16:00:18 <noonedeadpunk> #startmeeting openstack_ansible_meeting
16:00:19 <openstack> Meeting started Tue Jan 12 16:00:18 2021 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:22 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:00:31 <noonedeadpunk> #topic bug triage
16:00:55 <noonedeadpunk> hey! have been a while since last meeting and there are a lot of new stuff in launchpad...
16:01:29 <noonedeadpunk> https://bugs.launchpad.net/openstack-ansible/+bug/1908510
16:01:30 <openstack> Launchpad bug 1908510 in openstack-ansible "Keystone federation part is not executed during play" [Undecided,New]
16:01:42 <noonedeadpunk> I think jrosser you have laready covered that?
16:04:56 <noonedeadpunk> oh, ok, fix is not working
16:05:11 <noonedeadpunk> specificly this part https://review.opendev.org/c/openstack/openstack-ansible/+/760431/1/playbooks/os-keystone-install.yml
16:06:37 <jrosser> o/ hello
16:06:52 <noonedeadpunk> but was also fixed afterwards....
16:07:37 <noonedeadpunk> uh, this broken gerrit->launchpad integgration makes me mad
16:07:39 <jrosser> it is fixed here i think https://review.opendev.org/c/openstack/openstack-ansible/+/767513
16:08:02 <noonedeadpunk> yeah, alredy found :(
16:08:33 <noonedeadpunk> ok, next thing https://bugs.launchpad.net/openstack-ansible/+bug/1908544
16:08:34 <openstack> Launchpad bug 1908544 in openstack-ansible "ROLE lxc_hosts: variable lxc_hosts_container_image_url has no effect" [Undecided,New]
16:10:53 <jrosser> oh this is always true https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/defaults/main.yml#L204
16:11:13 <jrosser> so this test does not work https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/tasks/lxc_cache_prestage.yml#L52
16:12:04 <noonedeadpunk> but it should be always false then?
16:12:38 <noonedeadpunk> ah, because of the default, yeah
16:13:00 <noonedeadpunk> that does not make sense actually...
16:13:14 <noonedeadpunk> we don;t need to set_fact at all I guess
16:14:25 <jrosser> for debian and redhat it's always set to something https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/vars/debian.yml#L16
16:14:34 <noonedeadpunk> that is totally broken ,confirming
16:15:18 <jrosser> the set fact is required for ubuntu i think to resolve the container name?
16:15:39 <jrosser> the later when: clauses are supposed to only make it run for ubuntu?
16:18:07 <noonedeadpunk> you mean that checksum is not defined?
16:19:35 <noonedeadpunk> I think yI could broke that when was fixing ubuntu image somehow
16:20:16 <jrosser> certainly the vars are set up differently between ubuntu and the other distros
16:20:18 <noonedeadpunk> anyway, it should be length test at least, not bool
16:21:02 <jrosser> and we must change the default for lxc_hosts_container_image_url to be just empty
16:21:14 <noonedeadpunk> yeah, agree
16:21:46 <noonedeadpunk> or not - I think that will break everythin except ubuntu?
16:22:21 <jrosser> the whole thing needs overhauling i guess
16:22:29 <jrosser> again :(
16:23:14 <jrosser> it might be better to define a bool in vars/<distro>.yml that says if the magic code to resolve the latest container is needed or not
16:23:26 <jrosser> the its just simpler and 'true' for ubuntu and false for everything else
16:23:39 <noonedeadpunk> `_lxc_hosts_container_image_url` for ubuntu should not be defined, but I think empty string will give false
16:24:11 <noonedeadpunk> well, yes, we might go this way as well
16:24:40 <jrosser> if _lxc_hosts_container_image_url | length = 0 and _lxc_hosts_container_image_resolve ....
16:24:59 <jrosser> i.e only do it if there is no provided URL and the OS requires that we resolve
16:25:15 <jrosser> +/- an underscore there
16:25:48 <noonedeadpunk> yeah, ok, agreed
16:26:01 <noonedeadpunk> https://bugs.launchpad.net/openstack-ansible/+bug/1908703
16:26:03 <openstack> Launchpad bug 1908703 in openstack-ansible "federation domain not configured correct" [Undecided,New]
16:27:03 <noonedeadpunk> I really was trying not to touch that part of the code as never used it with my own
16:27:26 <ThiagoCMC> Hey guys, which hosts are supposed to be under 'masakari-monitor_hosts:' ? The controllers or the compute nodes? The Pacemaker and Corosync were installed at the baremetal! Outside of the Masakari container. Is this expected?
16:27:45 <noonedeadpunk> these are compute hosts
16:27:54 <ThiagoCMC> Damn... ok
16:29:43 <noonedeadpunk> I think this has been fixed with https://opendev.org/openstack/openstack-ansible-os_keystone/commit/4c9642765e632d409c6d9028aae26f4514af7033
16:32:11 <noonedeadpunk> uh, no, it's different probably...
16:32:22 <noonedeadpunk> but I can hardly understand what goes wrong here....
16:32:38 <ThiagoCMC> I'm wondering about the releasenotes for 22.0.0.0b1, where it says: "Support is added for deploying OpenStack on Ubuntu Focal (20.04) with source based installs. Ubuntu Cloud Archive is not available for Focal at this point so it is not possible to offer distro package based installs for Focal." - But Cloud Archive is available to focal!
16:34:28 <noonedeadpunk> yes but do not contain Ussuri packeges - I guess this reno is from U.
16:34:55 <ThiagoCMC> Hmmm... Got it
16:35:58 <jrosser> gshippey: any ideas on https://bugs.launchpad.net/openstack-ansible/+bug/1908703
16:36:00 <openstack> Launchpad bug 1908703 in openstack-ansible "federation domain not configured correct" [Undecided,New]
16:36:16 <ChiTo> Hi openstack-ansible team, happy new year!
16:36:29 <gshippey> thank you - was literally about to go look on eavesdrop!
16:36:58 <noonedeadpunk> nice to see you around gshippey :)
16:38:49 <jrosser> this task https://opendev.org/openstack/openstack-ansible-os_keystone/src/branch/master/tasks/keystone_federation_sp_idp_setup.yml#L124
16:39:07 <jrosser> takes trusted_id_list.domain_id
16:40:10 <jrosser> i think that federated_identities is dealt with completely somewhere else, so i'm really not sure what the bug is about
16:40:14 <ChiTo> I am trying to install Octavia for my cloud but I am hitting this issue: FAILED! => { "msg": "'dict object' has no attribute 'address'", am I missing some specific value? it is hit on the post-installation task
16:40:23 <jrosser> as the data and task highlighted do not seem to be related
16:41:03 <ChiTo> I have tried with vlan as well as flat types and both raise the error, my deployment does not use lxc, and it is installed on bare metal (i mean the neutron components and octavia ones)
16:41:26 <jrosser> ChiTo: can we look at this after the meeting?
16:41:51 <noonedeadpunk> pto is not around so...
16:42:01 <ChiTo> jrosser: Sure thing, thanks for your reply
16:42:12 <jrosser> i think we should query if the task is related to the data
16:42:24 <jrosser> becasue domain can be specified in several places
16:43:01 <noonedeadpunk> can I ask that somebody from team picked this up? As I have no experience in that area at all :(
16:44:29 <jrosser> sure i'll see if we can try to find out whats going on there
16:46:13 <noonedeadpunk> Thanks
16:46:21 <noonedeadpunk> next https://bugs.launchpad.net/openstack-ansible/+bug/1872169
16:46:22 <openstack> Launchpad bug 1872169 in openstack-ansible "train failed deploy setup_openstack.yaml " [Undecided,New]
16:46:22 <ThiagoCMC> I did something bad yesterday, where I configured the 'masakari-monitor_hosts:' with the controllers host/ip, now, even after cleaning it up the 'openstack_deploy/conf.d/masakari.yml' with just the compute nodes, the playbooks are still trying to install pacemaker/corosync at the controllers! I'm guessing that the inventory is now polluted, how to clean it up?
16:46:34 <noonedeadpunk> I guess it's because `"/openstack/venvs/rally-20.1.1.dev1/bin/pip", "install", "-U", "--constraint", "/openstack/venvs/rally-20.1.1.dev1/global-constraints.txt", "--constraint", "/openstack/venvs/rally-20.1.1.dev1/constraints.txt", "--log", "/var/log/python_venv_build.log", "--find-links", "http://localhost/os-releases/20.1.1.dev1/ubuntu-18.04-x86_64", "--trusted-host", "localhost", "pip", "setuptools", "wheel"`
16:48:19 <jrosser> connection refused to localhost:80
16:48:46 <noonedeadpunk> yeah, it should not be localhost
16:48:57 <noonedeadpunk> and we define it here https://opendev.org/openstack/openstack-ansible/src/branch/master/playbooks/defaults/source_install.yml#L27
16:49:56 <jrosser> i wonder why localhost is an inventory target, thats not right
16:50:31 <noonedeadpunk> it's not a target, but it's a utility host?
16:51:00 <noonedeadpunk> localhost is there because it's default for https://opendev.org/openstack/ansible-role-python_venv_build/src/branch/master/defaults/main.yml#L79
16:54:32 <ThiagoCMC> The Senlin endpoint on Victoria is being installed under "RegionOne", it's ignoring the configured 'service_region' name. Is there a quick fix available? I tried to declare 'senlin_service_region: "{{ service_region }}"' within 'inventory/group_vars/senlin.yml' but no effect.
16:55:10 <noonedeadpunk> I really don't see how this might happen even for 20.0.0
16:56:53 <noonedeadpunk> um, you should be able to set senlin_service_region jsut in user_variables
16:57:14 <noonedeadpunk> or openstack_deploy/group_vars/senlin_all.yml
16:57:47 <jrosser> if someone makes internal_lb_vip_address: localhost then that would go all kinds of wrong
16:58:11 <jrosser> perhaps for that bug we need to ask to either 1) join IRC and chat or 2) attach o_u_c.yml
16:58:13 <noonedeadpunk> I can't imagin somebody doing that... but....
16:58:20 <noonedeadpunk> +1
16:59:21 <jrosser> we should talk about pip resolver and also whatever is broken with debian right now
16:59:47 <noonedeadpunk> #topic office hours
17:00:03 <noonedeadpunk> yeah, my fault...
17:00:35 <noonedeadpunk> regarding debian - I read through, and I started deploying aio to look closer
17:01:42 <jrosser> ok thats cool i'd not got that far
17:01:53 <jrosser> backports repo is there by default in the CI images i think
17:02:08 <noonedeadpunk> we need it for the newer libvirt iirc
17:02:21 <jrosser> right, and thats where it all starts to go wrong
17:02:24 <noonedeadpunk> as otherwise nova fails
17:02:31 <noonedeadpunk> well, yes :(
17:02:44 <jrosser> i expect taking the paste i made and trying to apt-install the stuff off there might get you to the same error quickly
17:02:50 <jrosser> rather than run the whole AIO
17:03:29 <jrosser> i looked in the infra mirror and the libfdt1 seems to be there with a new enough version
17:08:34 <noonedeadpunk> uh https://opendev.org/openstack/openstack-ansible-openstack_hosts/src/branch/master/vars/debian.yml#L89
17:09:22 <noonedeadpunk> eventually need to end meeting I guess and continue to check out debian
17:09:25 <noonedeadpunk> #endmeeting