frickler | jrosser_: noonedeadpunk: this would be needed to merge now to fix zuul config errors, you may also want to cherry-pick to unmaintained branches https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/915525 | 05:20 |
---|---|---|
opendevreview | Dr. Jens Harbott proposed openstack/openstack-ansible-os_tempest stable/zed: Remove tripleo CI jobs https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/915942 | 05:25 |
frickler | also ^^ | 05:25 |
opendevreview | Andrew Bonney proposed openstack/openstack-ansible-os_neutron stable/2023.2: Fix multiline yaml formatting in neutron systemd services https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/915948 | 08:39 |
opendevreview | Merged openstack/openstack-ansible-os_tempest stable/zed: Remove tripleo CI jobs https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/915942 | 09:15 |
opendevreview | Merged openstack/ansible-role-python_venv_build stable/zed: Remove TripleO jobs https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/915525 | 10:09 |
opendevreview | Merged openstack/openstack-ansible master: [Feature] Add skyline deployment capability https://review.opendev.org/c/openstack/openstack-ansible/+/859446 | 10:45 |
nixbuilder | I am still battling this AIO install... just finished installing 28.0.2 (as suggested) and still getting 503 errors. So, as suggested, I looked in the haproxy config and noticed that there is no backend installed for horizon. Instead it is configured "use_backend %[path,map_reg(/etc/haproxy/base_regex.map)]" and the '/etc/haproxy/base_regex.map' is essentially empty (just a comment line). | 13:22 |
jrosser_ | nixbuilder: is this a metal or LXC deployment? | 14:07 |
nixbuilder | jrosser_: Metal | 14:07 |
nixbuilder | jrosser_: Used 'export SCENARIO='aio_metal'' | 14:08 |
jrosser_ | well, the default on a metal deploy is not to install horizon https://github.com/openstack/openstack-ansible/blob/master/tests/roles/bootstrap-host/vars/main.yml#L42-L44 | 14:10 |
jrosser_ | which may of course now not be valid, but in the past metal and horizon were mutually exclusive | 14:11 |
jrosser_ | but i beleive you could force it with `export SCENARIO='aio_metal_horizon'` | 14:12 |
jrosser_ | *metal and horizon were mutually exclusive _in the AIO_ | 14:13 |
jrosser_ | though its very likely that is no longer the case and we should be deploying horizon for metal AIO now | 14:14 |
nixbuilder | jrosser_: Well... the AIO document says "The horizon web interface provides a graphical interface for interacting with the AIO deployment. By default, the horizon API is available on port 443 of the host (or port 80, if SSL certificate configuration was disabled). As such, to interact with horizon, simply browse to the IP of the host." | 14:15 |
jrosser_ | yes but you have moved away from the default deployment of the AIO | 14:15 |
jrosser_ | which is SCENARIO=aio_lxc | 14:15 |
nixbuilder | Oh... ok. Well I will re-install using your suggested scenario :-D | 14:16 |
jrosser_ | you should be able to re-run the bootstrap_aio i think | 14:17 |
jrosser_ | if you want to do a production metal deploy with horizon then it would be worth getting it fixed | 14:17 |
jrosser_ | in the past there was insufficient care taken to bind the horizon backend only to an internal IP so port 443 was not available for the loadbalancer to use on the external IP | 14:18 |
nixbuilder | jrosser_: Cool. I'll try that | 14:18 |
jrosser_ | but i did a *ton* of work through the whole of OSA to be much more strict about port / IP bindings | 14:18 |
jrosser_ | so it should either now 1) work 2) be extremely close to working for metal+horizon | 14:19 |
jrosser_ | nixbuilder: if you get stuck again i can try to replicate it, as this should really be fixed | 14:20 |
nixbuilder | jrosser_: For my sanity sake I will do a complete re-install and let you know how it turns out. | 14:22 |
jrosser_ | ok, cool | 14:22 |
nixbuilder | jrosser_: FWIW... on the previous install using only 'export SCENARIO='aio_metal'', the process did install '/etc/openstack_deploy/conf.d/horizon.yml' and setup the dashboard_hosts. | 14:30 |
jrosser_ | hmm | 14:33 |
nixbuilder | jrosser_: Never mind... my bad. Looked at the wrong server blade. Sorry. | 14:34 |
jrosser_ | frickler: i am pretty much at a loss knowing how to unpick the mess on our unmaintained branches to cherry pick the zuul config errors you fixed earlier | 14:54 |
jrosser_ | i can't locally cherry-pick / git-review the changes because .gitreview is not updated for the new unmaintained branches | 14:55 |
jrosser_ | and that can't be merged because the buster nodeset is no longer available https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/913067 | 14:56 |
frickler | jrosser_: you'll likely need to combine all these fixes into a single commit | 15:00 |
jrosser_ | i suspect that the errors are across multiple repos | 15:00 |
jrosser_ | as the job definition referencing buster will be elsewhere again, openstack-ansible-tests most likley | 15:01 |
jrosser_ | i'll try to fix that first | 15:01 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 15:05 |
frickler | jrosser_: let me know if something needs force merging | 15:11 |
jrosser_ | frickler: the trouble is the branch renaming has made a synchronous catastrophic breakage across like two dozen repos on three branches atomically | 15:11 |
jrosser_ | there is no sequence of patches other than noop-ing the entire lot that i can see to even begin to dig any of this back to working | 15:12 |
jrosser_ | this is 1000x easier for mono or small-number repo projects where you can just make a large-ish patch to fix all the things in one go | 15:13 |
* jrosser_ frustrated | 15:17 | |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 15:28 |
nixbuilder | jrosser_: THANK YOU for your assistance... after re-installing with the 'aio_metal_horizon' scenario everything seems to work now! | 15:51 |
nixbuilder | jrosser_: :-D:-D:-D | 15:51 |
jrosser_ | nixbuilder: excellent! if you want to make a patch to remove the logic that prevents horizon deploying by default....... | 15:51 |
jrosser_ | ^ or i can do this? | 15:51 |
nixbuilder | jrosser_: I am not familiar with how to make patches for your OSA. | 15:52 |
jrosser_ | ok no problem i can do it, but there is some background on the setup you need to contribute code here https://docs.openstack.org/contributors/code-and-documentation/quick-start.html | 15:53 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Deploy horizon by default with metal AIO scenarios https://review.opendev.org/c/openstack/openstack-ansible/+/916005 | 15:56 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 16:00 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 16:13 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 16:25 |
opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-tests unmaintained/victoria: Update .gitreview for unmaintained/victoria https://review.opendev.org/c/openstack/openstack-ansible-tests/+/913070 | 16:29 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!