| opendevreview | Merged openstack/openstack-ansible stable/2023.1: Set correct language for docs https://review.opendev.org/c/openstack/openstack-ansible/+/939161 | 08:18 |
|---|---|---|
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [DNM] Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 08:21 |
| opendevreview | Ivan Anfimov proposed openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939203 | 08:39 |
| opendevreview | Ivan Anfimov proposed openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939204 | 08:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [DNM] Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 08:51 |
| opendevreview | Ivan Anfimov proposed openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939204 | 08:53 |
| opendevreview | Ivan Anfimov proposed openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939204 | 08:55 |
| opendevreview | Ivan Anfimov proposed openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939204 | 08:56 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 09:23 |
| jrosser | o/ morning | 09:24 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 09:25 |
| noonedeadpunk | o/ | 09:28 |
| opendevreview | Merged openstack/ansible-role-httpd master: Initial commit to the role https://review.opendev.org/c/openstack/ansible-role-httpd/+/938245 | 09:32 |
| opendevreview | Merged openstack/ansible-role-systemd_networkd master: Replace functional tests with molecule https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938517 | 09:36 |
| jrosser | should we recheck this? https://review.opendev.org/c/openstack/project-config/+/935695 | 09:38 |
| opendevreview | Merged openstack/ansible-role-systemd_networkd master: Install role pre-requisite packages https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/938560 | 09:46 |
| noonedeadpunk | jrosser: nope, governance patch is still not merged.... | 09:49 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Optimize generation of required roles/collections https://review.opendev.org/c/openstack/openstack-ansible/+/939221 | 10:43 |
| noonedeadpunk | doh, I didn't cover upgrade case in 939151 | 11:18 |
| opendevreview | Merged openstack/openstack-ansible stable/2023.1: Remove senlin/sahara/murano roles from required project https://review.opendev.org/c/openstack/openstack-ansible/+/939073 | 11:50 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 12:22 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 12:34 |
| birbilakos | Hi good folks. I'm trying to deploy octavia using the following guide: https://docs.openstack.org/openstack-ansible-os_octavia/latest/configure-octavia.html | 13:56 |
| birbilakos | Having a question about what the host_bind_override should be. | 13:57 |
| birbilakos | I have a dedicated bridge (br-lbaas) in my hosts that should serve this network. This has already vlan config set on the servers | 13:58 |
| opendevreview | Merged openstack/ansible-config_template master: Incorrect example in ansible-config_template docs https://review.opendev.org/c/openstack/ansible-config_template/+/939204 | 13:59 |
| birbilakos | I can't understand what this should be in my case: host_bind_override: "bond0" # Defines neutron physical network mapping | 14:00 |
| opendevreview | Merged openstack/openstack-ansible-tests stable/2023.1: Remove sahara from zuul required projects https://review.opendev.org/c/openstack/openstack-ansible-tests/+/939056 | 14:06 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 14:13 |
| kleini | birbilakos, you don't need to override, if you already have the br-lbaas bridges. | 14:29 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 14:55 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 14:56 |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 15:00 |
| opendevmeet | Meeting started Tue Jan 14 15:00:18 2025 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_ansible_meeting' | 15:00 |
| noonedeadpunk | #topic rollcall | 15:00 |
| noonedeadpunk | o/ | 15:00 |
| damiandabrowski | hi! | 15:04 |
| noonedeadpunk | #topic office hours | 15:05 |
| noonedeadpunk | so, it seems all current patches for molecule has been landed | 15:06 |
| noonedeadpunk | #link https://review.opendev.org/q/topic:%22osa/molecule%22 | 15:06 |
| noonedeadpunk | meaning, that only plugins repo left on the old functional tests. And we can deprecate the repo once it's done | 15:06 |
| noonedeadpunk | afaik there were some complications with running lxc inside of docker :D | 15:07 |
| noonedeadpunk | jrosser was working on that, so not sure if there are more updates | 15:07 |
| jrosser | oh i am looking at the just now | 15:08 |
| jrosser | (keep getting distracted so slooow progress) | 15:08 |
| jrosser | but i basically have lxc containers being created in docker now and am just trying to port how things work in the test repo inventory to molecule, so that lxc_containers_create will work | 15:09 |
| noonedeadpunk | btw, damiandabrowski, presence of molecule should help out while creating change for the pki role for support of vault "driver" | 15:09 |
| noonedeadpunk | but I guess you'd need to create a new scenario for that path | 15:09 |
| noonedeadpunk | jrosser: ok, amazing, sounds like really good progress | 15:10 |
| damiandabrowski | awesome! | 15:10 |
| noonedeadpunk | and we will be able to finally drop tests repo after so many years | 15:10 |
| noonedeadpunk | (drop -> deprecate) | 15:10 |
| damiandabrowski | just to let everyone know: I plan to start work on adding hashicorp vault support to ansible-role-pki in like 2 weeks | 15:10 |
| noonedeadpunk | next thing - httpd role | 15:12 |
| noonedeadpunk | #link https://review.opendev.org/q/topic:%22osa/httpd_role%22 | 15:12 |
| noonedeadpunk | while repo still not officially reconginzed by TC (https://review.opendev.org/c/openstack/governance/+/935694) | 15:13 |
| noonedeadpunk | it has all required votes to land | 15:14 |
| noonedeadpunk | then I've seen a review on https://review.opendev.org/c/openstack/openstack-ansible/+/938275 | 15:19 |
| noonedeadpunk | I'm not 100% sure, but it somehow feels being a chicken-egg rather then anything else | 15:20 |
| jrosser | that seems so | 15:23 |
| noonedeadpunk | though I do see same issue with skyline | 15:24 |
| noonedeadpunk | so I'm thinking if it would makse to disable upgrade jobs and try to land requirements | 15:28 |
| noonedeadpunk | as I ws not able to replicate issue running upgrade job locally | 15:28 |
| jrosser | that sounds sensible | 15:28 |
| noonedeadpunk | ok, will try to edit the patch and propose a follow-up with re-enabling upgrade jobs | 15:30 |
| noonedeadpunk | btw, on weekends I was working on some "healthchecks" and come up with a playbook (well, playbook and an include task for it, so more of a role) for testing network connectivity | 15:31 |
| noonedeadpunk | which I aimed to run after setup-hosts | 15:31 |
| noonedeadpunk | so it discovers hosts/networks from inventory and runs pings and iperf between hosts[0] and hosts[1:] | 15:32 |
| jrosser | MTU check would be another good thing to test | 15:32 |
| noonedeadpunk | good point | 15:32 |
| noonedeadpunk | will add that. | 15:32 |
| noonedeadpunk | and I was also thinking if it's worth adding to plugins/healtchecks or ops repo? | 15:33 |
| noonedeadpunk | or nobody cares about stuff like that ?:D | 15:33 |
| jrosser | well we have something similar already | 15:33 |
| jrosser | but i expect that not all deployments have the same interfaces to test between | 15:34 |
| noonedeadpunk | well, it's getting data from openstack_inventory | 15:34 |
| noonedeadpunk | So if there's a `container_networks` defined - it would work nicely | 15:35 |
| jrosser | we also do 'negative testing' to ensure that the different networks are not accidentally routed together | 15:35 |
| noonedeadpunk | that would be an interesting one | 15:36 |
| jrosser | but on the subject of networking | 15:36 |
| noonedeadpunk | and tricky given they could be just firewalled | 15:36 |
| jrosser | removal of linuxbridge will bring for us some giant migration need | 15:36 |
| noonedeadpunk | I was just leveraging systemd_networkd for network setup so realized that would need some kind of test to ensure that all vlans are actually reachable in expected places | 15:37 |
| jrosser | yes this would be interesting to see | 15:38 |
| noonedeadpunk | negative testing should be also doable though, except when we're talking about networks that are not defined in osa, but present on hosts | 15:38 |
| noonedeadpunk | as then it would know nothing about them | 15:38 |
| noonedeadpunk | ok, I can try to push smth (at my free time) and will be totally fine if it's rejected or abandoned | 15:39 |
| noonedeadpunk | and next probably worth starting looking into CentOS 10.... | 15:42 |
| noonedeadpunk | oh, btw | 15:43 |
| noonedeadpunk | there was an interesting ML regarding rabbitmq quorum queues | 15:43 |
| noonedeadpunk | #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/message/XRS3IQFHWX6LHYMUI4XIFPDLYIQGPYHD/ | 15:44 |
| noonedeadpunk | and sounds like we need to do some improvements/health checks there | 15:44 |
| opendevreview | Merged openstack/openstack-ansible master: Fix inventory adjustment for legacy container naming https://review.opendev.org/c/openstack/openstack-ansible/+/939110 | 15:53 |
| NeilHanlon | o/ better late than never? 🙃 | 15:57 |
| noonedeadpunk | sure thing! | 15:58 |
| noonedeadpunk | #endmeeting | 15:59 |
| opendevmeet | Meeting ended Tue Jan 14 15:59:59 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:59 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-14-15.00.html | 15:59 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-14-15.00.txt | 15:59 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2025/openstack_ansible_meeting.2025-01-14-15.00.log.html | 15:59 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 16:25 |
| jrosser | noonedeadpunk: this is WIP (it doesnt verify anything) but should give you enough to play with docker + the lxc roles | 16:25 |
| birbilakos | kleini: does that look like a valid config for octavia? I basically want to use the br-lbaas bridge which is already in the hosts | 16:28 |
| birbilakos | - network: container_bridge: "br-lbaas" container_type: "veth" container_interface: "eth14" network_interface: "br-lbaas" ip_from_q: "octavia" type: "flat" net_name: "octavia" group_binds: - neutron_linuxbridge_agent - octavia-worker - octavia-housekeeping - octavia-health-manager | 16:28 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 16:29 |
| birbilakos | syntax highlighed got messed up, here's how it should look: https://pastebin.com/rRRPh7sd | 16:30 |
| noonedeadpunk | are you doing linux bridges? | 16:32 |
| birbilakos | yes, relevant netplan config: https://pastebin.com/L1qxCqjJ | 16:35 |
| noonedeadpunk | i mean - as neutron driver | 16:35 |
| noonedeadpunk | ml2.lxb | 16:35 |
| birbilakos | ah, no, I'm using ovs | 16:35 |
| noonedeadpunk | then it should be neutron_openvswitch_agent - not neutron_linuxbridge_agent fwiw | 16:36 |
| birbilakos | got it | 16:36 |
| birbilakos | is the 'network_interface' setting correct? | 16:37 |
| birbilakos | I'm not quite sure if I need to use host_bind_override or network_interface to be honest :S | 16:44 |
| birbilakos | I reckon based on this: 'The host_bind_override override is used for LinuxBridge-based deployments, and requires a physical interface name which will then be used by the LinuxBridge agent for flat and vlan-based provider and tenant network traffic.' | 16:49 |
| birbilakos | I nned to use network_interface instead. But I'm unsure of the value | 16:49 |
| noonedeadpunk | so host_bind_override is kinda useful when you do need to define an interface for some host that doesn't use containers | 16:50 |
| noonedeadpunk | or well | 16:50 |
| noonedeadpunk | I think I;ve used that some time back for passing SR-IOV network cards into LXC containers without bridges | 16:50 |
| jrosser | i would also say that if you already have some vlan based provider networks, then don't add another flat network to neutron for lbaas | 16:50 |
| noonedeadpunk | yeah, vlan should be doing jsut fine | 16:51 |
| jrosser | just create the provider network as a vlan in neutron by picking an existing free vlan | 16:52 |
| jrosser | and also connect that up to br-lbaas on your controllers for the control plane services to connect to | 16:53 |
| jrosser | what did i miss here ? https://zuul.opendev.org/t/openstack/build/d8884a8323e74284a07586cbc646c23c | 16:56 |
| noonedeadpunk | good question | 16:57 |
| noonedeadpunk | I'd expect this being pulled by molecule | 16:57 |
| jrosser | me too and it seems ok locally | 16:58 |
| birbilakos | I'm not sure I understand :( What I have is a linux bridge in m hosts, namely br-lbaas. I want to use this bridge for octavia and not mess with vlans. Is that doable? | 16:58 |
| jrosser | birbilakos: is you external network a flat network? | 17:00 |
| noonedeadpunk | birbilakos: you need to ensure connectivity between octavia-api which runs in containers with VMs in openstack, which will be running AMphora | 17:00 |
| noonedeadpunk | that's why br-lbaas is existing | 17:00 |
| birbilakos | yes, ext is flat | 17:00 |
| jrosser | hmm ok | 17:00 |
| birbilakos | let me share the full config | 17:00 |
| noonedeadpunk | so you need to put in it a network which can be present in vms | 17:01 |
| jrosser | imho, it is never a good idea to use flat for provider networks | 17:01 |
| noonedeadpunk | (and I'm not sure you can do that with flat, unless flat is a "static" vlan) | 17:01 |
| jrosser | ^ yes you can do it with flat | 17:01 |
| birbilakos | https://pastebin.com/rvjAdBUE | 17:01 |
| noonedeadpunk | well, I kinda used flat nicely, jsut my flats were mappings to vlans | 17:01 |
| jrosser | but it requires a reconfiguratoin of literally everything each time you want to add/remove one? | 17:02 |
| birbilakos | everything besides the lbaas stuff works just fine for my env now. I cannot have vlans through the ext unfortunately | 17:03 |
| birbilakos | btw, br-ext is how my env reaches the outside world | 17:04 |
| noonedeadpunk | I wonder how this is happening as well... Shouldn't ansible_user_dir be always defined? https://zuul.opendev.org/t/openstack/build/1f550326229941f2b43b3f2a045302eb | 17:04 |
| noonedeadpunk | well. kind of - you need to add a new mapping indeed | 17:04 |
| noonedeadpunk | but given that external net3s are added not frequently and some might want to have a control over it - it could be used | 17:05 |
| birbilakos | external net never changes - its a flat /22 network we get | 17:06 |
| noonedeadpunk | I can totally recall actually issue like that, but it was the case when molecule-plugin wasn't installed or smth like that (returning to your issue) | 17:06 |
| jrosser | that should be coming from tox i think | 17:06 |
| noonedeadpunk | I'd expect this to be added to the list of collections: https://github.com/ansible-community/molecule-plugins/blob/95141070006d996a5d43ce1f9301873342c9bfc1/src/molecule_plugins/docker/driver.py#L279 | 17:09 |
| noonedeadpunk | and be appended during run or smth: https://github.com/ansible/molecule/blob/23200bc984a8f9eb4bc5c20f7292632b75ca52c5/src/molecule/shell.py#L88-L91 | 17:10 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-role-systemd_service master: DNM - test molecule jobs https://review.opendev.org/c/openstack/ansible-role-systemd_service/+/939263 | 17:12 |
| jrosser | ^ sanity check | 17:12 |
| noonedeadpunk | jrosser: --destroy=never ? | 17:14 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/d8884a8323e74284a07586cbc646c23c/log/job-output.txt#1000 | 17:14 |
| jrosser | oh whoops | 17:14 |
| noonedeadpunk | but I don't see it in tox itself | 17:14 |
| noonedeadpunk | r well. I do :D | 17:15 |
| jrosser | that should not matter in a clean vm tbh | 17:15 |
| noonedeadpunk | it could be it's skipping requirements part then... but dunno | 17:15 |
| jrosser | left over from speeding up the hack cycles :) | 17:15 |
| noonedeadpunk | or well, driver requirement parts | 17:15 |
| birbilakos | Based on the following from the docs: "Octavia needs connectivity between the control plane and the load balancing VMs. For this purpose a provider network should be created which gives L2 connectivity between the octavia services on the controllers (either containerised or deployed on metal) and the octavia amphora VMs." i understand that this is what the br-lbaas network will be responsible for. Is that accurate? | 17:16 |
| noonedeadpunk | as it kinda `Running default > dependency\n WARNING Skipping, missing the requirements file.` | 17:16 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 17:18 |
| jrosser | birbilakos: br-lbaas is an artefact really of how the CI tests are structured, and also conflates what has to happen on the control plane and on the computes which is not necessarily always the same | 17:18 |
| jrosser | for the control plane, you need an interface which somehow brings the lbaas-mgmt network to the controllers, and if it's an LXC based control plane then you need that to be a bridge so that the containers can be hooked up to it | 17:20 |
| jrosser | on the computes, it needs to match up with how you describe it in neutron as a provider network | 17:20 |
| jrosser | and it's up to you if you make that flat or vlan | 17:21 |
| jrosser | and there is perhaps no need for it actually to be a bridge on the computes, becasue ultimately you give an interface to neutron | 17:22 |
| birbilakos | so how is the control plane (i.e. octavia containers) are supposed to be able to reach the load balancing VMs? I though the above network_provider definition of br-lbaas would server that purpose | 17:23 |
| jrosser | you provision a network (however you want to) on your network infrastructure to do that | 17:23 |
| birbilakos | i did provision this network already, no? All servers can communicate through it -> br-lbaas linux bridge | 17:24 |
| birbilakos | 172.29.232.0/22 | 17:26 |
| noonedeadpunk | and VMs spawned in openstack need to able to connect to it as well | 17:26 |
| jrosser | you already made that a vlan 40 on bond1? | 17:27 |
| noonedeadpunk | as octavia spwans it's balancers as regular openstack vms | 17:27 |
| jrosser | i dont think you need to assign any IP to br-lbaas anywhere - in fact thats probably a security disaster | 17:28 |
| birbilakos | so, this needs to be defined similarly to my br-public interfaces. Yes, br-lbaas is on vlan 40 on bond 1 | 17:32 |
| jrosser | is there a reason you cannot combine these networks all onto a trunk port? | 17:34 |
| jrosser | lbaas, public and public2? | 17:34 |
| jrosser | noonedeadpunk: molecule jobs look all broken on systemd_service role too https://zuul.opendev.org/t/openstack/build/4168176ce1a646a79235bbe3e1bf17ed | 17:36 |
| jrosser | i just picked systemd_service completely randomly to check there was not some underlying issue | 17:37 |
| noonedeadpunk | doh | 17:40 |
| noonedeadpunk | it was fast | 17:40 |
| noonedeadpunk | wtf has changed to make things break | 17:41 |
| noonedeadpunk | like this ran today morning: https://zuul.opendev.org/t/openstack/build/782419d5ce5649c9a49a972b9677536f | 17:42 |
| noonedeadpunk | btw the task is just skipped here: https://zuul.opendev.org/t/openstack/build/782419d5ce5649c9a49a972b9677536f/log/job-output.txt#1082-1084 | 17:43 |
| noonedeadpunk | but then again - dependency must exist and be installed somehow | 17:43 |
| noonedeadpunk | I can only guess some issues with galaxy that are just hidden by molecule | 17:49 |
| noonedeadpunk | like https://galaxy.ansible.com/ui/search/?keywords=community.docker | 17:49 |
| noonedeadpunk | also interestingly - https://galaxy.ansible.com/community/docker gives 404 at first and then redirects | 17:51 |
| noonedeadpunk | not sure though if it can be an issue or just me assuming search working like that | 17:51 |
| jrosser | ok i can reproduce this locally now | 17:55 |
| jrosser | community.docker is cached in my ~/.ansible | 17:55 |
| noonedeadpunk | locally now I got even `ERROR ERROR! Unexpected Exception, this is probably a bug: Non integer values in LooseVersion ('master')` | 17:55 |
| noonedeadpunk | for `ansible-galaxy collection install --pre -r /home/dr5005/Documents/ansible/ansible-role-systemd_service/requirements.yml` | 17:56 |
| jrosser | yeah you can'y put "master" as a collection version | 17:56 |
| noonedeadpunk | you can but with force iirc | 17:56 |
| noonedeadpunk | https://opendev.org/openstack/ansible-role-systemd_service/src/branch/master/requirements.yml#L6 | 17:57 |
| noonedeadpunk | and that worked couple of days ago | 17:57 |
| noonedeadpunk | and we should have a very specific version of ansible-core | 17:57 |
| noonedeadpunk | ansible-galaxy collection install --force -r /home/dr5005/Documents/ansible/ansible-role-systemd_service/requirements.yml does work | 17:58 |
| noonedeadpunk | https://opendev.org/openstack/ansible-role-systemd_service/src/branch/master/molecule/default/molecule.yml#L6 | 17:58 |
| noonedeadpunk | so wtf... | 17:58 |
| jrosser | we are also very specific about the versions of molecule | 18:00 |
| jrosser | so that should not have changed | 18:00 |
| noonedeadpunk | there was none released | 18:00 |
| jrosser | jamesdenton__: are you doing anything with ironic + neutron segmented networks? | 18:07 |
| noonedeadpunk | I really don't get what has changed in molecule so it's now ignoring quite a list of things | 18:16 |
| jrosser | i just confirmed that the version of molecue and molecule-plugins in the tox venv are the ones we specify in test-requirements.yml | 18:18 |
| birbilakos | jrosser: would the following config work? a) get rid of the br-lbaas bridge, b) hook up another veth pair to br-ext (similar veth pairs I use for nets br-public and br-public2 already) | 18:18 |
| noonedeadpunk | it could be some requirement, but I'm not getting which one | 18:19 |
| jrosser | birbilakos: well i think i'm still trying to understand why you have done br-public and br-public2 like that | 18:19 |
| birbilakos | c) use the new veth pair to host_bind_override | 18:19 |
| birbilakos | actually in reality only br-public is used | 18:20 |
| jrosser | veth pairs should not be necessary, nor bridges | 18:20 |
| birbilakos | br-public2 is not used for external network access | 18:20 |
| jrosser | ok sure | 18:20 |
| birbilakos | the reason I opted for the veth pairs was because I could not use the br-ext interface itself | 18:21 |
| jrosser | but i am still trying to understand why it is more favourable to have multiple falt networks rather than a trunk | 18:21 |
| birbilakos | which is practically the management interface of said hosts | 18:21 |
| jrosser | ooohhh right | 18:21 |
| jrosser | ewwwwww | 18:21 |
| birbilakos | via these pairs I can define the ext networks in openstack and assign floating ips | 18:21 |
| birbilakos | which are in the 10.x range | 18:22 |
| jrosser | if it was me, i would be treating all of public, public2, whatever else, and lbaas as a trunk port | 18:23 |
| birbilakos | but how would you assign an ip to the servers then? | 18:24 |
| birbilakos | in order to have them be reachable (e.g. management plane) | 18:24 |
| jrosser | by having some netplan or whatever that de-encapsulates one vlan from the trunk, and connects it to an interface with an ip | 18:24 |
| noonedeadpunk | jrosser: hmmmmm https://pypi.org/project/ansible-compat/ | 18:26 |
| noonedeadpunk | 3 hours ago | 18:26 |
| noonedeadpunk | `Refactor search_galaxy_paths to use pathlib` among changes | 18:27 |
| birbilakos | i'm not sure I understand :( at the moment there's just single physical interface which has vlan segregation to host all openstack networks, including ext net | 18:27 |
| jrosser | ok | 18:28 |
| jrosser | so why do you want to use flat networks rather than let neutron deal with the encap/decap? | 18:29 |
| birbilakos | well, I'm not that experienced with openstack i guess :) | 18:29 |
| birbilakos | I'm not looking for a redesign here tbh so I'm thinking how I can expand the curent config to support the new use case of lbaas | 18:30 |
| jrosser | of course it's your choice | 18:30 |
| birbilakos | what I'm still unclear about is whether the communication / network of the controller nodes to the VMs needs to happen over the br-ext net (10.x) or can do this over some internal net (e.g. 172.29.232.0/22) | 18:31 |
| jrosser | but i wanted to show how that adding the lbaas network as a new flat network will require a reconfiguration of every compute/network node and new physnet mappings adding to all the neutron config files, so that is pretty invasive | 18:32 |
| jrosser | however, if you instead had your provider networks as a trunk into OVS you could do that with a single CLI operation to neutron, to define a new provider network for lbass and tell it which vlan-id to use | 18:33 |
| jrosser | if you think there might be a need to ever add more provider networks, or change the ones you have, it would be operationally much much eaiser if they are not flat networks | 18:34 |
| jrosser | so about communication between controllers and VMs | 18:35 |
| jrosser | that is done over a new "provider network", which is basically the same as you external network, except it is carrying an internal network | 18:35 |
| birbilakos | ok, if you can help me how to implement this in my env, I can try it. From a networking point of view, all servers have 1 NIC, vlan segregation for 172.x networks is done on the switch level | 18:36 |
| jrosser | there is quite a nice blog post here https://satishdotpatel.github.io/openstack-ansible-octavia/ | 18:37 |
| jrosser | in your case you will need to switch out br-vlan for whatever your single interface is | 18:38 |
| jrosser | anyway | 18:38 |
| jrosser | really nothing stops you doing it also as an additional flat network | 18:38 |
| jrosser | it is just more things to get right in openstack_user_config and more complexity, in my opinion | 18:39 |
| jrosser | there should be enough info there for you to make it work either way | 18:39 |
| birbilakos | thank you jrosser, I will study this link | 18:40 |
| jrosser | you might also need some adjustment for ovs perhaps, as that example is linuxbridge based for the openstack_user_config part | 18:42 |
| birbilakos | correct, basically: 10.102.194.102 | 18:55 |
| birbilakos | I mean to say: group_binds: - neutron_openvswitch_agent | 18:55 |
| birbilakos | the problem i see with the solution in the link is that it requires a br-vlan which I dont have | 18:59 |
| jrosser | jrosser> in your case you will need to switch out br-vlan for whatever your single interface is | 19:00 |
| jrosser | sorry if this is too complicted | 19:01 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Pin ansible-compat up to 25.0.0 https://review.opendev.org/c/openstack/openstack-ansible/+/939274 | 19:06 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 19:09 |
| noonedeadpunk | huh, https://docs.ansible.com/ansible/latest/playbook_guide/playbooks_vars_facts.html made me smile | 19:16 |
| noonedeadpunk | `"ansible_user_dir": "/home/zuul"` | 19:16 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 19:19 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 19:21 |
| noonedeadpunk | so the only difference between the job that succeeded in the morning and which failing now - ansible-compat-24.10.0 vs 25.0.0 | 19:31 |
| noonedeadpunk | but feels like something more has changed.... | 19:32 |
| noonedeadpunk | ah, damn | 19:32 |
| noonedeadpunk | depends-on doesn't work.... | 19:33 |
| noonedeadpunk | jrosser: you said you could reproduce issue locally? | 19:33 |
| jrosser | yeah i can | 19:33 |
| noonedeadpunk | can you check if this helps? https://review.opendev.org/c/openstack/openstack-ansible/+/939274/1/test-requirements.txt | 19:33 |
| noonedeadpunk | as I somehow don't see it still... | 19:34 |
| jrosser | i'm just trying to use that as a local test-requirements.yml | 19:36 |
| jrosser | maybe i still have some underlying error | 19:36 |
| jrosser | ok this at least gets past the community.docker error and is now running the prepare/converge | 19:37 |
| noonedeadpunk | ++ | 19:37 |
| jrosser | i do not yet have a good idea how to make that work with depends-on | 19:37 |
| jrosser | unless we can write some externally called shell script to retrieve that file either via http or from a file, depending on some env var | 19:38 |
| noonedeadpunk | well. that what was the most pita with functional jobs - is that dependencies didn't really work | 19:42 |
| noonedeadpunk | but hopefully test-requirements won;t be chaning too frequently, so it might be not _that_ annoying | 19:43 |
| jrosser | well | 19:47 |
| jrosser | right now in the molecule env definition in tox.ini, we have a var for TOX_CONSTRAINTS_FILE | 19:48 |
| jrosser | we could have a similar one for TEST_REQUIREMENTS_FILE, and just define that in our zuul jobs | 19:48 |
| jrosser | and let it default to the usual url othewise | 19:48 |
| noonedeadpunk | so I think the issue would be, that repo is on "localhost" while tox is running on node | 19:51 |
| noonedeadpunk | so we'd need also to copy the file | 19:51 |
| noonedeadpunk | but then there's also a question about requirements.yml, as we define there our own collections and roles from time to time | 19:52 |
| jrosser | but thats in the repo under test, so that should work out ok | 19:52 |
| jrosser | unless there is a further case of depends-on | 19:53 |
| noonedeadpunk | I was thinking abotu this usecase as next one that can be problematic: https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257/5/requirements.yml | 19:53 |
| noonedeadpunk | but yeah, I agree that test-requirements.txt should be solvable... | 19:54 |
| jrosser | the repos are on the test node though? | 19:54 |
| noonedeadpunk | oh. | 19:54 |
| noonedeadpunk | probably you're right | 19:54 |
| noonedeadpunk | so you think we can also generate a proper requirements file instead of using one that's in-repo? | 19:55 |
| jrosser | you are right though that there is no embedded ansible in this case | 19:55 |
| jrosser | we almost have that code anyway in get-role/collection requirements | 19:56 |
| noonedeadpunk | well | 19:56 |
| noonedeadpunk | I was just moving it out lol | 19:56 |
| noonedeadpunk | in https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 19:57 |
| jrosser | right but perhaps thats actually good | 19:57 |
| noonedeadpunk | as now there's a separate playbook doing that... | 19:57 |
| jrosser | as our base molecule job is defined in the openstack-ansible repo we could have a pre playbook that reformats any requirements.yml file | 19:57 |
| noonedeadpunk | yeah | 19:58 |
| jrosser | and then that would automatically apply to all molecule jobs | 19:58 |
| noonedeadpunk | true | 19:58 |
| jrosser | and same trick with an env var to switch to it in tox.ini | 19:58 |
| jrosser | that really gives some direction into how to move/refactor that code to make it most useful | 19:59 |
| noonedeadpunk | sounds doable... | 19:59 |
| noonedeadpunk | so, now I'm calling (at least trying to call) this one https://review.opendev.org/c/openstack/openstack-ansible/+/939151/19/zuul.d/playbooks/pre-osa-requirements.yml from run-upgrade.sh | 20:00 |
| noonedeadpunk | we can make just src and dest configurable for collections | 20:01 |
| noonedeadpunk | another thing I was looking at, is that ansible-collection-requirements do no allow to deal with `roles` stanza, which we might need to fix as well | 20:02 |
| jrosser | sure - that sounds good | 20:08 |
| jrosser | and configurable src/dest are probably most of what we need so long as the files are all proper galaxy requirements format | 20:08 |
| jamesdenton__ | jrosser I am not yet doing anything with ironic+segments, but I do recall testing it with some success | 20:41 |
| jrosser | interesting - we have found networking-generic-switch to be missing some understanding of segments | 20:42 |
| jamesdenton__ | that is true. i am not using NGS in this case | 20:43 |
| jamesdenton__ | curious if networking-baremetal would be better? | 20:43 |
| jrosser | well - that us all very confusing | 20:43 |
| jamesdenton__ | :D | 20:43 |
| jamesdenton__ | TBH i have not tested networking-baremetal, yet. | 20:44 |
| jrosser | because you need ironic-neutron-agent (which is part of netowrking-baremetal) to fill out the relevant database bits about segments | 20:44 |
| jrosser | and extremely confusingly networking-baremetal has grown the ability to twiddle netconf things, and also understands segments quite well | 20:44 |
| jrosser | but for !netconf you need n-g-s, and thats not so good for segments | 20:45 |
| jrosser | we made a start here https://review.opendev.org/c/openstack/networking-generic-switch/+/939211 | 20:45 |
| jrosser | but it seems there are also changes needed in neutron and nova | 20:45 |
| jamesdenton__ | i found NGS in practice to be a bit slow for me | 20:47 |
| jamesdenton__ | thanks for the patch, i will try and look at it in Q3 | 20:48 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Molecule to respect depends-on for test-requirements update https://review.opendev.org/c/openstack/openstack-ansible/+/939290 | 22:49 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-lxc_container_create master: Re-introduce functional tests with molecule https://review.opendev.org/c/openstack/openstack-ansible-lxc_container_create/+/939257 | 22:52 |
| noonedeadpunk | so you're saying that it was that easy and should jsut work? | 22:52 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_service master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-systemd_service/+/939292 | 22:56 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_service master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-systemd_service/+/939292 | 22:57 |
| noonedeadpunk | it seems it indeed "just works" | 23:00 |
| noonedeadpunk | python -I -m pip install -r /home/zuul/src/opendev.org/openstack/openstack-ansible/test-requirements.txt -c /home/zuul/src/opendev.org/openstack/requirements/upper-constraints.txt | 23:01 |
| noonedeadpunk | ok, that was way easier then I thought when we were discussing it | 23:01 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move zuul preparation for role/collection bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/939151 | 23:20 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Optimize generation of required roles/collections https://review.opendev.org/c/openstack/openstack-ansible/+/939221 | 23:21 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-apt_package_pinning master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/openstack-ansible-apt_package_pinning/+/939299 | 23:25 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-frrouting master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-frrouting/+/939300 | 23:27 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-pki master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-pki/+/939301 | 23:28 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-config_template master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-config_template/+/939302 | 23:29 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_mount master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-systemd_mount/+/939303 | 23:29 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_networkd master: Use OSA_TEST_REQUIREMENTS_FILE for molecule job https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/939304 | 23:30 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!