| opendevreview | Neil Hanlon proposed openstack/openstack-ansible-plugins master: DNM: try forcing systemd-udev to install for CentOS Stream https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/878803 | 00:51 |
|---|---|---|
| opendevreview | Neil Hanlon proposed openstack/openstack-ansible master: Add rockylinux-9 distro metal job to check pipeline https://review.opendev.org/c/openstack/openstack-ansible/+/878115 | 00:54 |
| NeilHanlon | oh.. this is a fun one... 🙃https://zuul.opendev.org/t/openstack/build/ed9ac39a6127407a8ecb5e0952ebb3cb | 00:57 |
| NeilHanlon | jrosser, noonedeadpunk: seeing CI failures on rocky and CentOS now due to.. something. there's a difference in the rocky packages from yesterday vs today. in particular, libibverbs, but, there's more than just that. https://paste.opendev.org/show/bUXUY3qlG3QULRixC5kU/ | 01:15 |
| jrosser | NeilHanlon: `nothing provides libmlx5.so.1(MLX5_1.24)(64bit)` if i had to completely guess than than version number almost suggests that library originating from the OFED driver? | 05:31 |
| jrosser | which actually i think is not the case, but the naming is pretty unusual | 05:33 |
| hamidlotfi | Hi there, | 07:04 |
| hamidlotfi | I wanted to scale up my OSA environment with your guide published at `https://docs.openstack.org/openstack-ansible/latest/admin/scale-environment.html` | 07:04 |
| hamidlotfi | I want to add infra04 in my env as a controller | 07:04 |
| hamidlotfi | But in the `setup-infrastructure.yml` show me this error: | 07:04 |
| hamidlotfi | `hostvars[groups[rabbitmq_host_group][0]]['ansible_facts']['hostname'] }}: 'dict object' has no attribute 'hostname'` | 07:04 |
| gokhanisi | Good morning folks, I need to install OSA on offline environment. I need your advises. for pypi mirror I am thinking to use python-pypi-mirror lib for mirroring and for pypi server using pypiserver. for distrubution packects use apt-mirror. Can you share your advises for offline OSA deployment? | 07:21 |
| jrosser | hamidlotfi: it is possible that the documentation does not match with changes made in the code | 07:25 |
| jrosser | hamidlotfi: the error is that facts have not been gathered from the first rabbitmq host, and that is likley due to the --limit in step 3 | 07:26 |
| jrosser | hamidlotfi: is it step 4 that fails for you? | 07:27 |
| jrosser | gokhanisi: i have previously used devpi as a caching proxy for pypi, so not completely offline but almost. I think that noonedeadpunk also has some experience with offline installs so might also have a suggestion for the python mirror | 07:28 |
| hamidlotfi | :jrosser I use this command `openstack-ansible playbooks/setup-infrastructure.yml --limit localhost,infra04,infra04-host_containers` | 07:31 |
| hamidlotfi | @jrosser | 07:31 |
| hamidlotfi | @jrosser | 07:32 |
| jrosser | hamidlotfi: i'm not sure thats going to work - step 4 says `openstack-ansible playbooks/setup-infrastructure.yml -e galera_force_bootstrap=true` | 07:35 |
| jrosser | and it also gives a specific reason for that, the rabbit and galera clusters need expanding to include the new node, so it's reasonable that data is required about all of the infra nodes during that process | 07:36 |
| gokhanisi | thanks jrosser I am also waiting for noonedeadpunk suggestions :) | 07:38 |
| gokhanisi | jrosser, it seems you used like this https://gist.github.com/aschmu/4bdb40f5212eea6bf3ea698eeea2d60c | 07:41 |
| hamidlotfi | @jrosser It's this mean that I run this `openstack-ansible playbooks/setup-infrastructure.yml -e galera_force_bootstrap=true` plus a command like this for rabbit or just run this for both configurations? | 07:44 |
| jrosser | gokhanisi: well we specifically didnt use bandersnatch as if you mirror the whole of pypi it is absolutely gigantic | 07:45 |
| jrosser | thats why we made a caching proxy instead, but like i said the environment was not totally isolated and the apt mirror and devpi servers had interfaces on inside and outside networks | 07:47 |
| jrosser | hamidlotfi: that command should deploy everything "infrastructure" onto your infra host, like repo server, haproxy, rabbitmq, galera etc | 07:48 |
| jrosser | gokhanisi: if you want something that is absolutely isolated, like air-gap networks then you will need something thats is more like a true mirror | 07:49 |
| gokhanisi | jrosser, yes it is absolutely isolated | 07:50 |
| jrosser | ok sure | 07:51 |
| hamidlotfi | @jrosser That command stoped with this error `mysqlcheck: Got error: 1102: Incorrect database name '#mysql50#tmp.J7AnZnVRKB' when selecting the database` | 07:59 |
| jrosser | hamidlotfi: can you tell me which release of OSA you are using? | 08:00 |
| hamidlotfi | ZED | 08:00 |
| jrosser | hmm ok | 08:00 |
| noonedeadpunk | gokhanisi: we're using pulp and aptly just in case | 08:24 |
| noonedeadpunk | no idea what's the best choice to be frank, wasn't doing any comparison, but these 2 worked fine | 08:25 |
| noonedeadpunk | yeah, having ovs dependency on melanox stuff is just wrong | 08:45 |
| noonedeadpunk | I don't think it's related in any way to Rocky, as ovs is taken from another repo | 08:45 |
| noonedeadpunk | that is very annoying though | 08:49 |
| noonedeadpunk | I kind of wonder if that might be tighten to rdo release 2023.1 or smth | 08:50 |
| *** kleini_ is now known as kleini | 08:59 | |
| noonedeadpunk | It seems that's kind of the reason https://review.rdoproject.org/r/c/rdoinfo/+/48014 | 09:02 |
| noonedeadpunk | As I was quite sure that 3.1 was not the case before... And now it's backported down to W... | 09:04 |
| jrosser | noonedeadpunk: well tbh rdo/RH product cycle is totally decoupled from regular openstack releases isnt it? | 09:37 |
| jrosser | just happens that the branches are the same name | 09:37 |
| noonedeadpunk | rhop is totally decoupled, but RDO is kind trailing thing | 09:38 |
| noonedeadpunk | but I mean - it's not under governance, so... | 09:38 |
| jrosser | right - but like we see with tripleo they only care about very specific branches? not each release | 09:38 |
| noonedeadpunk | yeah, kind of decoupled | 09:38 |
| noonedeadpunk | tripleo switched to independent release cycles since Wallaby | 09:39 |
| jrosser | so where that rdo commit says "OpenStack networking team has requested to backport this up to Wallaby." i have no idea how that is compatible with stable branch policy for wider opensack | 09:39 |
| noonedeadpunk | the problem with tripleo is that we should not have allowed them to make same branch naming | 09:39 |
| noonedeadpunk | but rdo does have all releases and they care about them kinda | 09:40 |
| noonedeadpunk | So if you check this - https://www.rdoproject.org/rdo/release-cadence/ it's kinda aligned with openstack releases despite they're not uner governance | 09:40 |
| jrosser | so question really is if we as OSA can commit to supporting stable branches deployed with rdo packages when they might make huge backports like that | 09:43 |
| noonedeadpunk | I'm not sure we have better place to take ovs from? | 09:47 |
| noonedeadpunk | as this is not distro path I'm talking about | 09:47 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/f295470ef7fe4ad3ae25d5c985c3d304/log/job-output.txt#18951 | 09:48 |
| noonedeadpunk | as native ovs provided in centos repos used to be too old to meet requirements | 09:48 |
| noonedeadpunk | maybe it's not the case anymore - don't know | 09:49 |
| admin1 | 26.0.1 .. trying to add compute note .. get this -> https://paste.openstack.org/raw/bCuztg9kMITqiXNqzhH5/ | 10:05 |
| admin1 | i don't think i have seen this error before . | 10:05 |
| noonedeadpunk | yeah, we have a bug for that already | 10:06 |
| noonedeadpunk | Not really sure how to workaround this properly. I truly believe that it's result of things being changed in Z | 10:06 |
| noonedeadpunk | in ansible | 10:06 |
| admin1 | i ran the nova playbook without limit as well ( so that facts are refreshed ) .. stuck in the same | 10:15 |
| admin1 | i can delelete ansible_facts and try again | 10:15 |
| admin1 | if its some cache thing . | 10:15 |
| noonedeadpunk | So, you need to gather facts for both computes and api/scheduler | 10:18 |
| noonedeadpunk | And that might be also troublesome if you have some computes down | 10:18 |
| admin1 | if i delete ansible_facts, and re-run setup-hosts and setup infra with no limits, might it fix ? | 10:25 |
| admin1 | i do have some hosts that are down and we are waiting for replacement parts | 10:25 |
| * jrosser still thinks we should add `-e adding_compute_node=true` | 10:25 | |
| * admin1 wants (start) as well .. | 10:26 | |
| admin1 | (star) | 10:26 |
| admin1 | ignore my last 2 lines .. was testing the /me | 10:26 |
| noonedeadpunk | NeilHanlon: so... with rdo folks upating ovs/ovn, now rocky misses libmlx5.so.1. They claim that package comes from CentOS Base at the moment and it's required due to mlx now is in enabled drivers by default https://git.centos.org/rpms/openvswitch/blob/c9s-sig-nfv-openvswitch-3.1/f/SPECS/openvswitch3.1.spec#_309 | 10:28 |
| noonedeadpunk | CentOS is broken differently though then Rocky - it fails on tempest with https://git.centos.org/rpms/openvswitch/blob/c9s-sig-nfv-openvswitch-3.1/f/SPECS/openvswitch3.1.spec#_309 | 10:29 |
| noonedeadpunk | *https://git.centos.org/rpms/openvswitch/blob/c9s-sig-nfv-openvswitch-3.1/f/SPECS/openvswitch3.1.spec#_309 | 10:29 |
| noonedeadpunk | (if forget about udev issue for a second) | 10:29 |
| noonedeadpunk | Which goes down to https://zuul.opendev.org/t/openstack/build/64d8812538584e4b9d2b95e4c0907299/log/logs/host/ovn/ovn-northd.log.txt#32 | 10:36 |
| noonedeadpunk | we also have now couple of ssl issues logged here https://zuul.opendev.org/t/openstack/build/64d8812538584e4b9d2b95e4c0907299/log/logs/host/ovn/ovsdb-server-nb.log.txt | 10:38 |
| noonedeadpunk | jamesdenton: we might need some your help here to be frank... | 10:40 |
| jamesdenton | hi noonedeadpunk - i will try and check that out this moprnming | 10:45 |
| jamesdenton | *morning | 10:45 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_mount master: Ensure systemd-udev is installed for gluster https://review.opendev.org/c/openstack/ansible-role-systemd_mount/+/878851 | 10:51 |
| noonedeadpunk | so they've basically upgraded ovs/ovn from 2.17/22.06 to 3.1/22.12 | 10:52 |
| jamesdenton | ok | 10:52 |
| noonedeadpunk | ssl probably not an issue though. It's that aio is not registered somewhat, despite we obviously running command to set hostname | 11:03 |
| jamesdenton | at one time there was a command to set the hostname within ovs, but i believe at some point that became problematic, not sure if it's still there | 11:04 |
| jamesdenton | I'll see if i can stand up a Rocky AIO and tshoot the OVS issue. The MLX library requirement is likely due to that being the DPDK build of OVS | 11:05 |
| noonedeadpunk | I will check that, but at least we're running this thing https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/tasks/providers/setup_ovs_ovn.yml#L18 | 11:05 |
| jamesdenton | ahh yes, that's the one. | 11:06 |
| noonedeadpunk | I'm spawning aio for centos metal already | 11:06 |
| NeilHanlon | if we install rdma-core we should be okay, provided the soname is still provided | 11:06 |
| noonedeadpunk | rdma-core is... a repo or a package? | 11:07 |
| NeilHanlon | rdma-core provides libibverbs, which provides the mlx lib | 11:07 |
| NeilHanlon | package | 11:07 |
| noonedeadpunk | aha, ok | 11:07 |
| jamesdenton | ugly ugly. :| | 11:07 |
| noonedeadpunk | yeah, that sucks for sure | 11:07 |
| noonedeadpunk | damn, I pasted same link 3 times :( | 11:08 |
| jamesdenton | noonedeadpunk at one point, the hostname was never set and that had to be manually do. I believe with newer OVS it is done automatically, so we *might* be able to do without that command, now. Can try and patch it out and see | 11:08 |
| jamesdenton | need to get the kids off to school, but will be back in a bit | 11:09 |
| noonedeadpunk | yeah, sure, no huge rush | 11:09 |
| noonedeadpunk | the error neutron fails to bind the port is `Refusing to bind port c6249ef3-801f-4615-8630-5dcaf2a7c3ee due to no OVN chassis for host: aio1` https://zuul.opendev.org/t/openstack/build/64d8812538584e4b9d2b95e4c0907299/log/logs/host/neutron-server.service.journal-02-32-13.log.txt#6545 | 11:10 |
| noonedeadpunk | So it kinda feels that hostname is not set or set wrongly, but I'm pretty sure it should be fine... | 11:11 |
| noonedeadpunk | jamesdenton: you're right - ovs now auto-adds hostname, but it does that incorrectly | 11:29 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/64d8812538584e4b9d2b95e4c0907299/log/logs/etc/host/openvswitch/conf.db.txt#20 | 11:31 |
| noonedeadpunk | (and L8) | 11:31 |
| noonedeadpunk | Oh! And then L64 | 11:32 |
| noonedeadpunk | would be great to have a way to disable that... | 11:36 |
| jrosser | oh! is this my fault? i recently added that aio1.openstack.local fqdn to /etc/hosts in the aio | 11:52 |
| admin1 | noonedeadpunk, do you know of any way ( even manual or hacky) that will allow me to add that new compute node ? | 11:53 |
| noonedeadpunk | I think it's ovs 3+ thingy | 11:53 |
| jrosser | admin1: to work around that compute node add issue we allowed the playbook to run without --limit until it had gathered facts | 11:54 |
| jrosser | then we CTRL-C | 11:54 |
| noonedeadpunk | admin1: eventually it fails _after_ compute is deployed | 11:54 |
| jrosser | and run it again with the limit, but in our case we did not have any nodes that were offline | 11:54 |
| jrosser | so i think that this is still going to be troublesome for you | 11:54 |
| noonedeadpunk | We already have a bug report coming https://bugs.launchpad.net/openstack-ansible/+bug/2013189 | 12:01 |
| noonedeadpunk | So... ovs solutions likely are either to set `RECORD_HOSTNAME=no` or `FULL_HOSTNAME=no` and drop our task for setting it | 12:43 |
| noonedeadpunk | and it's `/usr/share/openvswitch/scripts/ovs-ctl` that messes things around | 12:47 |
| noonedeadpunk | https://github.com/openvswitch/ovs/blob/master/utilities/ovs-ctl.in#L42-L52 | 12:47 |
| noonedeadpunk | does anybody has ubuntu aio with ovn to check smth? | 12:58 |
| damiandabrowski | noonedeadpunk: i can | 13:11 |
| noonedeadpunk | can you place my key there then ?:) | 13:11 |
| damiandabrowski | sure, one sec | 13:11 |
| jrosser | does anyone have a moment to look at https://lists.openstack.org/pipermail/openstack-discuss/2023-March/033008.html | 13:29 |
| jrosser | its kind of an OSA question not designate | 13:29 |
| noonedeadpunk | it slipped me due to the tag... | 13:30 |
| admin1 | noonedeadpunk, i have multi deplpoyment | 13:35 |
| admin1 | osa + ovn | 13:35 |
| admin1 | 1 controller + 2 compute ( also as network nodes ) | 13:36 |
| admin1 | testing k8s in that | 13:36 |
| admin1 | k8s + octavia | 13:36 |
| admin1 | k8s + octavia on top of osa+ovn :D | 13:37 |
| noonedeadpunk | that's already too hard for me :D | 13:44 |
| *** spotz__ is now known as spotz_ | 13:51 | |
| noonedeadpunk | jamesdenton: NeilHanlon so just to update you a bit what we've found with RDO folks regarding OVS and their update - that openvswitch 3.1 has broken down `ovs-vsctl add` command. As it intends to adding things rather then replacing | 14:00 |
| noonedeadpunk | So it's regression of https://github.com/openvswitch/ovs/commit/2221e8b034298bd77dc9282895c4b32fd11583e8 basically | 14:01 |
| noonedeadpunk | or well. not exactly, but idea is that `add` should not override things, while it obviously does for 3.1. It's not for 3.0.1 which ubuntu is running and worked properly in 2.17 | 14:03 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Workaround ovs bug that resets hostname with add command https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/878911 | 14:24 |
| noonedeadpunk | I feel we're in a horrible dependency loop as we have 3 separate issues in rhel distros that arised for the last couple of days | 14:26 |
| noonedeadpunk | not counting rabbit on stable branches... | 14:31 |
| * NeilHanlon sighs | 14:55 | |
| * NeilHanlon wonders if it's acceptable to crack a beer at 11am | 14:55 | |
| noonedeadpunk | Yeah... I'm was thinking the same but wrt more strong drinks... | 14:56 |
| NeilHanlon | for the rocky problem missing libibverbs (mlx SO) -- something "feels" wrong about that since a CI job from a day ago is pulling in this package, and now all of a sudden.. not | 14:58 |
| NeilHanlon | I wonder if some previous package from centos sig-nfv had an explicit `Requires` that is no longer there | 15:00 |
| noonedeadpunk | day ago it was ovs2.17, now it's 3.1 | 15:00 |
| noonedeadpunk | I have centos spawned and now trying to spawn rocky to check | 15:00 |
| NeilHanlon | i'm rebuilding my AIO as well... | 15:00 |
| noonedeadpunk | well, yeah, libibverbs-44.0-2 is being installed from base | 15:02 |
| jrosser | this feels like the sort of trouble you get when packages are not build in clean environments | 15:03 |
| * jrosser thinks back to days with pbuilder for deb packages | 15:03 | |
| NeilHanlon | stream is _just_ far enough ahead that we run into these sorts of things :( | 15:32 |
| jrosser | i guess that any degree of parallelism in the builds of different packages is a complete minefield as well | 15:34 |
| NeilHanlon | i'll write up a _something_ and pass it over to Amy to see what we can do overall. i can imagine this won't be the last time something like this happens, so it'd be good to have more/better communication coverage over these impactful sorts of things due to the delta between CentOS Stream and RHEL | 15:39 |
| noonedeadpunk | I'd still propose quick dirty fix to install old ovs until this solved so we could unblock gates | 15:52 |
| noonedeadpunk | As discussion sounded to me like not this week fix for sure | 15:52 |
| opendevreview | Neil Hanlon proposed openstack/openstack-ansible-os_neutron master: Pin openvswitch package on RHEL-likes to 2.17 https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/878924 | 15:53 |
| NeilHanlon | noonedeadpunk: ^ | 15:54 |
| noonedeadpunk | NeilHanlon: this is not working jsut in case ^ | 15:54 |
| noonedeadpunk | I had to add `exclude = *rdo-openvswitch*3.1*,*rdo-ovn*3.1*` to avoid this from being isntalled | 15:55 |
| NeilHanlon | ah :( | 15:55 |
| * NeilHanlon sighs | 15:55 | |
| noonedeadpunk | https://paste.openstack.org/show/bGFidAmbzq9Uom2Buze8/ | 15:55 |
| noonedeadpunk | That is annoying, I know | 15:56 |
| noonedeadpunk | I think we're adding repo in openstack_hosts repo | 15:56 |
| noonedeadpunk | I will push a patch now | 15:57 |
| NeilHanlon | we could maybe pin that rdo-openvswitch pkg instead? | 15:58 |
| noonedeadpunk | Um, might be but I struggled with that a bit | 16:02 |
| NeilHanlon | no prob. exclude is a good/better option | 16:05 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Pin openvswitch package on Rocky to 2.17 https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878926 | 16:06 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Pin openvswitch package on Rocky to 2.17 https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878926 | 16:06 |
| noonedeadpunk | I;m just not sure if exclude for everything or jsut rocky | 16:07 |
| noonedeadpunk | As we're having serious dependency loop now | 16:07 |
| NeilHanlon | stream doesn't need it | 16:07 |
| NeilHanlon | afaik | 16:07 |
| noonedeadpunk | As we need to merge 3 patches in 3 repos now to fix all breakages | 16:07 |
| noonedeadpunk | and backport all that | 16:08 |
| NeilHanlon | right | 16:11 |
| NeilHanlon | IMO we only need the openvswitch change for rocky tests; then we need the hostname change only for stream, i guess? | 16:12 |
| NeilHanlon | Or if we do the openvswitch change for both for now, maybe we don't need that hostname patch | 16:12 |
| noonedeadpunk | Yeah, I kind of thinking what would be easier | 16:15 |
| noonedeadpunk | But we also have this things as well https://review.opendev.org/c/openstack/ansible-role-systemd_mount/+/878851 that breaks stream | 16:16 |
| noonedeadpunk | maybe indeed we can merge using old ovs repo easy enough | 16:17 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Pin openvswitch package on RHEL to 2.17 https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878926 | 16:18 |
| opendevreview | Neil Hanlon proposed openstack/openstack-ansible-plugins master: Ensure systemd-udev is installed for gluster https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/878803 | 16:22 |
| NeilHanlon | removed the DNM from that one just in case.. | 16:23 |
| noonedeadpunk | are you sure it's needed? | 16:24 |
| noonedeadpunk | As it looks to be in place? https://opendev.org/openstack/openstack-ansible-plugins/src/branch/master/roles/glusterfs/vars/redhat.yml#L22 | 16:24 |
| noonedeadpunk | Ah... | 16:25 |
| noonedeadpunk | I think we're having ordering issue | 16:25 |
| noonedeadpunk | So L60 should be kinda moved to L44 | 16:25 |
| damiandabrowski | jrosser: can you please explain why did you include haproxy_base_service in horizon_haproxy_services? | 16:26 |
| damiandabrowski | https://opendev.org/openstack/openstack-ansible/src/commit/7f358cd906d27582da8affb11de4d7b1c3b7465a/inventory/group_vars/horizon_all.yml#L49 | 16:26 |
| damiandabrowski | it's not working properly with separated haproxy config because horizon_all group does not have an access to haproxy_base_service var. | 16:26 |
| damiandabrowski | But horizon works fine on my AIO without that.* | 16:27 |
| damiandabrowski | * after fixing this bug: https://opendev.org/openstack/openstack-ansible/src/commit/4195a2937c0281b49bdad9633fb4b237d58fc77e/inventory/group_vars/haproxy/haproxy.yml#L262 | 16:27 |
| damiandabrowski | it should be '.* horizon-back' instead of '.* horizon' | 16:27 |
| noonedeadpunk | or well... should it... | 16:27 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-systemd_mount master: Ensure systemd-udev is installed for gluster https://review.opendev.org/c/openstack/ansible-role-systemd_mount/+/878851 | 16:28 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Workaround ovs bug that resets hostname with add command https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/878911 | 16:30 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Workaround ovs bug that resets hostname with add command https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/878911 | 16:31 |
| noonedeadpunk | doh | 16:37 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Ensure systemd-udev is installed for gluster https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/878803 | 16:41 |
| noonedeadpunk | I think I'm a bit lost now in all dependencies we need... | 16:43 |
| jrosser | damiandabrowski: doesnt that patch come from before the work with haproxy maps was done? | 16:43 |
| jrosser | damiandabrowski: remember that the step before doing maps was moving to 'default_backend' for horizon on the base service | 16:44 |
| jrosser | so i think that now that the maps work is done, that line is no longer needed | 16:47 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Pin openvswitch package on RHEL to 2.17 https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878926 | 16:47 |
| damiandabrowski | the 'Prepare main repo for separated haproxy config' patch? yes, be AFAIK we agreed to rebase it on top of your haproxy maps changes | 16:48 |
| noonedeadpunk | huh, whaat, zuul now re-schedules automatically if you update dependant patch? | 16:49 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Workaround ovs bug that resets hostname with add command https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/878911 | 16:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Limit usage of old OVS to Rocky only https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878929 | 16:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Limit usage of old OVS to Rocky only https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/878929 | 16:54 |
| noonedeadpunk | so we kinda have this topic to track patches https://review.opendev.org/q/topic:osa%252Frdo_ovs_3.1 | 16:56 |
| noonedeadpunk | Se we will have troubles with plugins repo patching at very least, as we in fact don't install collection version we're currently testing | 17:48 |
| noonedeadpunk | I think this is due to setting git+file here - https://opendev.org/openstack/openstack-ansible/src/branch/master/scripts/get-ansible-collection-requirements.yml#L63 | 17:48 |
| noonedeadpunk | I think worth trying to switch type there to `dir` | 17:48 |
| noonedeadpunk | so we'll likely need to disable centos jobs for plugins repo until ovs patches land | 17:49 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Ensure systemd-udev is installed for gluster https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/878803 | 17:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Restore CentOS jobs https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/878933 | 17:53 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Use a map file to select haproxy horizon backend from the base frontend https://review.opendev.org/c/openstack/openstack-ansible/+/876851 | 19:50 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Enable TLS frontend for repo_server by default https://review.opendev.org/c/openstack/openstack-ansible/+/876426 | 19:50 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Implement separated haproxy service config https://review.opendev.org/c/openstack/openstack-ansible/+/871189 | 19:50 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Update letsencrypt docs https://review.opendev.org/c/openstack/openstack-ansible/+/878939 | 19:50 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!