| *** jamesdenton has joined #openstack-ansible | 00:00 | |
| *** thorst has joined #openstack-ansible | 00:02 | |
| *** thorst has quit IRC | 00:07 | |
| *** agrebennikov has joined #openstack-ansible | 00:09 | |
| *** ricardoas has quit IRC | 00:09 | |
| *** yifei has joined #openstack-ansible | 00:09 | |
| *** acormier has joined #openstack-ansible | 00:17 | |
| *** acormier has quit IRC | 00:17 | |
| *** acormier has joined #openstack-ansible | 00:18 | |
| *** askb has joined #openstack-ansible | 00:19 | |
| *** acormier has quit IRC | 00:22 | |
| *** acormier has joined #openstack-ansible | 00:23 | |
| *** jamesdenton has quit IRC | 00:24 | |
| *** acormier has quit IRC | 00:27 | |
| *** jamesdenton has joined #openstack-ansible | 00:27 | |
| *** acormier has joined #openstack-ansible | 00:29 | |
| *** acormier has quit IRC | 00:31 | |
| *** markvoelker has quit IRC | 00:35 | |
| *** jamesdenton has quit IRC | 00:37 | |
| *** dxiri has quit IRC | 00:41 | |
| *** acormier has joined #openstack-ansible | 00:41 | |
| *** acormier has quit IRC | 00:45 | |
| *** acormier has joined #openstack-ansible | 00:46 | |
| *** markvoelker has joined #openstack-ansible | 00:47 | |
| *** n3v3rm0r3r has quit IRC | 00:56 | |
| *** chyka has joined #openstack-ansible | 01:03 | |
| *** chyka has quit IRC | 01:04 | |
| *** gouthamr has quit IRC | 01:05 | |
| *** chyka has joined #openstack-ansible | 01:07 | |
| *** kristian__ has joined #openstack-ansible | 01:26 | |
| *** cathrich_ has joined #openstack-ansible | 01:28 | |
| *** cathrichardson has quit IRC | 01:28 | |
| *** n3v3rm0r3r has joined #openstack-ansible | 01:29 | |
| *** kristian__ has quit IRC | 01:31 | |
| *** galstrom_zzz is now known as galstrom | 01:43 | |
| *** dxiri has joined #openstack-ansible | 01:47 | |
| *** dxiri has quit IRC | 01:51 | |
| *** acormier_ has joined #openstack-ansible | 01:51 | |
| *** acormier has quit IRC | 01:55 | |
| *** n3v3rm0r3r has quit IRC | 01:58 | |
| *** dxiri has joined #openstack-ansible | 02:05 | |
| *** thorst has joined #openstack-ansible | 02:08 | |
| *** dxiri has quit IRC | 02:10 | |
| *** thorst has quit IRC | 02:13 | |
| *** acormier_ has quit IRC | 02:25 | |
| *** n3v3rm0r3r has joined #openstack-ansible | 02:26 | |
| *** jwitko has quit IRC | 02:30 | |
| *** esberglu has joined #openstack-ansible | 03:00 | |
| *** esberglu has quit IRC | 03:00 | |
| *** dxiri has joined #openstack-ansible | 03:02 | |
| *** chhavi has joined #openstack-ansible | 03:02 | |
| *** dxiri has quit IRC | 03:06 | |
| *** schwicht has quit IRC | 03:14 | |
| *** galstrom is now known as galstrom_zzz | 03:26 | |
| *** cpuga has joined #openstack-ansible | 03:26 | |
| *** cpuga has quit IRC | 03:30 | |
| *** udesale has joined #openstack-ansible | 03:35 | |
| *** cpuga has joined #openstack-ansible | 03:44 | |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible master: Idempotent nova db privilege grants https://review.openstack.org/483176 | 03:52 |
|---|---|---|
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible master: Idempotent nova db privilege grants https://review.openstack.org/483176 | 03:53 |
| *** esberglu has joined #openstack-ansible | 03:53 | |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible master: Idempotent nova db privilege grants https://review.openstack.org/483176 | 03:55 |
| *** esberglu has quit IRC | 03:58 | |
| *** gkadam has joined #openstack-ansible | 03:59 | |
| *** albertcard1 has quit IRC | 04:28 | |
| *** agrebennikov has quit IRC | 04:29 | |
| *** hybridpollo has quit IRC | 04:46 | |
| *** ivve has joined #openstack-ansible | 05:25 | |
| *** prometheanfire has quit IRC | 05:25 | |
| *** gouthamr has joined #openstack-ansible | 05:34 | |
| *** marst has joined #openstack-ansible | 05:36 | |
| *** tobberydberg has joined #openstack-ansible | 05:41 | |
| *** kristian__ has joined #openstack-ansible | 05:43 | |
| *** kristian__ has quit IRC | 05:47 | |
| *** thorst has joined #openstack-ansible | 05:57 | |
| *** thorst has quit IRC | 06:03 | |
| *** dxiri has joined #openstack-ansible | 06:05 | |
| *** cshen_ has joined #openstack-ansible | 06:05 | |
| *** yifei has quit IRC | 06:09 | |
| *** yifei has joined #openstack-ansible | 06:09 | |
| *** dxiri has quit IRC | 06:09 | |
| *** Oku_OS-away is now known as Oku_OS | 06:10 | |
| *** dxiri has joined #openstack-ansible | 06:33 | |
| *** esberglu has joined #openstack-ansible | 06:37 | |
| *** dxiri has quit IRC | 06:38 | |
| *** esberglu has quit IRC | 06:38 | |
| *** esberglu has joined #openstack-ansible | 06:38 | |
| *** esberglu has quit IRC | 06:38 | |
| *** chyka has quit IRC | 06:44 | |
| *** pcaruana has joined #openstack-ansible | 06:52 | |
| *** kristian__ has joined #openstack-ansible | 06:53 | |
| *** cshen_ has quit IRC | 06:55 | |
| *** electrofelix has joined #openstack-ansible | 06:57 | |
| *** armaan has joined #openstack-ansible | 06:57 | |
| *** coolkil has joined #openstack-ansible | 07:02 | |
| *** armaan_ has joined #openstack-ansible | 07:06 | |
| *** armaan has quit IRC | 07:06 | |
| *** mbuil has joined #openstack-ansible | 07:09 | |
| *** armaan_ has quit IRC | 07:11 | |
| *** armaan has joined #openstack-ansible | 07:11 | |
| *** armaan_ has joined #openstack-ansible | 07:14 | |
| *** armaan has quit IRC | 07:15 | |
| *** dxiri has joined #openstack-ansible | 07:17 | |
| *** gouthamr has quit IRC | 07:22 | |
| *** dxiri has quit IRC | 07:22 | |
| *** armaan_ has quit IRC | 07:27 | |
| *** armaan has joined #openstack-ansible | 07:27 | |
| *** jvidal has joined #openstack-ansible | 07:27 | |
| *** charcol has quit IRC | 07:30 | |
| *** sufyan68 has joined #openstack-ansible | 07:31 | |
| *** andreas_s has joined #openstack-ansible | 07:37 | |
| andreas_s | coolkil, around? I'm one of the colleagues of markus_z who is working on openstack ansible on s390x | 07:38 |
| *** shardy has joined #openstack-ansible | 07:52 | |
| *** kristian__ has quit IRC | 07:54 | |
| *** kristian__ has joined #openstack-ansible | 07:55 | |
| *** thorst has joined #openstack-ansible | 07:59 | |
| *** tobberyd_ has joined #openstack-ansible | 07:59 | |
| coolkil | ey andreas yea im around | 08:01 |
| *** tobberydberg has quit IRC | 08:03 | |
| *** thorst has quit IRC | 08:03 | |
| *** askb has quit IRC | 08:04 | |
| *** dxiri has joined #openstack-ansible | 08:06 | |
| *** markvoelker has quit IRC | 08:08 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_swift master: Add support for the openSUSE Leap distributions https://review.openstack.org/483253 | 08:10 |
| *** dxiri has quit IRC | 08:10 | |
| hwoarang | good morning | 08:14 |
| *** Lirt has joined #openstack-ansible | 08:15 | |
| hwoarang | anyone to give me +2 for https://review.openstack.org/#/c/482529/ so get neutron in? it blocks some other roles :) TIA | 08:15 |
| hwoarang | s/so/to | 08:15 |
| Lirt | Hello OSA, I have one question about external_lb_vip_address and about external LB configuration. Is there somebody who has know-how in this topic? | 08:17 |
| *** jbadiapa_ is now known as jbadiapa | 08:18 | |
| *** cpuga has quit IRC | 08:18 | |
| *** armaan_ has joined #openstack-ansible | 08:23 | |
| *** armaan has quit IRC | 08:26 | |
| *** cshen__ has joined #openstack-ansible | 08:27 | |
| *** cshen__ has quit IRC | 08:31 | |
| *** tobberyd_ has quit IRC | 08:38 | |
| *** tobberydberg has joined #openstack-ansible | 08:38 | |
| coolkil | can anybody tel me why the console definitions are needed acros all containers? i think it is only needed on actual compute nodes. | 08:44 |
| coolkil | all nova containers that is btw | 08:47 |
| *** dxiri has joined #openstack-ansible | 08:48 | |
| andymccr | coolkil: you mean in the conf files? | 08:52 |
| coolkil | yes | 08:52 |
| andymccr | coolkil: hmm you're probably right it doesnt seem like the api/metadata etc should need that conf, so i guess its just an oversight that didnt cause any issues so nobody bothered to fix :) | 08:53 |
| *** dxiri has quit IRC | 08:53 | |
| coolkil | il take it out when adding the serial console | 08:54 |
| andymccr | coolkil: sounds good | 08:54 |
| *** cshen__ has joined #openstack-ansible | 09:01 | |
| *** cshen__ has quit IRC | 09:03 | |
| *** kristia__ has joined #openstack-ansible | 09:18 | |
| *** admin0 has joined #openstack-ansible | 09:20 | |
| admin0 | morning all | 09:21 |
| *** kristian__ has quit IRC | 09:21 | |
| *** winggundamth has joined #openstack-ansible | 09:22 | |
| *** lostRhino has joined #openstack-ansible | 09:25 | |
| *** maybebuggy has quit IRC | 09:27 | |
| *** firebat has quit IRC | 09:29 | |
| *** lostRhino has left #openstack-ansible | 09:29 | |
| *** maybebuggy has joined #openstack-ansible | 09:30 | |
| *** kysse has quit IRC | 09:33 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron master: Add support for the openSUSE Leap distributions https://review.openstack.org/482529 | 09:33 |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Ensure PIP requirements are always met https://review.openstack.org/483281 | 09:35 |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Ensure mysql-python dependency met https://review.openstack.org/483282 | 09:35 |
| eoseq | hi | 09:55 |
| eoseq | do you know how to debug jinja2 template? | 09:56 |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Ensure mysql-python dependency met https://review.openstack.org/483282 | 09:56 |
| eoseq | I hit that bug: https://bugs.launchpad.net/openstack-ansible/+bug/1691130 | 09:56 |
| openstack | Launchpad bug 1691130 in openstack-ansible "fails at repo_build : Build package requirements file" [Undecided,Incomplete] | 09:56 |
| *** thorst has joined #openstack-ansible | 10:00 | |
| odyssey4me | eoseq I'll comment in the bug | 10:01 |
| eoseq | odyssey4me: ok, thanks | 10:01 |
| kristia__ | Hi, odyssey4me. Lxc containers are not getting assigned ip addresses "aio1_utility_container-f8c09a57 RUNNING 1 onboot, openstack - -". What shall I do? | 10:02 |
| kristia__ | running lxc-ls --line | 10:03 |
| odyssey4me | kristia__ sorry, but I can't help with networking | 10:04 |
| *** thorst has quit IRC | 10:04 | |
| kristia__ | does someone here know lxc networking? and help me debug this issue? | 10:05 |
| *** ivveh has joined #openstack-ansible | 10:05 | |
| *** charcol has joined #openstack-ansible | 10:07 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Remove run-playbooks https://review.openstack.org/482253 | 10:07 |
| *** vnogin has joined #openstack-ansible | 10:07 | |
| *** markvoelker has joined #openstack-ansible | 10:09 | |
| *** toddnni has quit IRC | 10:11 | |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Ensure PIP requirements are always met https://review.openstack.org/483281 | 10:13 |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Ensure mysql-python dependency met https://review.openstack.org/483282 | 10:13 |
| kristia__ | odyssey4me: when I console into the utility container I get asked for user and pass, is that normal? because last time I didnt get asked | 10:14 |
| odyssey4me | kristia__ how did you console to it? | 10:14 |
| kristia__ | lxc-console -n aio1_utility_container-f8c09a57 | 10:15 |
| kristia__ | lxc-attach works | 10:15 |
| *** esberglu has joined #openstack-ansible | 10:16 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ops master: Change mirror used to fetch venvs by default https://review.openstack.org/483290 | 10:16 |
| *** esberglu has quit IRC | 10:16 | |
| *** cshen_ has joined #openstack-ansible | 10:19 | |
| odyssey4me | kristia__ yep, lxc-attach is the correct way | 10:20 |
| kristia__ | need the networking | 10:21 |
| odyssey4me | lxc-console will bring up the console, as the name suggests | 10:21 |
| kristia__ | should I remove all lxc containers, purge all lxc packages, delete lxcbr and run bootstrap scripts? | 10:21 |
| odyssey4me | I don't know anything about your env, and I'm tied up with other work. If that makes you happy, do it. | 10:22 |
| odyssey4me | Your test env should ideally be something you can delete and rebuild, so set yourself up with that. | 10:22 |
| odyssey4me | purging packages and conf files is very, very messy. | 10:23 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ops master: Change mirror used to fetch venvs by default https://review.openstack.org/483290 | 10:24 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Remove run-playbooks https://review.openstack.org/482253 | 10:26 |
| kristia__ | yeah I know | 10:30 |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Change mirror used to fetch venvs by default https://review.openstack.org/483290 | 10:31 |
| *** dxiri has joined #openstack-ansible | 10:33 | |
| *** udesale has quit IRC | 10:36 | |
| *** dxiri has quit IRC | 10:38 | |
| *** markvoelker has quit IRC | 10:43 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone master: Ensure that keystone restarts after db sync https://review.openstack.org/483036 | 10:46 |
| DimGR | kristia__ lxcbr0 down ; lxcbr0 up | 10:46 |
| *** acormier has joined #openstack-ansible | 10:53 | |
| *** acormier has quit IRC | 10:55 | |
| *** acormier has joined #openstack-ansible | 10:55 | |
| Lirt | Hello OSA, I have one question about external_lb_vip_address and about general external LB configuration. Is there somebody who has know-how in this topic? | 10:57 |
| maybebuggy | Hi all, not sure if somebody is able to help here, but I'm trying to do a test openstack-ansible deployment on an OpenStack (Kilo). I've built 3 controllers, 1 deploy host and 3 computes (using terraform) and created three separate tenant networks (in the hosting openstack, which uses gre as overlay network). Now I have the issue, that OpenStack by default filters outgoing packets from an unknown mac... So the containers built by | 10:59 |
| maybebuggy | OSA cannot reach other containers. Is anyone using OSA on OpenStack with tenant networks? | 10:59 |
| *** jamesdenton has joined #openstack-ansible | 10:59 | |
| *** thorst has joined #openstack-ansible | 11:01 | |
| *** cmart has quit IRC | 11:02 | |
| odyssey4me | Lirt best to just ask, and if someone can answer then they will | 11:02 |
| odyssey4me | maybebuggy kilo is long EOL and doesn't just deploy due to various changes in Ubuntu Trusty since it went EOL | 11:03 |
| maybebuggy | odyssey4me: sorry, missunderstanding. the hosting openstack is kilo :) - the thing osa tries to deploy is ocata, but i'm not yet passt the setup-infrastructure playbook | 11:05 |
| *** thorst has quit IRC | 11:05 | |
| maybebuggy | up to now, the containers are created, but it fails during the galera install because the galera container is unable to reach the repo container | 11:05 |
| *** charcol has quit IRC | 11:05 | |
| odyssey4me | maybebuggy yeah, and this is inside an openstack environment? | 11:06 |
| maybebuggy | the control machine is an instance inside that openstack kilo. yes. | 11:06 |
| odyssey4me | it's very likely that your host environment has spoofing protection which will prevent network comms by anything other than the host MAC address, so to make that work you will have to ensure that the bridges are tunneled | 11:06 |
| odyssey4me | you'll have to do a fairly complicated set of veth pairs plugged into a vxlan/gre mesh | 11:07 |
| maybebuggy | odyssey4me: yes, it's the spoofing protection ;) thats how far i got. I had hoped somebody has an idea how to easily disable that spoofing protection on kilo | 11:07 |
| Lirt | @maybebuggy You can try to disable port protection to be sure this is an issue, right? | 11:08 |
| maybebuggy | Lirt: i'm not sure how to disable port protection, if thats easily doable i would highly appreciate that | 11:08 |
| Lirt | neutron port-update --no-security-groups <ID_PORT> and neutron port-update --port-security-enabled=False <ID_PORT> | 11:09 |
| odyssey4me | maybebuggy you really shouldn't disable it because it will allow all sorts of scary things to be done in your hosting cloud | 11:09 |
| odyssey4me | but yeah, what Lirt said may work | 11:09 |
| *** winggundamth has quit IRC | 11:09 | |
| Lirt | If he will stay unconnected to other parts of the network and keep it only for test, it may be ok. | 11:09 |
| odyssey4me | I put up https://gist.github.com/odyssey4me/863e84a0f6271712c6d48980cffb958d some time ago which was using the bootstrap-host thing for testing inside an openstack cloud, that might still work. | 11:09 |
| maybebuggy | odyssey4me: i'm the only one on that tenant and it's a tenant only network | 11:09 |
| odyssey4me | You might want to try standing one up, then examine how it's all setup and works and apply it to your test env. | 11:10 |
| *** esberglu has joined #openstack-ansible | 11:10 | |
| Lirt | Ok I have one question, afaik the external load balancer should only forward traffic to internal_lb_vip_address with no port forwarding changes. Therefore I can just install VM with haproxy and setup transparent forwarding(keeping dst ports) to internal_lb_vip_address? | 11:11 |
| *** toddnni has joined #openstack-ansible | 11:11 | |
| odyssey4me | that won't work exactly as-is due to some ansible changes in 2.2 (which we use for ocata), but that's easily solvable by changing ansible_ssh_host to ansible_host | 11:11 |
| odyssey4me | Lirt the way that we (Rackspace) deploy is without haproxy using an external LB. | 11:11 |
| Lirt | <odyssey4me> you mean hardware LB | 11:12 |
| *** jamesdenton has quit IRC | 11:12 | |
| odyssey4me | The external LB needs to be pre-prepared with the IP/port mappings to the containers. Bot the external and internal LB address are on the external LB. | 11:12 |
| odyssey4me | But that will require you to be able ot reverse engineer the mappings from the inventory before you deploy. We use a script for that which builds an f5 config. | 11:13 |
| *** jamesdenton has joined #openstack-ansible | 11:13 | |
| *** cshen_ has quit IRC | 11:13 | |
| odyssey4me | see the discussion here http://eavesdrop.openstack.org/irclogs/%23openstack-ansible/%23openstack-ansible.2017-07-11.log.html#t2017-07-11T14:49:17 | 11:14 |
| odyssey4me | jamesdenton gave the command a little further down | 11:14 |
| *** esberglu has quit IRC | 11:14 | |
| odyssey4me | of course you *cloud* use the hardware LB to be in front of haproxy, that might be simpler if you're OK with the extra layer | 11:15 |
| *** thorst has joined #openstack-ansible | 11:17 | |
| odyssey4me | andymccr it would appear that the ceph gate is not as reliable as I thought and it's causing a bit of blockage for master/ocata | 11:17 |
| odyssey4me | I know that logan- is working on a fix, but perhaps we should move it back to non-voting while we do that? | 11:17 |
| Lirt | <odyssey4me> Maybe I quite don't understand that. Internal LB is configured well with haproxy playbook with all backends. Then external LB also needs to have backend configuration to all containers? Will this not work if this external lb will only proxy traffic to internal lb? I find this part of installation not very well documented. | 11:19 |
| jamesdenton | both internal and external lb can be handled by haproxy. it's not really setup to split internal and external between two different load balancers. You'll want different IPs for each, but they can be in the same subnet, or not | 11:20 |
| odyssey4me | Lirt no, let me try to be more clear - *we* use a hardware LB *without* haproxy and therefore we need to configure our hardware LB to know about all the containers. | 11:20 |
| odyssey4me | so if you want to use haproxy, then what jamesdenton said applies | 11:20 |
| odyssey4me | if you want to use the hardware LB for the external facing address, I guess that could be done - off the top of my head I can't think exactly how you'd do it - my head is trying to solve other problems right now | 11:21 |
| *** dxiri has joined #openstack-ansible | 11:22 | |
| Lirt | Y ofc I want to internal lb to be in private network and external to be as anycast public IP with software LB. I just don't want to do the configuration of external LB (which I want to be haproxy) by myself, when OSA already can generate it. So I was thinking about just reproxying it to internal LB. Maybe that is not the best idea from infrastructure and HA point of view. | 11:23 |
| *** smatzek has joined #openstack-ansible | 11:24 | |
| *** dxiri has quit IRC | 11:26 | |
| jamesdenton | that's beyond the scope of the current playbook, but you can probably make it go with some work on your end | 11:26 |
| *** vnogin has quit IRC | 11:27 | |
| Lirt | OK, thank you for help :-) | 11:27 |
| ivveh | i encountered issues with consoles when setting it up like that (private network where internal LB was isolated from external network where external LB was put). not sure if that is intended | 11:28 |
| *** toddnni has quit IRC | 11:29 | |
| jamesdenton | Nova can be configured to use a different proxy address - in that case you'd want to specify the external VIP or FQDN | 11:29 |
| jamesdenton | it may use internal by default | 11:30 |
| jamesdenton | there should be an override | 11:30 |
| ivveh | ah | 11:30 |
| ivveh | do you by change know the config parameter? | 11:30 |
| jamesdenton | not offhand but i can find it | 11:31 |
| ivveh | ill try by myself first then | 11:31 |
| ivveh | :) | 11:31 |
| ivveh | im guessting its the nova-consoleauth | 11:31 |
| Lirt | And one small question, will be scaling supported in Pike release as is written in this maintenance guide? https://docs.openstack.org/openstack-ansible/latest/admin/maintenance-tasks/scale-environment.html#add-a-new-infrastructure-host | 11:32 |
| *** lostRhino has joined #openstack-ansible | 11:32 | |
| lostRhino | good morning - | 11:32 |
| lostRhino | are all containers supposed to be sending logs to the loghost? Or is it just specific containers and hosts? If so can what machines send logs to the loghost | 11:33 |
| *** jamesden_ has joined #openstack-ansible | 11:34 | |
| *** toddnni has joined #openstack-ansible | 11:35 | |
| lostRhino | the rsyslog_all group - states children": [ | 11:36 |
| lostRhino | "rsyslog" | 11:36 |
| lostRhino | ], | 11:36 |
| lostRhino | "hosts": [] | 11:36 |
| *** jamesdenton has quit IRC | 11:37 | |
| ivveh | jamesden_: could it be mksproxy_base_url & html5_proxy_base_url? | 11:37 |
| jamesden_ | i'm using nova_spice_html5proxy_base_url | 11:38 |
| jamesden_ | with 'https://domain.com:6082/spice_auto.html' as the value | 11:38 |
| *** vnogin has joined #openstack-ansible | 11:39 | |
| ivveh | is there any difference with the internal and external anyways? except for the api's? | 11:39 |
| ivveh | security wise | 11:40 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone master: Ensure that keystone restarts after db sync https://review.openstack.org/483309 | 11:40 |
| lostRhino | I ask because the utility container (currently) is not sending logs but the repo container is - unsure why one would get set to send logs and the other would not | 11:40 |
| jamesden_ | external is configured for https, while internal is configured for http by default, IIRC | 11:40 |
| *** markvoelker has joined #openstack-ansible | 11:41 | |
| ivveh | ah thats a big difference :) | 11:42 |
| odyssey4me | ivveh yeah, by default with haproxy we set external as https and internal as http | 11:42 |
| odyssey4me | the reason for external vs internal is actually more to do with openstack's service catalog which has external and internal endpoints | 11:43 |
| ivveh | hmm just tested on my testrig, it doesn't allow http on internal tho | 11:43 |
| odyssey4me | we have set it out expecting that the infra itself will use the internal endpoints and the clients using openstack will use the external endpoints | 11:43 |
| jamesden_ | ideally, internal would be used by the services themselves, and external would be used by clients. | 11:43 |
| jamesden_ | what odyssey4me said | 11:44 |
| odyssey4me | I expect that you may have set them both to be the same ip then? | 11:44 |
| ivveh | no different subnets even | 11:44 |
| odyssey4me | either that or you're testing using an openstack client and have not told it to use the internal endpoint | 11:44 |
| ivveh | (and vlan) | 11:44 |
| odyssey4me | the default behaviour for the clients is to use the external endpoint | 11:44 |
| ivveh | ah | 11:46 |
| *** acormier has quit IRC | 11:46 | |
| *** toddnni has quit IRC | 11:46 | |
| jamesden_ | and even that breaks from time to time | 11:46 |
| *** acormier has joined #openstack-ansible | 11:47 | |
| ivveh | but i would want consoles to be reachable both from external and internal, isn't that intended? | 11:47 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone master: Ensure that keystone restarts after db sync https://review.openstack.org/483309 | 11:47 |
| jamesden_ | your consoles only need to be reached from 'public' facing clients, usually ones coming in thru horizon | 11:48 |
| jamesden_ | And that horizon URL would probably be an externally-routed IP (doesn't have to be public, just something within your org, and the proxy url should match | 11:49 |
| jamesden_ | both services are load balanced and have the same listener address, but different ports (443 vs 6082, but same VIP) | 11:49 |
| ivveh | okay | 11:49 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone stable/ocata: Replace db sync check with alternative process https://review.openstack.org/483033 | 11:49 |
| ivveh | what really the point of the internal? | 11:50 |
| jamesden_ | The internal VIP ought to be reserved for the openstack services to communicate with one another, or for clients that might be logged into the utility container, for example. But often you'll find people install clients on their workstations and whatnot,and you'll want them using the 'public' endpoint, and thus the external VIP. | 11:50 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone stable/ocata: Replace db sync check with alternative process https://review.openstack.org/483033 | 11:50 |
| ivveh | makes sense | 11:51 |
| jamesden_ | ivveh Ask such questions and you'll anger the elders | 11:51 |
| *** acormier has quit IRC | 11:51 | |
| ivveh | :) | 11:51 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone stable/ocata: Replace db sync check with alternative process https://review.openstack.org/483033 | 11:51 |
| *** armaan_ has quit IRC | 11:51 | |
| ivveh | reason for me asking was that in the case i would want to access a console "as an admin" and if i would do it via the private network, i wouldn't be able | 11:51 |
| *** armaan has joined #openstack-ansible | 11:52 | |
| ivveh | if such a need would arise from issues or whatever | 11:52 |
| jamesden_ | there are certain services that have a subset of commands that can only be used thru the 'admin' endpoint, which is often the same address as the internal endpoint, but doesn't have to be. An approach to security that really wasnt adopted by the majority of services, but it's a sort of legacy holdover | 11:52 |
| jamesden_ | naw - If you have the URL and the respective token you can reach it admin or otherwise | 11:52 |
| jamesden_ | and the console is a 'graphical' console, not a text console, so it has to be done thru a browser, not thru SSH | 11:53 |
| odyssey4me | ivveh the 'internal' endpoint is not meant as your admin network... it's internal to the infra | 11:54 |
| ivveh | yeah that was what i meant | 11:54 |
| ivveh | through a browser via the internal network | 11:54 |
| ivveh | (maybe i was abit unclear) | 11:54 |
| odyssey4me | jamesden_ yeah, it's a legacy holdover - the admin and internal endpoints are no longer different in any way from a service standpoint | 11:54 |
| andymccr | odyssey4me: hmm i guess we could, thats pretty weird though - it was working pretty flawlessly (even better than the full build on master at least). | 11:54 |
| odyssey4me | andymccr from the discussion/investigation yesterday it's down to a bit of a race condition | 11:55 |
| odyssey4me | we implement nova-compute, then implement the ceph client - so until ceph client is there, nova errors out because it can't connect to its storage | 11:55 |
| odyssey4me | that is now coming to a head due to the rolling upgrade changes implementing things in a different order than previously | 11:56 |
| odyssey4me | previously we kinda bashed it with a hammer until it worked, now that we're doing a final reload (not restart) it seems we're only sometimes doing it at just the right time for the driver | 11:56 |
| ivveh | jamesden_: so just to make it perfectly clear, its not possible to reach the graphical console via a browser when on the internal network if it is totally isolated from the public one :) | 11:57 |
| andymccr | ok. well i guess there's not much we can do until thats fixed then - rolling upgrades is a real pita, cant wait to see the back of that. | 11:57 |
| jamesden_ | ivveh I would consider the internal network internal to the cloud itself and try not to use it as a client | 11:57 |
| odyssey4me | the right solution is to get the ceph client there at the right time so that we don't have to restart twice (once for nova, then once for ceph client) and logan- is working that out | 11:57 |
| ivveh | alright, so "not intended" then | 11:57 |
| jamesden_ | ivveh it absolutely is possible, just probably not best practice to use it as a client if you can avoid | 11:57 |
| ivveh | roger that! | 11:58 |
| *** Oku_OS has quit IRC | 11:58 | |
| *** toddnni has joined #openstack-ansible | 11:58 | |
| odyssey4me | yeah, it's really bringing up a lot of weird bugs due to the change in behaviour... it'll be worth it, but it's a pita right now | 11:58 |
| odyssey4me | andymccr so annoying though, see https://review.openstack.org/#/c/482090/ has great ceph times and sucky non-ceph times | 11:59 |
| openstackgerrit | Chris Beukers proposed openstack/openstack-ansible-os_nova master: [WIP] adding nova-serialconsole support https://review.openstack.org/479844 | 11:59 |
| *** udesale has joined #openstack-ansible | 12:01 | |
| andymccr | yeah we are preparing for teh release time with gate uncertainty coming up ;P which is gonna be worse - im not sure how we can improve our run times significantly though | 12:01 |
| *** foutatoro has joined #openstack-ansible | 12:06 | |
| *** thorst has quit IRC | 12:06 | |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Improve openstack-release file discovery https://review.openstack.org/482608 | 12:09 |
| *** armaan has quit IRC | 12:09 | |
| *** armaan_ has joined #openstack-ansible | 12:09 | |
| *** thorst has joined #openstack-ansible | 12:11 | |
| foutatoro | hi all, I would like to know if we can deploy hypervisors VMware vSphere and Hyper-V on compute nodes with openstack-ansible ? | 12:12 |
| ivveh | blasphemy | 12:12 |
| *** markvoelker has quit IRC | 12:13 | |
| foutatoro | ivveh: for specific use cases | 12:16 |
| odyssey4me | foutatoro nope, no-one's ever put in the work to do that | 12:20 |
| odyssey4me | personally, I'd love to see hyper-v support as I think it's well suited to how nova works | 12:21 |
| odyssey4me | vsphere's nova implementation is quite weird and I'm not a fan | 12:21 |
| *** schwicht has joined #openstack-ansible | 12:21 | |
| odyssey4me | andymccr Well, I hate to do it, but perhaps with the demise of OSIC we should increase the timeout for the integrated build to 2 hrs? | 12:22 |
| andymccr | odyssey4me: im tempted | 12:22 |
| andymccr | odyssey4me: we are just over 90mins, it starts a slippery slope, but id rather we are able to merge working code than not. | 12:22 |
| odyssey4me | yep | 12:22 |
| odyssey4me | ok, lemme push up a patch for that | 12:23 |
| andymccr | and spending a lot of time trying to optimize tiny things for minimal gains isnt that fun :P | 12:23 |
| odyssey4me | and to make ceph non-voting again until we work out the issue there | 12:23 |
| andymccr | yeah | 12:23 |
| andymccr | damn thats annoying - but in logan- we trust :P | 12:23 |
| odyssey4me | I think perhaps a ceph scenario in the nova role would be a good idea | 12:23 |
| andymccr | true | 12:23 |
| *** toddnni has quit IRC | 12:30 | |
| *** schwicht has quit IRC | 12:30 | |
| *** markvoelker has joined #openstack-ansible | 12:32 | |
| Tahvok | Not sure what's going on.. | 12:36 |
| lostRhino | can anyone help me with my rsyslog question, please? | 12:36 |
| Tahvok | The provision at some point is stopping apache2 at the keystone container, and then failing, cuz keystone is not available | 12:37 |
| jamesden_ | Tahvok that is a known issue at the moment, and a patch is being worked on | 12:37 |
| *** schwicht has joined #openstack-ansible | 12:37 | |
| Tahvok | jamesden_: any workarounds? | 12:37 |
| *** woodard has quit IRC | 12:38 | |
| *** woodard has joined #openstack-ansible | 12:39 | |
| *** deep-book-gk_ has joined #openstack-ansible | 12:39 | |
| jamesden_ | Tahvok You can try making a backup of /etc/ansible/os_keystone/tasks/keystone_db_setup.yml and implement the one here: https://review.openstack.org/483033, then rerun the playbook | 12:40 |
| *** deep-book-gk_ has left #openstack-ansible | 12:40 | |
| jamesden_ | i'msorry, that's /etc/ansible/roles/os_keystone... | 12:41 |
| logan- | o/ | 12:42 |
| *** yifei has quit IRC | 12:42 | |
| logan- | it looks like the include_role for ceph_client is working but i keep hitting timeout roulette so it is hard to verify anything | 12:42 |
| logan- | i guess i'll push glance/cinder versions and get things finalized based on whats there if nobody raises and big concerns | 12:43 |
| logan- | re: https://review.openstack.org/#/c/483061/ and https://review.openstack.org/#/c/483062/ | 12:43 |
| errr | hello, where would I add ansible_ssh_common_args: '-o StrictHostKeyChecking=no' something like this normally goes in the host section of my ansible.cfg but since OSA uses the inventoy.json file would I need to add it in there somewhere? | 12:44 |
| *** gillesMo has joined #openstack-ansible | 12:45 | |
| strigazi | odyssey4me hello, I'm Spyros from the magnum team | 12:45 |
| odyssey4me | errr I think we have that on by default | 12:46 |
| odyssey4me | errr https://github.com/openstack/openstack-ansible/blob/master/scripts/openstack-ansible.rc#L46 | 12:46 |
| errr | odyssey4me: wow, a greo shows you are right | 12:46 |
| errr | grep* | 12:46 |
| errr | thanks | 12:47 |
| odyssey4me | andymccr logan- well, this may help: https://review.openstack.org/483338 | 12:47 |
| strigazi | odyssey4me in magnum we use heat and to create cluser of vms. That requires the vm to signal heat. In a standard OSA the endpoint are only in the management network. Any suggestion on how to create a route between the neutron private network to the OSA management network? | 12:47 |
| errr | odyssey4me: for some reason the git module ignores that | 12:48 |
| errr | odyssey4me: or so my co-worker claims anyway.. I havent tried to verify that yet myself | 12:49 |
| odyssey4me | errr that setting is for ansible's host connection, I don't think it applies to modules connecting to targets | 12:49 |
| errr | odyssey4me: see: https://docs.ansible.com/ansible/git_module.html | 12:50 |
| Tahvok | jamesden_: still failing at the same step | 12:50 |
| Tahvok | apache2 is stopped | 12:50 |
| odyssey4me | errr yeah, so the code path for that is likely different | 12:50 |
| *** sxc731 has joined #openstack-ansible | 12:51 | |
| odyssey4me | Tahvok I have a patch for that in review: https://review.openstack.org/#/q/Ide64927e43e7684f03be7a73b893283c80e89afc,n,z | 12:51 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_glance master: Include ceph_client role if needed https://review.openstack.org/483342 | 12:51 |
| jamesden_ | Tahvok, you'll need to start apache manually inside the keystone container after that first failure, since it did not get restarted, then run again. | 12:51 |
| jamesden_ | Sorry about that | 12:51 |
| Tahvok | Oh, ok | 12:52 |
| odyssey4me | errr we also set it here: https://github.com/openstack/openstack-ansible/blob/master/group_vars/all/all.yml#L105 | 12:52 |
| Tahvok | Trying now | 12:52 |
| errr | odyssey4me: yeah thats what I found via grep | 12:52 |
| odyssey4me | I would guess, though, that the best would be to make use of the git module's args instead? | 12:52 |
| Tahvok | odyssey4me: it's the same patch that jamesden_ provided? | 12:52 |
| odyssey4me | it's more explicit | 12:52 |
| errr | odyssey4me: how do I add the git module args correctly with out editing the playbook? | 12:53 |
| *** kristia__ has quit IRC | 12:55 | |
| *** kristian__ has joined #openstack-ansible | 12:56 | |
| *** kristian__ has quit IRC | 12:56 | |
| *** kristian__ has joined #openstack-ansible | 12:56 | |
| Tahvok | jamesden_: btw, I had a question about the f5 scripts you showed me yesterday | 12:58 |
| *** schwicht has quit IRC | 12:58 | |
| Tahvok | In the script, I can provide an ssl domain address, and it goes as 'destination' of ssl virtual servers | 12:59 |
| Tahvok | However, I get an error that no such virtual address is available. And that is correct. However, I cannot find a way to add fqdn as a virtual address in the f5 | 12:59 |
| Tahvok | How did you do it then? | 13:00 |
| jamesden_ | hmm | 13:01 |
| jamesden_ | The f5 gave you the error? Or did it occur on scriptgeneration | 13:01 |
| Tahvok | The f5 gave the error | 13:02 |
| *** winggundamth has joined #openstack-ansible | 13:02 | |
| Tahvok | jamesden_: a row example it gave me to run: https://gist.github.com/Tahvok/3a9c0a054ab30591ffd537a27d0e9f15 | 13:03 |
| jamesden_ | The --ssl-public-ip switch is where you specify the address you want to use as the external VIP address, and --ssl-domain-name is what will be the corresponding FQDN. I believe its used in the self-signed SSL cert generation | 13:03 |
| jamesden_ | the script is dumb in that it doesnt do any validation | 13:04 |
| Tahvok | This is how I've generated the configs: python f5-config.py --ssl-domain-name kloud-play.kenshoo.com --print | 13:05 |
| Tahvok | Everything else is provided in the inventory | 13:05 |
| odyssey4me | evrardjp answered your question in https://review.openstack.org/#/c/483309/2/tasks/keystone_db_setup.yml | 13:05 |
| evrardjp | thanks | 13:05 |
| jamesden_ | Tahvok try adding --ssl-public-ip with the IP for giggles and see if the output changes | 13:06 |
| evrardjp | odyssey4me: funny, without ready the code I'd have said the other way :p | 13:06 |
| Tahvok | jamesden_: and indeed it does | 13:06 |
| odyssey4me | evrardjp what do you mean? | 13:06 |
| jamesden_ | try to apply the new config | 13:06 |
| *** kylek3h has joined #openstack-ansible | 13:07 | |
| coolkil | odyssey4me: (or anyone else) what is the groupname for nova compute? | 13:07 |
| odyssey4me | coolkil group name for which purpose? to add to openstack_user_config/conf.d or to refer to in playbooks/roles? | 13:07 |
| Tahvok | jamesden_: I did this already, I've the second vip address as I wanted to continue with ansible | 13:08 |
| evrardjp | odyssey4me: nothing important. | 13:08 |
| Tahvok | But at least you explained me what was wrong yesterday :) | 13:08 |
| evrardjp | voted | 13:08 |
| *** gkadam has quit IRC | 13:08 | |
| Tahvok | Thanks a lot! | 13:08 |
| jamesden_ | sure thing | 13:08 |
| coolkil | well in nova.conf i got an entry like so {% if 'nova_console' in group_names %} that is nova console how to set nova compute? | 13:09 |
| odyssey4me | coolkil ok, let me show you where to find them :) | 13:10 |
| odyssey4me | coolkil this dict: https://github.com/openstack/openstack-ansible-os_nova/blob/master/defaults/main.yml#L430 | 13:10 |
| odyssey4me | for example, nova compute is here: https://github.com/openstack/openstack-ansible-os_nova/blob/master/defaults/main.yml#L442 | 13:10 |
| odyssey4me | the value of 'group' is the name for the nova-compute group | 13:11 |
| odyssey4me | ie nova_compute | 13:11 |
| *** lenserpgo has quit IRC | 13:11 | |
| coolkil | thnx! exactly what i needed to know | 13:11 |
| *** hw_wutianwei has quit IRC | 13:13 | |
| *** armaan_ has quit IRC | 13:16 | |
| *** schwicht has joined #openstack-ansible | 13:17 | |
| odyssey4me | andymccr evrardjp logan- so, related to https://review.openstack.org/483309 but also catching another issue is https://review.openstack.org/483033 | 13:17 |
| odyssey4me | right now if you run the keystone playbook more than once, keystone will be broken - that solves it | 13:18 |
| odyssey4me | and personally tested by jamesden_ himself :) | 13:18 |
| evrardjp | personally tested working? or personally tested that without it, it breaks? | 13:19 |
| odyssey4me | evrardjp both | 13:20 |
| jamesden_ | i haven't tested the latest patch, but patchset 1 worked | 13:20 |
| andymccr | this stuff is seriously snowballing isnt it :P | 13:20 |
| odyssey4me | I personally confirmed that it's broken yesterday, and made the patch to fix it. | 13:20 |
| odyssey4me | the iteration from patch set 1 was to cover the many, many different ways we configure keystone for newton/ocata. | 13:20 |
| odyssey4me | andymccr yeah, the unfortunate pain of porting so much work back is that it destabilises a stable branch | 13:21 |
| odyssey4me | I hope that we never have to do that level of backporting ever again. | 13:22 |
| evrardjp | I'll review asap | 13:22 |
| andymccr | likewise | 13:23 |
| *** armaan has joined #openstack-ansible | 13:24 | |
| *** cathrichardson has joined #openstack-ansible | 13:24 | |
| *** hw_wutianwei has joined #openstack-ansible | 13:24 | |
| *** cathrich_ has quit IRC | 13:25 | |
| *** acormier has joined #openstack-ansible | 13:25 | |
| *** acormier has quit IRC | 13:26 | |
| *** acormier has joined #openstack-ansible | 13:27 | |
| *** esberglu has joined #openstack-ansible | 13:28 | |
| Tahvok | jamesden_: can I mark the patchset as working in gerrit (never used gerrit before)? | 13:28 |
| Tahvok | It completed fine for me | 13:29 |
| mbuil | andymccr: do you have any news regarding the tacker role? did somebody from Openstack replied? | 13:29 |
| andymccr | mbuil: so the patches are up - its going to be a waiting game https://review.openstack.org/#/c/482873/ take a look and make sure i havnt typo'd anything :) | 13:30 |
| jamesden_ | Tahvok I suppose you can +1 it once you create an account, sure | 13:30 |
| andymccr | and there is a governance patch up too - https://review.openstack.org/#/c/482872/ but thats more of a waiting game and isnt needed for the initial sync so would worry less about that | 13:30 |
| jamesden_ | I'm sure odyssey4me would appreciate the feedback :) | 13:31 |
| Tahvok | jamesden_: I'm signed in.. Not sure where the +1 button is though | 13:31 |
| jamesden_ | Tahvok Once you click the Reply button it should appear | 13:31 |
| Tahvok | Got it | 13:31 |
| mbuil | andymccr: ok! thanks | 13:32 |
| mhayden | buenos dias | 13:32 |
| Tahvok | Thanks a lot guys! Ansible finished successfully for the first time. Not dashboard though, gonna investigate it now | 13:32 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_nova master: Include ceph_client role if needed https://review.openstack.org/483061 | 13:33 |
| jamesden_ | Tahvok happy to help. | 13:33 |
| *** thorst has quit IRC | 13:33 | |
| *** Oku_OS-away has joined #openstack-ansible | 13:34 | |
| *** Oku_OS-away is now known as Oku_OS | 13:34 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_cinder master: Include ceph_client role if needed https://review.openstack.org/483394 | 13:34 |
| kristian__ | andymccr: have you done lxc networking? My containers get ip addresses np, but dns is not working. I have attached and pointed one resolv.conf to google dns and then it worked. Do you know how to do it, that it will apply to all issues? | 13:34 |
| kristian__ | s/issues/containers/g | 13:34 |
| *** coolkil has quit IRC | 13:35 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 13:35 |
| andymccr | kristian__: so before that you couldnt resolve anything? | 13:37 |
| kristian__ | before changing the resolv.conf in the container, nope, ansible script faild at downloading packages | 13:37 |
| evrardjp | odyssey4me: commented on 483033 | 13:38 |
| evrardjp | detail but useful for readability/ara | 13:38 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_nova master: Include ceph_client role if needed https://review.openstack.org/483061 | 13:40 |
| *** weezS has joined #openstack-ansible | 13:41 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone stable/ocata: Replace db sync check with alternative process https://review.openstack.org/483033 | 13:41 |
| kristian__ | andymccr: does there need to be something setup in lxc configs or put dns in /etc/network/interfaces? | 13:41 |
| sufyan68 | Anyone knows how to add a certificate to horizon? because public apis does not work | 13:41 |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Improve openstack-release file discovery https://review.openstack.org/482608 | 13:41 |
| andymccr | Tahvok: thanks for testing that patch btw | 13:41 |
| odyssey4me | evrardjp good catch for https://review.openstack.org/483033 - fixed | 13:42 |
| *** klamath has joined #openstack-ansible | 13:42 | |
| *** klamath has quit IRC | 13:42 | |
| *** klamath has joined #openstack-ansible | 13:43 | |
| odyssey4me | kristian__ the containers will use the host via nat as a resolver, so if that's the issue it would seem that dnsmasq wasn't running right for the lxc bridge | 13:43 |
| odyssey4me | or that the lxc bridge wasn't up | 13:44 |
| *** smatzek has quit IRC | 13:44 | |
| kristian__ | should I restart lxcbr0? odyssey4me | 13:44 |
| odyssey4me | that also happens to be how they receive their first IP | 13:44 |
| odyssey4me | kristian__ yes, I think someone suggested that earlier | 13:44 |
| andymccr | kristian__: check lxbr0 also check you can resolve things from the host itself too. | 13:45 |
| jamesden_ | sufyan68 There are some overrides, namely haproxy_user_ssl_cert, haproxy_user_ssl_key, haproxy_user_ssl_ca_cert, and possibly horizon_server_name | 13:45 |
| kristian__ | yeah, I stopped the bridge and let the scripts to put it back up | 13:45 |
| jamesden_ | otherwise i'm pretty sure the haproxy playbook will just roll with a self-signed cert | 13:45 |
| *** lostRhino has left #openstack-ansible | 13:47 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Remove run-playbooks https://review.openstack.org/482253 | 13:49 |
| *** yifei has joined #openstack-ansible | 13:50 | |
| *** acormier has quit IRC | 13:52 | |
| *** acormier has joined #openstack-ansible | 13:53 | |
| *** acormier has quit IRC | 13:54 | |
| Tahvok | andymccr: np, hopefully I'll get more involved with the code, as we are moving to using ansible as our main provision tool for openstack | 13:54 |
| *** acormier has joined #openstack-ansible | 13:54 | |
| openstackgerrit | git-harry proposed openstack/openstack-ansible-ops master: Improve openstack-release file discovery https://review.openstack.org/482608 | 13:58 |
| odyssey4me | mhayden wanna up your vote now that the tests have passed? https://review.openstack.org/483070 | 13:58 |
| odyssey4me | andymccr I think we can likely save quite a bit of gate time by giving the galera and rabbitmq roles an overhaul | 14:00 |
| odyssey4me | we're doing a lot of fact setting where we could use a dynamic local fact script instead | 14:01 |
| andymccr | odyssey4me: yeah maybe, im wondering if we shouldn't have sets of playbooks for install vs upgrade - that way we can just blast all on install, and upgrade we are a bit more serialized about it. | 14:01 |
| andymccr | ptg topics for sure | 14:02 |
| odyssey4me | also, I'm kinda thinking that perhaps across the board we should adopt the strategy of packages being installed by roles with whatever repo config is already on the host - we should extract the repo config out into its own role so it's done once, in one place | 14:02 |
| odyssey4me | that will cut a huge ream of tasks and also cut out the constant apt refreshes all over the place | 14:02 |
| odyssey4me | andymccr I'm not sure how much time that would really save us, and having to try and cater for both execution strategies would probably be more difficult than helpful. | 14:04 |
| andymccr | the thing is if you have a large deployment, running in serial seems massively massively inefficient | 14:04 |
| odyssey4me | That said, I did do https://review.openstack.org/#/c/469605/1/playbooks/common-tasks/determine-execution-strategy.yml in an early patch set in order to be able to do exactly that. we could perhaps re-explore it at some point. | 14:04 |
| odyssey4me | the serial is only for the control plane | 14:04 |
| odyssey4me | the compute execution, for example, is set to 20% | 14:05 |
| andymccr | hmm yeah i guess that is true - but its stil something i shouldn;t have to care about if im doing a new install | 14:05 |
| odyssey4me | also, the control plane is mostly done using 1, 100% - meaning one is done first, then everything else | 14:05 |
| *** smatzek has joined #openstack-ansible | 14:06 | |
| odyssey4me | so I think we have a decent balance right now and I don't think trying to do a dual strategy approach is worth the added complexity | 14:06 |
| odyssey4me | I think we should rather focus on simplifying more. | 14:06 |
| *** acormier_ has joined #openstack-ansible | 14:07 | |
| *** acormier_ has quit IRC | 14:07 | |
| odyssey4me | We've had a lot of organic growth and I think it's best we regroup, step back and reconsider some of our patterns. | 14:07 |
| *** acormier_ has joined #openstack-ansible | 14:07 | |
| odyssey4me | a proper split between 'build' vs 'install' vs 'deploy' would be nice, but that will be a rather fundamental change and a *lot* of work | 14:08 |
| odyssey4me | that would aid my team's goals, so I'll be punting for us to get that as assigned work for queens | 14:09 |
| andymccr | i think that probably would've been a better first step to achieving seamless upgrades too though. the problem we have is we've added in a lot of complexity so making those simplifying changes is now harder, and we cant afford to do anymore of this kind of similar back-porting. | 14:10 |
| *** acormier has quit IRC | 14:10 | |
| odyssey4me | yeah, I'm not OK with porting this volume of change back at this frequency | 14:11 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_cinder master: Include ceph_client role if needed https://review.openstack.org/483394 | 14:12 |
| *** cpuga has joined #openstack-ansible | 14:16 | |
| kristian__ | odyssey4me: I have restarted lxcbr0 and recreated the containers, the same problem with the dns | 14:16 |
| andymccr | kristian__: can you resolve things from your host? | 14:16 |
| kristian__ | andymccr: yes I can | 14:17 |
| *** firebat has joined #openstack-ansible | 14:17 | |
| andymccr | are there any ip range overlaps for lxcbr0 ip range? | 14:17 |
| *** cpuga has quit IRC | 14:18 | |
| *** sxc731 has quit IRC | 14:18 | |
| firebat | Can anyone tell me where to look for how keystone adds the service accounts to itself? Only admin is being loaded into galera | 14:18 |
| andymccr | firebat: the roles do that individually, so when the glance role runs (for example) it will add a glance service account | 14:18 |
| *** cpuga has joined #openstack-ansible | 14:18 | |
| andymccr | that way you only get the service accounts that are actually going to be used by the deployed services | 14:18 |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Improve openstack-release file discovery https://review.openstack.org/482608 | 14:18 |
| andymccr | kristian__: if you check resolv.conf inside a container does it point to like 10.255.255.1 for example? and if you do a netstat -plnt | grep dnsmasq on your infra host do you see it listening there? | 14:19 |
| firebat | andymccr: Ok cool... so the glance role is failing then when it tries to authorize | 14:20 |
| firebat | What authentication should it be using to add those users to the keystone service? It does it via the API right? | 14:20 |
| andymccr | firebat: ahh ok - i think this is the same bug we're trying to get a fix in for asap (sorry about that): https://review.openstack.org/#/c/483033/ | 14:21 |
| firebat | I think I manually patched that in for testing wonder if i lost it | 14:22 |
| *** acormier has joined #openstack-ansible | 14:22 | |
| *** tobberyd_ has joined #openstack-ansible | 14:22 | |
| *** cpuga has quit IRC | 14:22 | |
| *** acormier has quit IRC | 14:23 | |
| *** acormier has joined #openstack-ansible | 14:23 | |
| *** yifei has quit IRC | 14:25 | |
| *** tobberydberg has quit IRC | 14:25 | |
| *** acormier_ has quit IRC | 14:25 | |
| firebat | andymccr: Yeah looks like I picked that commit and added it into my deploy process... just to clarify the glance service user should be added via a rest call right? | 14:26 |
| *** phschwartz has joined #openstack-ansible | 14:27 | |
| *** tobberyd_ has quit IRC | 14:28 | |
| odyssey4me | andymccr this one will help speed things up just a little: https://review.openstack.org/483003 | 14:28 |
| *** thorst has joined #openstack-ansible | 14:29 | |
| *** cshen has quit IRC | 14:32 | |
| andymccr | firebat: yeah we use the keystone client to add the user in the glance role, but i think the bug we have is that keystone isnt starting properly so it wont accept the request :( | 14:35 |
| *** cshen has joined #openstack-ansible | 14:35 | |
| *** galstrom_zzz is now known as galstrom | 14:37 | |
| openstackgerrit | Merged openstack/openstack-ansible-galera_client master: tasks: galera_client_install_zypper: Drop zypper workaround for Ansible < 2.2 https://review.openstack.org/482639 | 14:37 |
| *** acormier_ has joined #openstack-ansible | 14:37 | |
| *** gouthamr has joined #openstack-ansible | 14:37 | |
| *** marst has quit IRC | 14:37 | |
| *** acormier_ has quit IRC | 14:37 | |
| *** acormier_ has joined #openstack-ansible | 14:38 | |
| odyssey4me | hwoarang I'd like to thank you for working so diligently through all the roles, picking up and fixing things in sensible ways, and most especially for helping normalise them into good patterns which we just haven't been able to keep all the roles up to as a standard. | 14:39 |
| *** acormier has quit IRC | 14:40 | |
| DimGR | odyssey4me what help OSA would need with magnum now that i have it up and running | 14:40 |
| *** cpuga has joined #openstack-ansible | 14:41 | |
| odyssey4me | DimGR something along the lines of https://docs.openstack.org/openstack-ansible/latest/contributor/core-reviewers.html#general-responsibilities :) | 14:42 |
| odyssey4me | essentially care about whether the magnum role is actively tested in a sensible way, that it's actively adjusted as changes happen in the upstream service (conf option changes, etc), and that it is patched to improve it to make better use of ansible features and patterns which are developing in other roles | 14:43 |
| DimGR | alright | 14:43 |
| odyssey4me | the keystone, cinder, glance, nova, neutron and swift roles are the most actively maintained as they're used most heavily | 14:43 |
| hwoarang | odyssey4me: no problem. It's a pleasure :) | 14:44 |
| *** cpuga has quit IRC | 14:44 | |
| *** cpuga has joined #openstack-ansible | 14:44 | |
| odyssey4me | DimGR as magnum relies on heat, it'd be great if you could help with the heat role too | 14:44 |
| odyssey4me | if each of us keep an eye on developments and actively maintain a subset of roles, it'll be easier for everyone | 14:45 |
| *** sxc731 has joined #openstack-ansible | 14:46 | |
| *** lucasxu has joined #openstack-ansible | 14:46 | |
| *** marst has joined #openstack-ansible | 14:47 | |
| *** schwicht has quit IRC | 14:48 | |
| *** schwicht has joined #openstack-ansible | 14:50 | |
| *** tobberydberg has joined #openstack-ansible | 14:51 | |
| sxc731 | Greetings all! I was watching @andymccr's Boston pres and saw that OSA officially supports integrated Ceph deployment as of Ocata, which is just what I need ;-) | 14:55 |
| *** andreas_s has quit IRC | 14:55 | |
| sxc731 | Can anyone point me to the latest docs on how to achieve this. I have found a few blogs here and there but they seem somewhat out of date... | 14:56 |
| *** tobberydberg has quit IRC | 14:56 | |
| andymccr | sxc731: hey! hope the talk was useful | 14:59 |
| andymccr | the deploy guide is here: https://docs.openstack.org/project-deploy-guide/openstack-ansible/ocata/ it has a section about integrating ceph-ansible | 14:59 |
| sxc731 | The talk was great! ... and full of promises ;-) | 14:59 |
| andymccr | https://docs.openstack.org/project-deploy-guide/openstack-ansible/ocata/app-ceph.html | 14:59 |
| andymccr | its pretty straight forward, once you understand how OSA deployments are configured and work, it is just another host definition for ceph hosts and away you go! | 14:59 |
| *** SerenaFeng has joined #openstack-ansible | 14:59 | |
| andymccr | the deployment guide should hopefully explain all that to you - and then the ceph bits are an easy addition - but if you run into issues let us know how we can improve the docs | 15:00 |
| sxc731 | Yeah, I saw that.... So the blogs I've seen suggest some adjustments are necessary in user_variables.yml | 15:00 |
| admin0 | andymccr, is it possible to give basic examples there . assuming there are 3 servers for ceph that will act as mon and osd | 15:00 |
| openstackgerrit | Major Hayden proposed openstack/ansible-hardening master: Fix auditd remote conf check https://review.openstack.org/483436 | 15:01 |
| sxc731 | Hi @admin0, I think it was your FAQ entry I was reading here: https://www.openstackfaq.com/openstack-ansible-managed-ceph/ | 15:01 |
| odyssey4me | yeah, some user_variables entries will be needed | 15:02 |
| odyssey4me | we'd like to try and automate that away, but I don't think anyone's had the time to get it done | 15:02 |
| andymccr | admin0: yeah we should probably expand that | 15:02 |
| admin0 | sxc731, that was a jerry rigged setup :) .. i am trying to get more info no a more acceptable setup | 15:02 |
| sxc731 | TBH, this looked quite a bit more noob-friendly than the official docs (and I understand everyone is busy!) | 15:02 |
| admin0 | sxc731, what i did was took osa and rackspace's own playbook and kind of made it work | 15:03 |
| admin0 | i am here to test and document :) | 15:03 |
| sxc731 | Wow that's quite a tall - but obviously most welcome - order! | 15:03 |
| *** armaan has quit IRC | 15:04 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi master: templates: gnocchi-httpd: Ensure proper user control in gnocchi root https://review.openstack.org/482632 | 15:06 |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi master: Add support for the openSUSE Leap distributions https://review.openstack.org/482633 | 15:06 |
| sxc731 | So I presume @admin0's FAQ is still the best place to look? (despite the "not officially supported" introductory comment)? | 15:06 |
| admin0 | i was planning to test that for ocata again :) | 15:07 |
| odyssey4me | admin0 it'd be super-nice to have a patch pushed up to add an example config for a ceph environment | 15:07 |
| openstackgerrit | Merged openstack/openstack-ansible-os_heat master: Add support for the openSUSE Leap distributions https://review.openstack.org/482551 | 15:07 |
| odyssey4me | something like the 'prod' environment - but using ceph instead of nfs | 15:07 |
| spotz | andymccr: don't forget agenda. Meeting ping shortly apparently I can't find my notes:( | 15:08 |
| andymccr | spotz: haha thanks - yeah lemme quickly do that! | 15:08 |
| admin0 | i have a working ceph where ceph is deployed seperately, and i am also testing to use ceph but managed by proxmox :) .. so proxmox manages and gives ceph to OSA | 15:09 |
| sxc731 | admin0, odyssey4me: couldn't agree more; perhaps we can work on this together as I certainly have a use-case (trying to replace an old Fuel-deployed cloud; and Fuel certainly made that easy... if impossible to upgrade later...) | 15:09 |
| openstackgerrit | Merged openstack/openstack-ansible-galera_server master: tasks: galera_install_zypper: Drop zypper workaround for Ansible < 2.2 https://review.openstack.org/482638 | 15:09 |
| spotz | Greetings my fellow OSAers! cloudnull, DimGR, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, odyssey4me, serverascode, rromans, erikmwilson, mancdaz, _shaps_, BjoernT, claco, echiu, dstanek, jwagner, ayoung, prometheanfire, evrardjp, arbrandes, mhayden, scarlisle, luckyinva, ntt, javeriak, spotz, vdo, jmccrory, alextricity25, jasondotstar, admin0, michaelgugino, ametts, v1k0d3n, severion, bgmccollum, d | 15:09 |
| spotz | arrenc, JRobinson__, asettle, colinmcnamara, thorst, adreznec, eil397, qwang,nishpatwa_, cathrichardson, drifterza, sc68cal | 15:09 |
| spotz | Friendly reminder we will be meeting in 1 hour in #openstack-meeting-4 Agenda is available at https://wiki.openstack.org/wiki/Meetings/openstack-ansible#Agenda_for_next_meeting | 15:09 |
| *** lucasxu has quit IRC | 15:09 | |
| admin0 | sxc731, i merged osa and rackspace in that method .. so if some of the cores or the actual people from rackspace who use that can confirm that is a good way, we can start on it | 15:10 |
| admin0 | but there is also decapod these days that manages ceph | 15:10 |
| admin0 | ceph install, decapod, proxmox -- there are many ways to get ceph up | 15:11 |
| odyssey4me | admin0 that must be pre-ocata work, because in ocata we have the ability to deploy the ceph cluster too | 15:11 |
| odyssey4me | no need to mash it up | 15:11 |
| admin0 | odyssey4me, it was pre ocata :) | 15:11 |
| odyssey4me | admin0 if you push up the patch, then we can discuss whether it's right in review | 15:12 |
| odyssey4me | it's easier to discuss in review | 15:12 |
| admin0 | then let me test it with ocata | 15:12 |
| admin0 | find and fix the flaws and then submit a patch | 15:12 |
| odyssey4me | awesome, thanks | 15:12 |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: tasks: horizon_apache: Use the apache2_module Ansible module https://review.openstack.org/482923 | 15:13 |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: Add support for the openSUSE Leap distributions https://review.openstack.org/482924 | 15:13 |
| sxc731 | admin0, odyssey4me: indeed! Please let me know if I can help; I'm certainly up for testing | 15:13 |
| admin0 | sxc731, if your platfrom is ready to go, then just follow the docs and see if you can get it all up and running | 15:14 |
| odyssey4me | sxc731 the best is to setup a test environment that you're able to destroy and rebuild, often - then go ahead and deploy and figure it all out | 15:14 |
| odyssey4me | I would suggest starting with a well known config, perhaps the example configs from the appendices. That'll help you understand how the config maps to a result. | 15:15 |
| odyssey4me | Then try and work out the ceph bits afterwards, once you have your feet wet. | 15:15 |
| admin0 | sxc731, i use a big server and then multiple VMS ( as i have wrote in the site ) and with a virtual vyos as router, i am able to replicate all my production env, including the IPs being used | 15:15 |
| admin0 | even on the public side | 15:16 |
| kristian__ | andymccr: it is there "tcp 0 0 10.255.255.1:53 0.0.0.0:* LISTEN 22224/dnsmasq" | 15:17 |
| admin0 | odyssey4me, i have a test setup up . i will test ceph with ocata the integrated way | 15:18 |
| admin0 | its something i can re-create again and again | 15:18 |
| admin0 | just need to run ./redo.sh :D | 15:18 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh master: templates: aodh-httpd: Ensure proper user control in aodh root https://review.openstack.org/482333 | 15:18 |
| *** Oku_OS is now known as Oku_OS-away | 15:20 | |
| *** acormier has joined #openstack-ansible | 15:22 | |
| *** acormier has quit IRC | 15:22 | |
| admin0 | sxc731, what is your setup like ? | 15:23 |
| *** acormier has joined #openstack-ansible | 15:23 | |
| admin0 | how many nodes ? | 15:23 |
| sxc731 | odyssey4me: re "try and work out the ceph bits afterwards"... sure that's kind of where I was when I asked the q. I just thought admin0's page was going to bring me a little further than a bare pointer to ceph-ansible/group_vars/all.yml.sample | 15:25 |
| *** acormier_ has quit IRC | 15:25 | |
| admin0 | sxc731, i am playing with it tonight | 15:26 |
| admin0 | hands itching :D | 15:26 |
| sxc731 | admin0: it's a couple of bare metal boxes on which I have spun up a number of VMs (using Ubuntu MAAS and some automation to bring up the br-xxx etc) | 15:26 |
| sxc731 | That's my test env. Prod has 3 bare metal controllers and 6 compute boxes, all reasonably beefy | 15:26 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh master: Add support for the openSUSE Leap distributions https://review.openstack.org/482334 | 15:27 |
| sxc731 | admin0: OK let's both go for it an report back in 24 hours or so? | 15:27 |
| admin0 | yeah | 15:28 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Sort key-value pairs in backend dictionaries https://review.openstack.org/483070 | 15:28 |
| admin0 | one qustion to you though | 15:28 |
| odyssey4me | logan- small tweak on https://review.openstack.org/#/c/483342/1 needed | 15:28 |
| admin0 | odyssey4me, from experience, unlike OSA which is more zero touch after setup, with ceph you need to tinker it to optimize stuff .. need to add remove osds, change parmeters etc .. so what is the limit/feature of the ceph integration in ocata ? | 15:30 |
| admin0 | where can i read about that | 15:30 |
| logan- | yup thanks odyssey4me. will edit after the integrated test finishes. every time i edit one of the dependent patches it kills the integrated test and i have to retest heh | 15:30 |
| admin0 | or is it more providing an integrated file where the ansible-ceph reads and thats it ? | 15:30 |
| odyssey4me | logan- roger that | 15:31 |
| *** udesale has quit IRC | 15:31 | |
| odyssey4me | admin0 we're not implementing anything other than what ceph-ansible deploys | 15:31 |
| *** foutatoro has quit IRC | 15:31 | |
| odyssey4me | all we've done is add the playbooks and inventory bits so that it can be done | 15:32 |
| admin0 | ok .. clear now | 15:32 |
| odyssey4me | as you've suggested, the best practises are very opinionated based on how you're using the cluster so it's better not to try and codify those in - one size does not fit all | 15:32 |
| *** winggundamth has quit IRC | 15:33 | |
| *** acormier_ has joined #openstack-ansible | 15:35 | |
| firebat | Ok guys I found out my issue. For some reason the playbook doesn't seem to be getting the correct login parameters for keystone admin when it runs the os_glance role service task. Should I modify the task and try and get more output out of it to see what args it is receiving? With -vvv I don't see them | 15:38 |
| *** pcaruana has quit IRC | 15:38 | |
| firebat | Sorry if this is a dumb question I'm a bit of an ansible noob | 15:38 |
| odyssey4me | firebat hmm, that's odd | 15:38 |
| odyssey4me | are you changing anything up in user_variables.yml ? are your secrets populated in user_secrets.yml ? | 15:39 |
| *** acormier has quit IRC | 15:39 | |
| asura | It is safe to add openstack_service_publicuri_proto: http in user_variables.yml? | 15:41 |
| odyssey4me | asura yes of course, it's one of the example configs in the deploy guide | 15:42 |
| asura | Decided to just tear down and rebuild my cluster OSA test build again. Hopefully, I'll make it through today. | 15:42 |
| asura | Wish I had a redo.sh like admin0 :) Unfortunately, I have to clone all my vm's | 15:43 |
| openstackgerrit | Merged openstack/openstack-ansible-os_ceilometer master: test: test-install-mongodb: Do not mask mongodb_user failures https://review.openstack.org/482310 | 15:43 |
| openstackgerrit | Merged openstack/openstack-ansible-os_ceilometer master: tests: Provide mongodb configuration template https://review.openstack.org/482311 | 15:43 |
| admin0 | asura, my scripts are all documented and available on github .. blog is at openstackfaq.co | 15:44 |
| admin0 | * .com | 15:44 |
| asura | Yeah, I've read it | 15:44 |
| asura | Loved to point to VyOS | 15:44 |
| *** weezS has quit IRC | 15:44 | |
| admin0 | vyos allowed me to simulate 2 servers as 2 datacenters and do some cool tunneling in between | 15:44 |
| admin0 | mikrotik also works though | 15:45 |
| *** thorst has quit IRC | 15:45 | |
| asura | I'm using virtual box with Internal network "intnet" attached to the VMs. Networking appears to work. | 15:45 |
| asura | with the VyOS setup | 15:45 |
| *** thorst has joined #openstack-ansible | 15:46 | |
| asura | I suppose if it all works I could covert it to a Vagrant box for faster builds | 15:46 |
| *** lucasxu has joined #openstack-ansible | 15:49 | |
| *** thorst has quit IRC | 15:50 | |
| *** sxc731 has quit IRC | 15:51 | |
| boxrick1 | Hello, I am attempting to use a variant of what you guys use ( basically the LXD connection type rather than SSH ). I know you use a connection plugin and catch SSH then deploy via the host directly using lxc attach. | 15:53 |
| boxrick1 | I am just wondering, speed wise does this seem much slower? | 15:53 |
| *** vnogin has quit IRC | 15:53 | |
| boxrick1 | Since when I use the LXD connection plugin it is basically 3 minutes to gather facts about a host vs about 3 seconds via SSH directly. | 15:53 |
| *** armaan has joined #openstack-ansible | 15:53 | |
| odyssey4me | ouch | 15:53 |
| boxrick1 | Yea, its basically unusable and is causing me a whole amount of pain :/ | 15:54 |
| DimGR | where is the redo.sh admin0 on your page? can't find it | 15:54 |
| odyssey4me | I guess the lxd connection plugin might be gathering a bunch of extra facts, whereas the ssh plugin is not? | 15:55 |
| admin0 | there is a recreate.sh :D | 15:55 |
| odyssey4me | although we're specifically scoping the facts gathered down to a smaller set - have you done that? | 15:55 |
| admin0 | i will update the blog tonight with ocata | 15:55 |
| asura | Is there a reason you use library on your faq rather than octa? | 15:55 |
| asura | oh i see | 15:56 |
| boxrick1 | Everything runs slower in general, the gather facts was just an example. | 15:56 |
| asura | I might try that tomorrow if I don't get this working lol | 15:56 |
| admin0 | when i started with osa, it was that.. and then got super busy .. but now have taken steps to give time to the site and make it work .. but dedicated for production environments | 15:58 |
| asura | We'll honestly it was the best documentation I could find on OSA besides the online guide | 15:58 |
| * asettle stumbles in | 15:58 | |
| *** acormier_ has quit IRC | 15:58 | |
| asettle | Tell me more about how great the docs are | 15:58 |
| admin0 | asettle \o | 15:59 |
| asettle | HEy admin0 :) | 15:59 |
| *** armaan has quit IRC | 15:59 | |
| admin0 | well, it was my docs on openstackfaq.com he was talking about | 15:59 |
| asura | Well the online docs have a bug I submitted | 15:59 |
| *** acormier has joined #openstack-ansible | 15:59 | |
| asettle | Ohhhh wellllllllll sureeee :p | 15:59 |
| jamesden_ | asettle Would be nice if Google cache were updated... :( | 15:59 |
| asettle | I just hear "docs" and appear | 15:59 |
| asettle | jamesden_: I wish I worked magic, but alas... | 15:59 |
| admin0 | asura, well, the docs in osa are by the developers for the developers .. i am more an operator ( and a lazy dev) .. | 15:59 |
| *** armaan has joined #openstack-ansible | 15:59 | |
| asettle | admin0: hey, c'mon, we wrote a neat as ops guide | 15:59 |
| admin0 | i have a different view of the docs and guides | 15:59 |
| odyssey4me | boxrick1 unfortunately I haven't tried working with it just yet | 15:59 |
| spotz | meeting now!!!!! | 16:00 |
| asettle | And by 'we', I mean, the ops team | 16:00 |
| asura | There is an ops team? | 16:00 |
| admin0 | i am an unoffical ops validator :) | 16:00 |
| asettle | asura: Rackspace's ops team ;) since we use OSA, I whipped them into shape | 16:00 |
| odyssey4me | admin0 actually the deploy guide is for operators, as is the ops guide | 16:01 |
| asettle | Turned their knowledge into goodness | 16:01 |
| odyssey4me | the contributor/developer docs are for developers | 16:01 |
| admin0 | the docs are getting better now | 16:02 |
| asettle | "are getting better now" yo bro c'mon | 16:02 |
| asettle | :P | 16:02 |
| *** acormier has quit IRC | 16:03 | |
| admin0 | asettle, https://docs.openstack.org/project-deploy-guide/openstack-ansible/ocata/app-ceph.html :D | 16:04 |
| asettle | admin0: mostly just keen for ceph? :P | 16:04 |
| *** kristia__ has joined #openstack-ansible | 16:04 | |
| *** lucasxu has quit IRC | 16:04 | |
| *** sufyan68 has quit IRC | 16:04 | |
| admin0 | i am an operator .i want to utilize all the good work this team does in developing .. and its in their head .. so for example, there are 3 people now ( including me ) interested in getting ceph+osa runnign, and that docs is the only doc it is out there | 16:05 |
| admin0 | so trying to help there to get it better | 16:05 |
| admin0 | but if the devs give a bit more extra lines, it will help get started .. | 16:05 |
| *** schwicht has quit IRC | 16:06 | |
| admin0 | i have to run .. 6 PM ..office closing down .. will check back from home | 16:06 |
| *** agrebennikov has joined #openstack-ansible | 16:07 | |
| *** kristian__ has quit IRC | 16:07 | |
| andymccr | admin0: cool :) | 16:07 |
| asura | I'm willing to test the ceph+octa as well | 16:08 |
| asura | I have a ceph cluster up | 16:08 |
| *** dxiri has joined #openstack-ansible | 16:08 | |
| *** schwicht has joined #openstack-ansible | 16:08 | |
| *** gillesMo has quit IRC | 16:10 | |
| *** admin0 has quit IRC | 16:11 | |
| *** sxc731 has joined #openstack-ansible | 16:12 | |
| *** dxiri has quit IRC | 16:14 | |
| *** weezS has joined #openstack-ansible | 16:18 | |
| *** jamielennox has quit IRC | 16:19 | |
| kristia__ | andymccr: is there anything, how I could fix dns in lxcbr0? | 16:20 |
| andymccr | kristia__: so is the resolv.conf inside the containers pointing at 10.255.255.1 | 16:20 |
| andymccr | and can they reach the dnsmasq service? | 16:21 |
| kristia__ | it is, and I can successfully ping it | 16:21 |
| *** lucasxu has joined #openstack-ansible | 16:21 | |
| DimGR | kristia__ lxcbr0 down ; lxcbr0 up | 16:21 |
| *** openstacking_123 has joined #openstack-ansible | 16:22 | |
| kristia__ | did that, but gonna do it again (but did it ifdown and ifup) it might not be different or is it? | 16:22 |
| odyssey4me | logan- wow, that test build is only at neutron | 16:22 |
| openstacking_123 | Anyone else have trouble when linux bridge routers our in HA mode? | 16:22 |
| openstacking_123 | are* | 16:23 |
| odyssey4me | for some reason the jobs seem to be going very, very slowly this afternoon | 16:23 |
| kristia__ | DimGR: ifdown lxcbr0; ifup lxcbr0 right? | 16:23 |
| DimGR | yes | 16:24 |
| kristia__ | DimGR: then thats done, now restart the container? | 16:25 |
| *** dxiri has joined #openstack-ansible | 16:25 | |
| kristia__ | now it magically works, I did it many times before and it refused to ping google | 16:26 |
| DimGR | not needed | 16:26 |
| DimGR | :) | 16:26 |
| kristia__ | DimGR: it shouldnt be the problem that I did ifdown lxcbr0 and then ifup lxcbr0 and restart the container. Not restart the lxcbr0 in one line. Or is it? | 16:28 |
| asura | Why do I need internal_lb_vip_address and external_lb_vip_address ip addresses inside openstack_user_config.yml? So they run on separate hosts? | 16:28 |
| openstacking_123 | I get error 'Keepalived_vrrp[11377]: Netlink: filter function error' if I add a port into an ha linux bridge router. Then external IP on the router will go down for 3 minutes or so | 16:29 |
| asura | Sorry, I mean why do they need to be on separate ip addresses? | 16:30 |
| odyssey4me | kristia__ maybe, because it might need the time to properly close up the dnsmasq service | 16:30 |
| *** dxiri has quit IRC | 16:30 | |
| jamesden_ | asura Because https is enabled on external, and http on internal, and if you're using the same listener address it may cause issues for the client and/or internal service. | 16:32 |
| *** dxiri has joined #openstack-ansible | 16:32 | |
| jamesden_ | it's better to split them. they can even be addresses in the same subnet, but to avoid those types of issues it's best to define two unique addrs | 16:32 |
| odyssey4me | wel,, it physically cannot listen on the same address and port for both http and https | 16:33 |
| *** skape has quit IRC | 16:35 | |
| kristia__ | odyssey4me: deploying, fingers crossed | 16:36 |
| asura | kristia__ best of luck | 16:36 |
| kristia__ | thanks | 16:36 |
| kristia__ | install server packages no errors :D | 16:37 |
| kristia__ | at least, that was my todays bottleneck | 16:37 |
| *** schwicht has quit IRC | 16:38 | |
| kristia__ | currently on galera, should be a smooth ride | 16:38 |
| kristia__ | also do all services in /opt/openstack-ansible/etc/openstack_deploy/conf.d/*.yml.aio work? | 16:39 |
| logan- | i know odyssey4me i was noticing that too :( | 16:40 |
| logan- | it was just finishing nova with 10 mins left to timeout | 16:40 |
| *** gouthamr has quit IRC | 16:41 | |
| *** gouthamr has joined #openstack-ansible | 16:41 | |
| sc68cal | spotz: you should update your script or whatever that pings people, since I am no longer on that list https://wiki.openstack.org/wiki/Meetings/openstack-ansible | 16:41 |
| spotz | sc68cal: Yeah it's a few months old I'll admit | 16:42 |
| sc68cal | no worries | 16:42 |
| *** firebat has quit IRC | 16:43 | |
| *** dxiri has quit IRC | 16:44 | |
| odyssey4me | logan- the integrated job has passed for ceph, so I think the roles are good to go | 16:46 |
| *** thorst has joined #openstack-ansible | 16:46 | |
| odyssey4me | the role changes I mean | 16:46 |
| logan- | agreed | 16:46 |
| logan- | it looks like it all works | 16:46 |
| odyssey4me | I did notice a bug or two in the integrated build patch though | 16:46 |
| *** dxiri has joined #openstack-ansible | 16:46 | |
| logan- | we may want to reno the integrated build due to the var changes | 16:46 |
| odyssey4me | hmm, well - the vars aren't changing though - are they? | 16:47 |
| *** mbuil has quit IRC | 16:47 | |
| logan- | cinder_backends_rbd_inuse -> nova_cinder_rbd_inuse | 16:47 |
| logan- | i doubt cinder_backends_rbd_inuse is commonly overridden though | 16:48 |
| logan- | but who knows | 16:48 |
| odyssey4me | ah yes, makes sense to reno then | 16:48 |
| logan- | yea | 16:48 |
| logan- | i will update the integrated patch but we need to hold the role changes until the integrated is almost ready to go otherwise it will break the gate for a while. | 16:50 |
| logan- | the nova var namespacing change I just mentioned will break the ceph build until the integrated patch merges | 16:51 |
| odyssey4me | alternatively, we could do an interim state in the nova role to cater for both the old and new var, then remove the old var once the integrated patch merges | 16:52 |
| logan- | true.. | 16:55 |
| odyssey4me | odd, that one keystone job is stuck at creating containers | 16:55 |
| odyssey4me | it's been stuck there for almost an hour | 16:56 |
| *** weezS has quit IRC | 16:57 | |
| *** shardy has quit IRC | 16:58 | |
| *** dxiri has quit IRC | 16:59 | |
| *** dxiri has joined #openstack-ansible | 17:02 | |
| *** sxc731 has quit IRC | 17:03 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_ceilometer master: Add support for the openSUSE Leap distributions https://review.openstack.org/482312 | 17:08 |
| odyssey4me | hmm, it looks like the stuff using external downloads is causing the massive slowdown | 17:08 |
| odyssey4me | sometimes it just fails, sometimes it just hangs | 17:08 |
| *** stuartgr has quit IRC | 17:08 | |
| odyssey4me | it seems to mostly be happening with the lxc base cache download | 17:09 |
| openstackgerrit | Merged openstack/openstack-ansible-os_designate master: tests: Convert bind configuration file to template https://review.openstack.org/482490 | 17:11 |
| *** firebat has joined #openstack-ansible | 17:12 | |
| firebat | Hey guys so I tried pushing in correct variables to make sure the attempted authorization with keystone was using the correct information, but I'm still at a loss. I have verified via manually curling that I can get tokens as the admin user so there's that. Here's the error output I'm getting (it's a pastebin just shortened since I'm on a jump box): goo.gl/rZeZ74 | 17:21 |
| *** schwicht has joined #openstack-ansible | 17:23 | |
| *** electrofelix has quit IRC | 17:23 | |
| odyssey4me | firebat I didn't see an answer to my earlier questions | 17:23 |
| firebat | Could you repost I lost connection that's why I'm on a jump box now. I don't have an irc proxy at work | 17:23 |
| odyssey4me | http://eavesdrop.openstack.org/irclogs/%23openstack-ansible/%23openstack-ansible.2017-07-13.log.html#t2017-07-13T15:39:15 | 17:23 |
| firebat | Ahh thanks much | 17:24 |
| firebat | odyssey4me The secretes file is completely populated, and yes I do have some things in my user_variables.yml file I'll paste it in here. | 17:25 |
| *** firebat_paste has joined #openstack-ansible | 17:26 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_designate master: Add support for the openSUSE Leap distributions https://review.openstack.org/482491 | 17:26 |
| *** kristia__ has quit IRC | 17:28 | |
| firebat_paste | https://pastebin.com/e60zVtTd | 17:28 |
| *** kristian__ has joined #openstack-ansible | 17:28 | |
| firebat | odyssey4me There's not much in there as I'm just running a dev/test deploy right now, and I'm not sure how much stuff should be in there anyway | 17:29 |
| odyssey4me | the openstack_service settings are all defaults, so setting those is unnecessary | 17:29 |
| odyssey4me | right, so you have a default environment with debug enabled - that's fine | 17:30 |
| odyssey4me | you should not have to set anything unless you want to change default behaviour | 17:30 |
| *** ricardoas has joined #openstack-ansible | 17:30 | |
| odyssey4me | right - what else have you changed and where? | 17:30 |
| odyssey4me | and what tag/branch are you using? | 17:30 |
| firebat_paste | I'm on ocata/stable with your one change | 17:31 |
| firebat_paste | For the keystone service not being reset or whatever | 17:31 |
| firebat_paste | I can get the commit if you need | 17:31 |
| odyssey4me | ok, and what else? | 17:31 |
| odyssey4me | any other changes? | 17:32 |
| odyssey4me | can you pastebin/gist your openstack_user_config/conf.d file info? | 17:32 |
| odyssey4me | obviously sanitised | 17:32 |
| *** kristian__ has quit IRC | 17:32 | |
| *** toddnni has joined #openstack-ansible | 17:33 | |
| firebat_paste | Hmmm I think I may have made a mistake? All my stuff is in openstack_user_config.yml | 17:34 |
| odyssey4me | well, it depends on what's in there - you can send me a gist link privately if you'd prefer | 17:35 |
| odyssey4me | but yes, if all your behaviour change settings are in there, then it won't work because the inventory is a lower precedence | 17:36 |
| odyssey4me | all that should be in there are the network mappings, host:group mappings, and any host-specific vars (like cinder backends) | 17:36 |
| *** fxpester has quit IRC | 17:37 | |
| firebat_paste | Yeah I believe that's how I have it setup | 17:38 |
| firebat_paste | I sent you the private gist | 17:38 |
| odyssey4me | ok, so I see the problem | 17:39 |
| odyssey4me | you're using the same IP for both internal and external VIP's | 17:39 |
| odyssey4me | if you do that, you *have* to set the public proto to http | 17:39 |
| odyssey4me | the same IP cannot bind http and https on the same ip:port | 17:40 |
| firebat_paste | Interesting | 17:40 |
| firebat_paste | That makes a ton of sense | 17:40 |
| firebat_paste | Let me run with proto http and see if I can get a complete deploy then I can circle back around and fix my http vs https issue | 17:40 |
| odyssey4me | that's detailed in https://docs.openstack.org/project-deploy-guide/openstack-ansible/ocata/app-config-test.html | 17:41 |
| odyssey4me | specifically in https://docs.openstack.org/project-deploy-guide/openstack-ansible/ocata/app-config-test.html#user-variables | 17:41 |
| *** toddnni has quit IRC | 17:42 | |
| *** toddnni has joined #openstack-ansible | 17:44 | |
| *** schwicht has quit IRC | 17:45 | |
| odyssey4me | firebat_paste you'll likely have to build from scratch - or at least dump the db's - because the endpoints will all be configured in a way that can't be used | 17:47 |
| firebat_paste | Yeah looks to be the case | 17:48 |
| openstackgerrit | Merged openstack/openstack-ansible master: Set python interpreter for connection:local plays https://review.openstack.org/482090 | 17:53 |
| openstackgerrit | Merged openstack/openstack-ansible master: Restart glance services only when necessary https://review.openstack.org/482727 | 17:53 |
| *** tomtomtom has joined #openstack-ansible | 17:54 | |
| tomtomtom | anyone here know if the octavia apiv2 has been pushed into the playbooks yet? | 17:55 |
| odyssey4me | huzzah, with https://review.openstack.org/482090 in we might get an upgrade success tomorrow :) jmccrory | 17:55 |
| odyssey4me | tomtomtom you'll need to ask xgerman_ | 17:55 |
| jmccrory | awesome! | 17:55 |
| tomtomtom | ok thanks | 17:55 |
| xgerman_ | tomtomtom no | 17:56 |
| odyssey4me | jmccrory are you going to backport https://review.openstack.org/482727 ? I think it's necessary to do so asap | 17:56 |
| odyssey4me | should probably do https://review.openstack.org/483070 too | 17:56 |
| xgerman_ | it’s on my list of though. But first need to finish the scenario test | 17:56 |
| tomtomtom | thanks of the quick reply, it's appreciated | 17:56 |
| firebat | odyssey4me Just to clarify I don't have to change the deploying host right just the ones I deploy to? | 17:56 |
| odyssey4me | firebat I'm not sure I understand what you mean. | 17:57 |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible stable/ocata: Restart glance services only when necessary https://review.openstack.org/483487 | 17:57 |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible stable/newton: Restart glance services only when necessary https://review.openstack.org/483488 | 17:57 |
| *** weezS has joined #openstack-ansible | 17:57 | |
| tomtomtom | @xgerman any idea when it might be? | 17:57 |
| jmccrory | think there may be a similar issue with nova placement | 17:57 |
| firebat | odyssey4me I'm deplyoing from a VM where the openstack-ansible and all my customizations are... it doesn't have any state information in there that could get in the way of a deploy does it? | 17:58 |
| odyssey4me | firebat nope, except perhaps the fact cache which you can nuke | 17:58 |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible-os_cinder stable/ocata: Sort key-value pairs in backend dictionaries https://review.openstack.org/483489 | 17:58 |
| xgerman_ | tomtomtom before the Pike release :-) | 18:00 |
| tomtomtom | ok thanks | 18:00 |
| *** sxc731 has joined #openstack-ansible | 18:08 | |
| *** dmsimard is now known as dmsimard|afk | 18:08 | |
| *** schwicht has joined #openstack-ansible | 18:09 | |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible-os_cinder stable/newton: Sort key-value pairs in backend dictionaries https://review.openstack.org/483495 | 18:13 |
| *** toddnni has quit IRC | 18:14 | |
| *** kristian__ has joined #openstack-ansible | 18:15 | |
| *** kristian__ has quit IRC | 18:18 | |
| *** schwicht has quit IRC | 18:18 | |
| *** firebat_paste has quit IRC | 18:21 | |
| *** acormier has joined #openstack-ansible | 18:23 | |
| *** cshen_ has joined #openstack-ansible | 18:25 | |
| *** weezS has quit IRC | 18:26 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_glance master: Include ceph_client role if needed https://review.openstack.org/483342 | 18:31 |
| *** armaan has quit IRC | 18:32 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_nova master: Include ceph_client role if needed https://review.openstack.org/483061 | 18:35 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 18:36 |
| *** jamielennox has joined #openstack-ansible | 18:37 | |
| *** pcaruana has joined #openstack-ansible | 18:38 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 18:38 |
| *** kristian__ has joined #openstack-ansible | 18:41 | |
| *** toddnni has joined #openstack-ansible | 18:44 | |
| *** openstacking_123 has quit IRC | 18:44 | |
| *** thorst has quit IRC | 18:44 | |
| *** openstacking_123 has joined #openstack-ansible | 18:49 | |
| *** armaan has joined #openstack-ansible | 18:52 | |
| *** openstacking_123 has quit IRC | 18:54 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-galera_server master: Separate greenfield and upgrade tests https://review.openstack.org/483507 | 18:54 |
| odyssey4me | hwoarang andymccr ^ I think it's about time we do that. | 18:55 |
| *** dcdamien has joined #openstack-ansible | 19:00 | |
| *** dcdamien has quit IRC | 19:01 | |
| *** dcdamien has joined #openstack-ansible | 19:01 | |
| asura | jamesden_ I reloading three fresh nodes today and adjusted the internal_lb_vip_address and external to 172.29.236.11 and 172.29.236.12. Upon running, openstack-ansible setup-infrastructure.yml task pip_install fails with unable to connect to http://172.29.236.11:8181 https://pastebin.com/DD7mueC7 | 19:01 |
| asura | When I lxc-attach -n deploy-galera-container-8bc3d4ec, I'm able to ping 172.29.236.11 https://pastebin.com/WVASHXn3 -- I was under the impression that eth0 allows the containers access to the mgmt vlan. Q: Should deploy host (running ansible playbook) launch a service on port 8181 at 172.29.236.11 for the deploy-galera-container-8bc3d4ec container to connect to or is there something else happening? | 19:01 |
| asura | I can see no service (assuming it should be apache2) is running on 172.29.236.11:88 on host deploy. Does the load balancer or something else forward this connection to another container? How can I better understand where this is going wrong? Thanks. | 19:01 |
| asura | Reference: deploy openstack_user_config.yml https://pastebin.com/zWhNkPCX nic config https://pastebin.com/qFA4cYaX target host nic config https://pastebin.com/nD9RPr3w | 19:01 |
| jamesden_ | sure, lemme take a look | 19:02 |
| asura | Thanks | 19:02 |
| *** schwicht has joined #openstack-ansible | 19:05 | |
| asura | Meant to say I see no service is running on 172.29.236.11:8181 | 19:05 |
| *** dcdamien has quit IRC | 19:05 | |
| *** dcdamien has joined #openstack-ansible | 19:05 | |
| jamesden_ | Is 172.29.236.11 defined as your external or internal vip? | 19:05 |
| asura | internal | 19:06 |
| jamesden_ | and you're using haproxy? | 19:06 |
| dcdamien | Does anybody here know why this was added? https://git.openstack.org/cgit/openstack/openstack-ansible-os_cinder/commit/?id=4fb0bb48877f579b0d28e6ea18896f949df738e7 | 19:06 |
| dcdamien | As I know open-iscsi doesn't work in container | 19:07 |
| odyssey4me | dcdamien see the discussion in https://review.openstack.org/#/c/461133/ | 19:07 |
| asura | there is a section in openstack_user_config.yml defining # load balancer haproxy_hosts: deploy: ip: 172.29.236.11 | 19:07 |
| asura | So, I assume deploy node is acting as a load balancer? | 19:08 |
| asura | I've never dealt with load balancers, so I'm a bit confused by what is going on with the networking | 19:08 |
| odyssey4me | asura the load balancer is running on whichever node you told it to | 19:09 |
| odyssey4me | in that case, on your deploy node, yes | 19:09 |
| jamesden_ | asura 172.29.236.11:8181 should be configured as a VIP on the deploy host or whatever host is defined at the haproxy host. Check to see that it's configured on the br-mgmt interface. You can use netstat to ensure its listening on 8181. From that haproxy/deploy node, you can try to curl 172.29.236.11:8181 and see if you get a response. curl -v http://172.29.236.11:8181 and you should get a 200 OK and some dirs | 19:09 |
| asura | Yeah, I think I tried that is it got nothing | 19:10 |
| asura | I'll double check | 19:10 |
| jamesden_ | if you get a timeout or connection refused, then either a) haproxy isnt actually running or b) the backend pool member is down, in this case, the repo container is failing the monitor | 19:10 |
| asura | Its a connection refused | 19:11 |
| jamesden_ | on that haproxy node what is the output of 'netstat -an | grep :8181' | 19:11 |
| jamesden_ | and systemctl status haproxy | 19:12 |
| asura | deploy systemd[1]: haproxy.service: Start request repeated too quickly | 19:13 |
| asura | deploy systemd[1]: Failed to start HAProxy Load Balancer. | 19:13 |
| jamesden_ | well that's a start. /var/log/haproxy.log may have some insight | 19:14 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/newton: Sort key-value pairs in backend dictionaries https://review.openstack.org/483495 | 19:14 |
| *** kristian__ has quit IRC | 19:14 | |
| asura | I get a bunch on cannot bind errors to deploy haproxy-systemd-wrapper[1905]: [ALERT] 193/141307 (1909) : Starting frontend neutron_server-front-1: cannot bind socket [172.29.236.12:9696] | 19:14 |
| asura | on various ports | 19:15 |
| jamesden_ | ok. try configuring that address on br-mgmt, too | 19:15 |
| *** kristian__ has joined #openstack-ansible | 19:15 | |
| asura | My target node is using that address though, won't that conflict? | 19:15 |
| jamesden_ | yes, it will. try and find an address that won't conflict with any other node. I suggested .12 hoping it would be free | 19:16 |
| jamesden_ | update openstack_user_config.yml with the new addr you choose | 19:16 |
| jamesden_ | and you'll need to rerun haproxy-install.yml | 19:16 |
| asura | Thanks James | 19:16 |
| asura | I'll give it a try | 19:16 |
| jamesden_ | and may need to configure the addr on br-mgmt incase it doesnt do it, and maybe restart the service | 19:16 |
| asura | I hope you have a great day | 19:16 |
| jamesden_ | sure | 19:16 |
| jamesden_ | thx | 19:16 |
| jamesden_ | :) | 19:16 |
| dcdamien | @odysseycan I have tried today running os-cinder-install playbook but it fails when installing open-iscsi | 19:17 |
| dcdamien | https://bugs.launchpad.net/openstack-ansible/+bug/1704161 | 19:17 |
| openstack | Launchpad bug 1704161 in openstack-ansible "open-iscsi post-installation script fails in cinder-api-container" [Undecided,New] | 19:17 |
| *** jvidal has quit IRC | 19:19 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-galera_server master: Separate greenfield and upgrade tests https://review.openstack.org/483507 | 19:19 |
| odyssey4me | dcdamien hmm, that's odd - especially odd because we're not seeing that in any gating | 19:22 |
| *** acormier_ has joined #openstack-ansible | 19:24 | |
| *** acormier_ has quit IRC | 19:24 | |
| *** acormier_ has joined #openstack-ansible | 19:24 | |
| dcdamien | I tried to re-run this playbook after pulling 15.1.6 over 15.1.5.I can check it after recreating those containers. | 19:25 |
| openstackgerrit | German Eichberger proposed openstack/openstack-ansible master: [WIP] Octavia scenario https://review.openstack.org/483520 | 19:25 |
| *** acormier has quit IRC | 19:26 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Only gather facts when necessary https://review.openstack.org/483003 | 19:27 |
| *** dcdamien has quit IRC | 19:28 | |
| *** dcdamien has joined #openstack-ansible | 19:29 | |
| asura | So, if I have nodes deploy br-mgmt 172.29.236.11 target 172.29.236.12 storage 172.29.236.13; then I should configure br-mgmt on node deploy with a second ip address 172.29.236.14 to be used by the external_lb_vip_address address? or would it better to make a new host and assign it something like: haproxy_hosts: haproxyhost: ip: 172.29.236.14 inside openstack_user_config.yml? I'm uncertain if a br-mgmt on deploy can | 19:31 |
| *** kristian__ has quit IRC | 19:32 | |
| jamesden_ | configuring .14 as the external vip on the deploy host is fine | 19:32 |
| asura | openstack-ansible haproxy-install.yml /opt/openstack-ansible/playbooks/inventory/dynamic_inventory.py generate.MultipleIpForHostError: Host deploy has both 172.29.236.11 and 172.29.236.14 assigned | 19:35 |
| jamesden_ | one sec | 19:36 |
| *** chhavi has quit IRC | 19:38 | |
| *** kristian__ has joined #openstack-ansible | 19:39 | |
| jamesden_ | can you post up your openstack_user_config.yml? | 19:40 |
| *** cathrichardson has quit IRC | 19:40 | |
| *** cathrichardson has joined #openstack-ansible | 19:41 | |
| *** cshen_ has quit IRC | 19:41 | |
| dcdamien | @odyssey4me unfortunately not working even after recreating containers -> https://gist.github.com/anonymous/97117b8a7e994f26b1025c88552dfbe4 | 19:42 |
| *** kristian__ has quit IRC | 19:43 | |
| odyssey4me | dcdamien odd, are you using a special apt mirror or are there any other customisations that may come into account? | 19:43 |
| odyssey4me | I've deployed a few environments today in testing and have not seen that issue. | 19:43 |
| odyssey4me | can you post up any special config related to cinder services in the bug? | 19:44 |
| odyssey4me | perhaps things like whether it's in a container or not in your environment layout | 19:44 |
| *** kristian__ has joined #openstack-ansible | 19:45 | |
| jamesden_ | asura I just tested what i think is the same configuration without error | 19:45 |
| *** SerenaFeng has quit IRC | 19:46 | |
| *** SerenaFeng has joined #openstack-ansible | 19:47 | |
| *** SerenaFeng has quit IRC | 19:49 | |
| *** lucasxu has quit IRC | 19:50 | |
| *** pcaruana has quit IRC | 19:52 | |
| dcdamien | @odyssey4me sure, I can. Yep - it's in container, not metal. I'm using standard ubuntu-cloud repo. I have some customisations but looks unrelated. | 19:52 |
| odyssey4me | hmm, ok so that's the difference from the gating | 19:54 |
| odyssey4me | add that to the bug please | 19:54 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/ocata: Sort key-value pairs in backend dictionaries https://review.openstack.org/483489 | 19:59 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-galera_server master: Separate greenfield and upgrade tests https://review.openstack.org/483507 | 19:59 |
| *** dmsimard|afk is now known as dmsimard | 20:00 | |
| *** cshen_ has joined #openstack-ansible | 20:01 | |
| asura | jamesden_ https://pastebin.com/sVenYwvn sorry for the delay | 20:01 |
| jamesden_ | asura I would correct this: external_lb_vip_address: 172.29.236.12. Make it .14. And remove this stanza: haproxy_hosts | 20:05 |
| asura | opps :) | 20:06 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 20:15 |
| asura | well same error on TASK: pip_install : Install pip packages | 20:16 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 20:16 |
| jamesden_ | you may need to start haproxy by hand | 20:16 |
| jamesden_ | and check the log to make sure it started. a 'netstat -an | grep :8080' should show two listeners | 20:17 |
| asura | deploy haproxy-systemd-wrapper[20560]: [ALERT] 193/151841 (20562) : Starting frontend rabbitmq_mgmt-front-1: cannot bind socket [172.29.236.14:15672] | 20:19 |
| asura | https://pastebin.com/G8MwFgfT | 20:19 |
| jamesden_ | is 172.29.236.14 still bound to an interface? | 20:19 |
| *** cathrichardson has quit IRC | 20:20 | |
| jamesden_ | You can try ip addr add 172.29.236.14/32 dev br-mgmt | 20:20 |
| asura | ok its up | 20:20 |
| *** smatzek has quit IRC | 20:21 | |
| asura | feel like an idiot | 20:21 |
| asura | thanks so much | 20:21 |
| jamesden_ | join the club | 20:21 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Test containerised cinder-volume https://review.openstack.org/483541 | 20:21 |
| asura | So Haproxy runs despite the stanza | 20:21 |
| asura | guess after this is up i need to learn ansible so i can understand the backend more | 20:22 |
| jamesden_ | i guess it defaults to the deploy node, but if you defined haproxy_hosts it would've installed there, instead | 20:22 |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone master: Ensure that keystone restarts after db sync https://review.openstack.org/483309 | 20:23 |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone stable/ocata: Replace db sync check with alternative process https://review.openstack.org/483033 | 20:23 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_keystone stable/newton: Replace db sync check with alternative process https://review.openstack.org/483034 | 20:23 |
| *** cathrichardson has joined #openstack-ansible | 20:23 | |
| *** cshen_ has quit IRC | 20:27 | |
| *** dcdamien has quit IRC | 20:31 | |
| *** sxc731 has quit IRC | 20:35 | |
| *** cshen_ has joined #openstack-ansible | 20:37 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Test containerised cinder-volume https://review.openstack.org/483541 | 20:39 |
| dmsimard | mhayden: I totally just put icanhazip.com in a playbook and I'm not ashamed | 20:46 |
| * mhayden woots | 20:46 | |
| mhayden | you can haz | 20:46 |
| *** dcdamien has joined #openstack-ansible | 20:49 | |
| *** cshen_ has quit IRC | 20:55 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance master: Include ceph_client role if needed https://review.openstack.org/483342 | 20:59 |
| openstackgerrit | Merged openstack/openstack-ansible master: Idempotent nova db privilege grants https://review.openstack.org/483176 | 21:04 |
| openstackgerrit | Merged openstack/openstack-ansible stable/newton: Restart glance services only when necessary https://review.openstack.org/483488 | 21:04 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Include ceph_client role if needed https://review.openstack.org/483394 | 21:08 |
| *** esberglu has quit IRC | 21:12 | |
| *** smatzek has joined #openstack-ansible | 21:15 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible-os_nova master: Remove temporary default used for Ceph var migration https://review.openstack.org/483560 | 21:18 |
| *** smatzek has quit IRC | 21:20 | |
| *** kylek3h has quit IRC | 21:20 | |
| *** askb has joined #openstack-ansible | 21:25 | |
| *** esberglu has joined #openstack-ansible | 21:26 | |
| *** schwicht has quit IRC | 21:38 | |
| *** thorst has joined #openstack-ansible | 21:42 | |
| *** gouthamr has quit IRC | 21:47 | |
| *** thorst has quit IRC | 21:48 | |
| *** gouthamr has joined #openstack-ansible | 21:49 | |
| *** cpuga has quit IRC | 21:52 | |
| dcdamien | @odyssey4me https://review.openstack.org/483541 <- why cinder-volume not cinder-api? | 21:54 |
| *** schwicht has joined #openstack-ansible | 22:04 | |
| *** galstrom is now known as galstrom_zzz | 22:07 | |
| *** dcdamien has quit IRC | 22:09 | |
| *** gouthamr has quit IRC | 22:10 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/ocata: Idempotent nova db privilege grants https://review.openstack.org/483594 | 22:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/newton: Idempotent nova db privilege grants https://review.openstack.org/483595 | 22:36 |
| *** schwicht has quit IRC | 22:37 | |
| *** kristian__ has quit IRC | 22:47 | |
| *** tobberydberg has joined #openstack-ansible | 22:52 | |
| *** klamath has quit IRC | 22:52 | |
| *** tobberydberg has quit IRC | 22:56 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Remove ceph_client role execution from playbooks https://review.openstack.org/483062 | 22:59 |
| *** galstrom_zzz is now known as galstrom | 22:59 | |
| *** schwicht has joined #openstack-ansible | 23:11 | |
| *** kristian__ has joined #openstack-ansible | 23:12 | |
| *** kristian__ has quit IRC | 23:17 | |
| *** charcol has joined #openstack-ansible | 23:30 | |
| *** jamesden_ has quit IRC | 23:33 | |
| *** jamesdenton has joined #openstack-ansible | 23:34 | |
| *** openstacking_123 has joined #openstack-ansible | 23:36 | |
| *** thorst has joined #openstack-ansible | 23:48 | |
| *** vishwanathj has quit IRC | 23:53 | |
| *** openstacking_123 has quit IRC | 23:53 | |
| *** vishwanathj has joined #openstack-ansible | 23:54 | |
| *** thorst has quit IRC | 23:55 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!