| jrosser | good morning | 08:52 |
|---|---|---|
| jrosser | looks like the compatibility matix is not totally accurate for centos9-stream https://docs.openstack.org/openstack-ansible/latest/admin/upgrades/compatibility-matrix.html | 09:22 |
| jrosser | perhaps we should be starting another table below the existing one for antelope onward, to clear out the old columns that will never be supported again | 09:23 |
| noonedeadpunk | o/ | 09:32 |
| noonedeadpunk | Yeah, might be we need to create archive page, but we then should be moving only maintained releases | 09:32 |
| jrosser | we could keep the old one - just the useless data > useful data now | 09:33 |
| jrosser | and it hurts my eyes :) | 09:33 |
| jrosser | we've been looking at ironic consoles more as well | 09:34 |
| jrosser | and what vars like `nova_console_type` is actually for is really a big mess | 09:35 |
| * jrosser would like to hugely simplify this | 09:36 | |
| noonedeadpunk | Btw, what is inaccuracy for c9s? | 09:36 |
| jrosser | stein/train? | 09:38 |
| noonedeadpunk | ah lol | 09:38 |
| jrosser | yeah, thats what i mean about needing to tidy the table up a bitr | 09:39 |
| jrosser | no matter how hard you look its still invisible :) | 09:39 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Fix compatability matrix wrt CentOS 9 Stream https://review.opendev.org/c/openstack/openstack-ansible/+/877940 | 10:11 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Fix compatability matrix wrt CentOS 9 Stream https://review.opendev.org/c/openstack/openstack-ansible/+/877940 | 10:22 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Fix compatability matrix wrt CentOS 9 Stream and Rocky 9 https://review.opendev.org/c/openstack/openstack-ansible/+/877940 | 10:23 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Fix compatability matrix wrt CentOS 9 Stream and Rocky 9 https://review.opendev.org/c/openstack/openstack-ansible/+/877940 | 10:29 |
| admin1 | \o | 10:32 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Split compatability matrix for legacy releases https://review.opendev.org/c/openstack/openstack-ansible/+/877941 | 10:44 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Split compatability matrix for legacy releases https://review.opendev.org/c/openstack/openstack-ansible/+/877941 | 10:48 |
| *** arxcruz is now known as arxcruz|rover | 10:57 | |
| admin1 | checking if anyone already done ovs -> ovn in prod/or have notes | 12:01 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Add documentation for multi-architecture deployments. https://review.opendev.org/c/openstack/openstack-ansible/+/877954 | 12:44 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Split compatability matrix for legacy releases https://review.opendev.org/c/openstack/openstack-ansible/+/877941 | 13:11 |
| Losraio | Hello | 15:25 |
| Losraio | After fixing some problems with the user_config.yml, I now get the following error after running the setup-infrastructure.yml: | 15:25 |
| Losraio | https://paste.openstack.org/show/bn74XZR4VS0BNZgT3wdA/ | 15:25 |
| Losraio | The setup-hosts.yml has ran successfully though | 15:26 |
| jrosser | Losraio: have you done any debugging on that yet? | 15:45 |
| Losraio | I don't even know where to begin tbh | 15:45 |
| jrosser | well me neither from those log lines :) | 15:46 |
| Losraio | I checked inside the container, there is indeed no such file | 15:46 |
| jrosser | i don't even know which playbook that is..... | 15:46 |
| Losraio | It is the setup-infrastructure | 15:46 |
| jrosser | right - but look inside that, it calls a squence of others | 15:46 |
| Losraio | I'm looking at it, but it's not clear which playbook is causing the error | 15:52 |
| Losraio | Or which playbook is called, even | 15:52 |
| jrosser | you can run them one by one yourself, very easily | 15:52 |
| jrosser | `openstack-ansible playbooks/<playbook-name.yml>` | 15:53 |
| jrosser | there is no need at all to run them using the higher level setup-*.yml in order to do specific things | 15:53 |
| Losraio | hmm | 15:53 |
| jrosser | this is very useful for applying a config change only to one service, and will save you huge amounts of time in the future | 15:53 |
| Losraio | Ah yes, here it is | 15:54 |
| Losraio | It;s the repo-setup.yml | 15:54 |
| Losraio | Idk how it came to this, I have gotten numerous errors before but none like that | 15:55 |
| jrosser | so here? https://github.com/openstack/openstack-ansible/blob/master/playbooks/repo-install.yml#L33-L36 | 15:57 |
| Losraio | Yes | 15:57 |
| Losraio | The only thing I changed before deploying from scratch is declaring a different load balancer | 15:58 |
| jrosser | so look at the error - you have `Destination /var/lib/lxc/hua-openstack-controller_repo_container-93aa6afd/config does not exist` | 15:58 |
| jrosser | is that there? | 15:58 |
| Losraio | That will have to wait | 15:59 |
| Losraio | I am running the deployment again | 15:59 |
| Losraio | And some things are changing by Ansible already, | 16:00 |
| Losraio | oh well | 16:00 |
| jrosser | did you build an all-on-one yet for testing? | 16:00 |
| Losraio | Yes, the AIO works normally | 16:00 |
| Losraio | But I need a production cloud | 16:01 |
| Losraio | Even the bare minimum one | 16:01 |
| jrosser | so you should be able to compare | 16:01 |
| jrosser | i am guessing that you do not have a repo container for some reason | 16:01 |
| Losraio | Yeah I'm on the process of doing so | 16:01 |
| jrosser | if the config file is missing it would be a surprise if the container was existing properly | 16:01 |
| Losraio | Well, it should exist without a problem, because the setup_hosts.yml ran normally | 16:02 |
| jrosser | if you have a problem in openstack_user_config, it might be missing | 16:03 |
| Losraio | Let me paste it here, but I don't think there's a problem there | 16:03 |
| Losraio | https://paste.openstack.org/show/bkfNxhIaiYHNFUBRK4ds/ | 16:04 |
| Losraio | https://paste.openstack.org/show/bPXgr1ihIYQVb09qnwl4/ | 16:05 |
| Losraio | The latter is the correct one | 16:05 |
| Losraio | Other than the fact that the haproxy_hosts configuration is dumb but functional, I don't see any other problem | 16:05 |
| jrosser | - "10.3.0.32,10.2.0.34" <- this looks like a typo | 16:06 |
| Losraio | Oh | 16:06 |
| Losraio | Let me correct it | 16:06 |
| jrosser | in fact anything in used_ips which is not in cidr_networks is unnecessary | 16:06 |
| jrosser | which is a bunch of it | 16:06 |
| Losraio | right | 16:07 |
| jrosser | there is also bad indentation after `haproxy_hosts:` | 16:07 |
| jrosser | well maybe, i'm not sure what that is | 16:07 |
| Losraio | Weid, the linter shows no such thing | 16:07 |
| jrosser | how many haproxy hosts do you want? | 16:08 |
| Losraio | Three | 16:09 |
| Losraio | I have also configured keepalived according to the docs | 16:09 |
| jrosser | right - i see you have used the same ip for a VIP as for one of the hosts | 16:10 |
| jrosser | thats not going to be great with keepalived | 16:10 |
| Losraio | Yeah | 16:10 |
| jrosser | you know that keepalived is VRRP - and how that works? | 16:11 |
| Losraio | Sort of | 16:11 |
| jrosser | i would make the `internal_lb_vip_address` unique | 16:12 |
| jrosser | it is dynamically assigned to an interface on one of the three haproxy nodes | 16:12 |
| jrosser | so if you also manually assign it to one of those i don't know whats going to happen | 16:12 |
| Losraio | right | 16:12 |
| Losraio | Oh well, it seems like I should start from scratch | 16:17 |
| jrosser | thats probably not necessary in this case, but its really worth being able to rinse/repeat deployments easily in the early days | 16:22 |
| admin1 | what is the better way to do .. 24 -> 25 and after that upgrade from ubuntu20-22, or upgrade from 20-22 first and then upgrade from 24 -25/latest tags | 16:45 |
| jrosser | admin1: best consult the compatibility matrix https://docs.openstack.org/openstack-ansible/latest/admin/upgrades/compatibility-matrix.html | 17:31 |
| noonedeadpunk | if both releases do support OS - doesn't matter | 17:33 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Add documentation for multi-architecture deployments. https://review.opendev.org/c/openstack/openstack-ansible/+/877954 | 17:43 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Add documentation for multi-architecture deployments. https://review.opendev.org/c/openstack/openstack-ansible/+/877954 | 17:54 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Add documentation for multi-architecture deployments. https://review.opendev.org/c/openstack/openstack-ansible/+/877954 | 17:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Split compatability matrix for legacy releases https://review.opendev.org/c/openstack/openstack-ansible/+/877941 | 18:47 |
| noonedeadpunk | NeilHanlon: Hey! I was wondering what were reasons behind dependency on dockerhub for rocky-container element? Or well - were there reasons not to do that like centos with dnf? | 19:07 |
| noonedeadpunk | As we're building images for our clouds, including rocky, and in isolated regions we've faced with issue that we'd need to maintain dockerhub mirror or smth to build rocky images which sounds like overkill... | 19:08 |
| NeilHanlon | doesn't need to be from docker hub | 19:09 |
| NeilHanlon | They're on quay.io, too | 19:09 |
| NeilHanlon | or, I guess... are you asking why it's built from a container vs from qcow | 19:09 |
| noonedeadpunk | Well. We don't have internet connection somewhere :) | 19:09 |
| noonedeadpunk | I think it's more from container image downloaded from somewhere vs smth like https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/vars/redhat.yml#L22 | 19:10 |
| noonedeadpunk | As centos-minimal in infra seems doing smth like that as well | 19:11 |
| NeilHanlon | yeah i can add some context... finding the Change | 19:11 |
| noonedeadpunk | and at least with lxc dnf-download approach seems working nicely for us at least... | 19:11 |
| NeilHanlon | hrm. i can't find it in the change, annoyingly. the advice/direction i got from ianw and other dib folks at the time was to go the container route instead of the dnf download/qcow2-based direction | 19:14 |
| NeilHanlon | i could swear it was recorded in a comment somewhere | 19:14 |
| noonedeadpunk | ++ I will go and bug them then :) | 19:14 |
| NeilHanlon | https://review.opendev.org/c/openstack/diskimage-builder/+/805800 was recently abandoned as https://review.opendev.org/c/openstack/diskimage-builder/+/825957 implemented the rocky-container element | 19:15 |
| NeilHanlon | ah. i found my IRC logs | 19:16 |
| NeilHanlon | https://paste.opendev.org/show/bVA79Y8NtV5m4aMShEh0/ noonedeadpunk | 19:17 |
| NeilHanlon | boils down to compatibility of DNF in ubuntu, i guess | 19:17 |
| noonedeadpunk | meh... but yes | 19:18 |
| noonedeadpunk | thanks for the context. I will go annoy ianw then :D | 19:18 |
| NeilHanlon | OTOH - i am looking to potentially use diskimage builder to generate more of our Rocky images instead of the stuff I am doing today with Anaconda... | 19:18 |
| NeilHanlon | so you might just have my ear for making it work :P | 19:18 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!