprunge.us/MTab | 18:38 |
| captaindave | Plus an error about not being able to get facts from balancer1 | 18:39 |
| admin0 | i do not know ..one final thing i can do is to bring up balancer1, change ips to hostname ( some mockups ) and see if that fixes | 18:40 |
| admin0 | else, above my head :) | 18:40 |
| admin0 | i brb | 18:41 |
| captaindave | This will take a bit. | 18:45 |
| *** thorst has joined #openstack-ansible | 18:46 | |
| *** thorst has quit IRC | 18:46 | |
| mrtenio | logan-, the scripts finished, the patch https://review.openstack.org/508949 solved the problem with nova_console_port undefined. But it stopped with the errors I showed here http://paste.openstack.org/show/622486/ | 18:46 |
| *** thorst has joined #openstack-ansible | 18:48 | |
| *** dave-mccowan has joined #openstack-ansible | 18:59 | |
| *** idlemind has quit IRC | 19:04 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: [Docs] Clarify role development cycle https://review.openstack.org/504279 | 19:05 |
| *** idlemind has joined #openstack-ansible | 19:05 | |
| *** dxiri has quit IRC | 19:06 | |
| *** dxiri has joined #openstack-ansible | 19:06 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: [DNM] Build everything on metal https://review.openstack.org/504224 | 19:07 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: [Docs] Add the bump of the upstream repo https://review.openstack.org/502502 | 19:10 |
| *** smatzek has quit IRC | 19:12 | |
| *** smatzek has joined #openstack-ansible | 19:13 | |
| *** smatzek has quit IRC | 19:17 | |
| *** smatzek has joined #openstack-ansible | 19:21 | |
| *** smatzek has quit IRC | 19:25 | |
| *** smatzek has joined #openstack-ansible | 19:27 | |
| *** smatzek has quit IRC | 19:27 | |
| *** smatzek has joined #openstack-ansible | 19:28 | |
| *** hachi has quit IRC | 19:29 | |
| *** hachi has joined #openstack-ansible | 19:29 | |
| *** chyka has quit IRC | 19:35 | |
| *** chyka has joined #openstack-ansible | 19:35 | |
| *** thorst has quit IRC | 19:41 | |
| *** hachi has quit IRC | 19:46 | |
| *** hachi has joined #openstack-ansible | 19:47 | |
| tasker | during a run of 'os-keystone-install', at the task "gather software version list", the underlying ansible is throwing a "keyerror: ansible_local" at line 396 of "ansible/plugins/filter/core.py". I'm not seeing any active bugs on launchpad. has anyone seen this before? | 19:54 |
| *** thorst has joined #openstack-ansible | 19:56 | |
| tasker | this is ansible 14.2.8 | 19:56 |
| tasker | .. OSA 14.2.8 | 19:56 |
| * tasker is exhuasted | 19:56 | |
| tasker | previously I have been using a "--limit". going to try a run with no limit. | 19:57 |
| *** thorst has quit IRC | 20:00 | |
| *** gouthamr has quit IRC | 20:06 | |
| cloudnull | ansible_local is fact related | 20:06 |
| cloudnull | its possible the limit is omiting some fact generation which is causing an issue | 20:07 |
| tasker | sweet | 20:07 |
| tasker | that seems reasonable with the recent run having not blown up yet. | 20:07 |
| tasker | no limit, playbook completes successfully. | 20:10 |
| tasker | is this a design limitation to ansible or openstack-ansible? | 20:10 |
| cloudnull | its an osa issue. | 20:12 |
| cloudnull | retry limits have been broken for a while, we normally just turn them off. | 20:12 |
| tasker | . ( | 20:12 |
| cloudnull | and then use tags | 20:12 |
| tasker | i was hoping to take advantage of them to upgrade my control plane in thirds. | 20:13 |
| cloudnull | you can limit using groups | 20:13 |
| tasker | tell me more of these "groups" | 20:13 |
| cloudnull | something like keystone_all[0] | 20:13 |
| cloudnull | which would do the first node from that group | 20:13 |
| *** dave-mcc_ has joined #openstack-ansible | 20:13 | |
| tasker | is this something I can apply on the command-line? | 20:14 |
| cloudnull | yes. | 20:14 |
| cloudnull | openstack-ansible os-keystone-install.yml --limit 'keystone_all[0]' | 20:14 |
| tasker | oh! | 20:14 |
| cloudnull | if you wanted to do one, then the others you can do something like `openstack-ansible os-keystone-install.yml --limit 'keystone_all[0]',keystone_all[0:]` | 20:14 |
| cloudnull | opps. | 20:15 |
| cloudnull | `openstack-ansible os-keystone-install.yml --limit 'keystone_all[0],keystone_all[0:]'` | 20:15 |
| tasker | so basic python list indexing? | 20:15 |
| cloudnull | yes. | 20:15 |
| tasker | neat. | 20:15 |
| cloudnull | all of our services follow that pattern | 20:15 |
| cloudnull | ${service}_all | 20:15 |
| tasker | roger. | 20:16 |
| cloudnull | if you really want to hook into the groups you can pick through the inventory json | 20:16 |
| cloudnull | which has a complete list of all groups found in osa | 20:16 |
| *** dave-mccowan has quit IRC | 20:16 | |
| *** dave-mcc_ is now known as dave-mccowan | 20:17 | |
| *** dxiri has quit IRC | 20:19 | |
| *** dxiri has joined #openstack-ansible | 20:20 | |
| *** smatzek has quit IRC | 20:22 | |
| *** chas has joined #openstack-ansible | 20:23 | |
| tasker | that worked, but the ordering of containers will be different between services. could the inventory be edited by hand to force an ordering of the containers? | 20:25 |
| *** gouthamr has joined #openstack-ansible | 20:25 | |
| cloudnull | maybe? | 20:26 |
| cloudnull | its regenerated on the run which I think is sorted. | 20:26 |
| cloudnull | you can try it? | 20:26 |
| tasker | yeah. looks like it's maintaining whatever order I put in the inventory. | 20:27 |
| *** chas has quit IRC | 20:28 | |
| tasker | hmm. looks like all of my services were sorted in 2, 3, 1 order. | 20:28 |
| tasker | so, as long as that order is actually maintained, I should okay. | 20:30 |
| admin0 | cloudnull: how does pods work ? .. when an ocata install is upgraded to pike, is it automatically on pod/cell-1 .. and later we can define more pods ? | 20:31 |
| *** thorst has joined #openstack-ansible | 20:37 | |
| cloudnull | yes you can define more cells as needed. | 20:43 |
| *** esberglu has quit IRC | 20:43 | |
| cloudnull | though IMHO for the vast majority of deployments a single cell should be all that's required. | 20:43 |
| cloudnull | tasker: I want to say that wont change, unless you add more hosts to the environment. | 20:44 |
| admin0 | i have one tenant that creates like 10 vms per week, and another that creates 100 per hour and then destroyes in the next 3-4 hours ( training ) .. so the active one sometimes creates problem fo the other one as rabbit/neutron is on queue | 20:45 |
| admin0 | so was thinking if cells can solve it | 20:45 |
| tasker | cloudnull: I just discovered that using service_all[i] won't work when you get to complex groups that are made of sub-groups, such as nova. | 20:45 |
| tasker | in the case of nova, the full size of the list is 18. | 20:46 |
| cloudnull | yes you'd have to chain the groups together. like 'nova_all[0]:neutron_all[0]' | 20:46 |
| tasker | ahh | 20:46 |
| tasker | ok. that might work. | 20:46 |
| *** esberglu has joined #openstack-ansible | 20:55 | |
| *** smatzek has joined #openstack-ansible | 20:57 | |
| *** chas has joined #openstack-ansible | 20:58 | |
| *** smatzek has quit IRC | 20:58 | |
| *** smatzek has joined #openstack-ansible | 20:59 | |
| *** smatzek has quit IRC | 20:59 | |
| *** pbandark1 has joined #openstack-ansible | 21:00 | |
| *** dave-mccowan has quit IRC | 21:01 | |
| *** pbandark has quit IRC | 21:01 | |
| *** pbandark1 is now known as pbandark | 21:01 | |
| *** hachi has quit IRC | 21:04 | |
| *** mbuil has joined #openstack-ansible | 21:11 | |
| captaindave | admin0: I think I've found the problem. BBL. | 21:12 |
| *** captaindave has quit IRC | 21:12 | |
| *** mbuil has quit IRC | 21:40 | |
| admin0 | cloudnull: i have these variables for az in my config: https://gist.github.com/a1git/fa8abc77333cc91d58b69e46fbb758e3 .. but still, when i try to launch an instance, horizon shows nova as default .. what am i missing there ? | 21:41 |
| admin0 | hmm.. region is not same as az | 21:41 |
| admin0 | how to change default AZ from nova to somethign else ? | 21:43 |
| admin0 | nova_default_schedule_zone and cinder_storage_availability_zone ?? | 21:44 |
| *** askb has joined #openstack-ansible | 21:44 | |
| *** lucasxu has quit IRC | 21:47 | |
| *** chyka_ has joined #openstack-ansible | 21:48 | |
| *** chyka has quit IRC | 21:52 | |
| *** masber has joined #openstack-ansible | 22:07 | |
| *** thorst has quit IRC | 22:15 | |
| *** chas has quit IRC | 22:16 | |
| *** thorst has joined #openstack-ansible | 22:18 | |
| *** chas has joined #openstack-ansible | 22:18 | |
| *** thorst has quit IRC | 22:22 | |
| *** pmannidi has joined #openstack-ansible | 22:23 | |
| *** chas has quit IRC | 22:24 | |
| *** armaan_ has joined #openstack-ansible | 22:26 | |
| *** armaan_ has quit IRC | 22:27 | |
| *** kylek3h has quit IRC | 22:28 | |
| *** armaan has quit IRC | 22:29 | |
| *** acormier has quit IRC | 22:29 | |
| *** acormier has joined #openstack-ansible | 22:30 | |
| *** thorst has joined #openstack-ansible | 22:31 | |
| *** acormier has quit IRC | 22:34 | |
| *** thorst has quit IRC | 22:35 | |
| *** lbragstad has quit IRC | 22:36 | |
| *** gouthamr has quit IRC | 22:38 | |
| *** pbandark has quit IRC | 22:42 | |
| *** chyka_ has quit IRC | 23:02 | |
| *** chyka has joined #openstack-ansible | 23:15 | |
| *** chhavi has joined #openstack-ansible | 23:17 | |
| *** esberglu has quit IRC | 23:21 | |
| *** chhavi has quit IRC | 23:22 | |
| *** acormier has joined #openstack-ansible | 23:26 | |
| *** acormier has quit IRC | 23:30 | |
| *** weezS has quit IRC | 23:32 | |
| *** germs has joined #openstack-ansible | 23:57 | |
| *** germs has quit IRC | 23:58 | |
| *** dxiri_ has joined #openstack-ansible | 23:59 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!