| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Add multi-AZ deployment configuration sample https://review.opendev.org/c/openstack/openstack-ansible/+/939609 | 09:28 |
|---|---|---|
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Add multi-AZ deployment configuration sample https://review.opendev.org/c/openstack/openstack-ansible/+/939609 | 09:41 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [doc] Add multi-AZ deployment configuration sample https://review.opendev.org/c/openstack/openstack-ansible/+/939609 | 10:13 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Molecule to respect depends-on for test-requirements update https://review.opendev.org/c/openstack/openstack-ansible/+/939290 | 10:23 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add noble to molecule testing https://review.opendev.org/c/openstack/openstack-ansible/+/939306 | 10:23 |
| noonedeadpunk | fwiw, ansible-compat 25.0.0 was dropped from pypi: https://pypi.org/project/ansible-compat/25.0.0/ | 10:24 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add httpd role to a-r-r https://review.opendev.org/c/openstack/openstack-ansible/+/938271 | 10:24 |
| jrosser | hmm interesting | 10:24 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Ensure repo_server proper username https://review.opendev.org/c/openstack/openstack-ansible/+/938275 | 10:24 |
| jrosser | i wonder where the discussion about that is happening | 10:24 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Return upgrade jobs back to voting https://review.opendev.org/c/openstack/openstack-ansible/+/939307 | 10:24 |
| jrosser | anyway thats good - we have been wedged up for 1 week now on master branch, so we can start rechecking stuff again | 10:26 |
| noonedeadpunk | not in IRC for sure | 10:26 |
| noonedeadpunk | I also spent weekend on documenting AZ setup: https://zuul.opendev.org/t/openstack/build/4755b757ba944c9f80a40404e811f408/artifacts | 10:27 |
| noonedeadpunk | it includes also some smaller examples, which I guess could be avoided... But also I realized we don't have any example showing how to do them all together. | 10:28 |
| noonedeadpunk | to it's quite big one.... | 10:28 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-repo_server master: Use standalone httpd role https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/938270 | 10:39 |
| jrosser | i wonder if we should follow up this https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/939599 to remove the need for a privileged port | 10:44 |
| jrosser | to allow global use of priv ports on potentially a metal infra host when we could just move the backends to 8443/8080 is surprising | 10:47 |
| jrosser | this is likely from the past when direct access to the horizon backend by the user might be expected | 10:48 |
| opendevreview | Andrew Bonney proposed openstack/openstack-ansible-os_ironic master: Fix quorum queue support for ironic-inspector https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/939544 | 10:53 |
| opendevreview | Merged openstack/openstack-ansible master: Update prod example to reflect OVN setup https://review.opendev.org/c/openstack/openstack-ansible/+/938981 | 11:03 |
| f0o | I'm just ever so slightly confused by the plethora of "action required" emails haha | 11:31 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-role-systemd_networkd master: DNM - test vars inclusion https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/939628 | 12:26 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-haproxy_server master: Make sysctl configuration path configurable Defaults to /etc/sysctl.conf to retain current behavior https://review.opendev.org/c/openstack/openstack-ansible-haproxy_server/+/939601 | 12:27 |
| opendevreview | Merged openstack/openstack-ansible-os_swift master: Make sysctl configuration path configurable Defaults to /etc/sysctl.conf to retain current behavior https://review.opendev.org/c/openstack/openstack-ansible-os_swift/+/939603 | 12:57 |
| noonedeadpunk | f0o: yeah, it looks like weve introduced couple of issues lately | 13:00 |
| noonedeadpunk | but gerrit is quite noisy indeed | 13:01 |
| opendevreview | Merged openstack/openstack-ansible-repo_server stable/2023.2: Fix tags usage for the role https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/939051 | 13:01 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-ops master: Upgrade magnum-cluster-api to 0.24.2 https://review.opendev.org/c/openstack/openstack-ansible-ops/+/936229 | 13:02 |
| f0o | noonedeadpunk: no worries; I'm running my changes happily locally | 13:07 |
| f0o | if there _IS_ something I need to address, feel free to shout at me ;) | 13:07 |
| noonedeadpunk | sure thing, but it looks like issues are on our side for now | 13:14 |
| opendevreview | Merged openstack/openstack-ansible-os_horizon master: Make sysctl configuration path configurable Defaults to /etc/sysctl.conf to retain current behavior https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/939599 | 13:21 |
| jrosser | huh | 13:41 |
| jrosser | so why only warnings here for rabbitmq and galera https://zuul.opendev.org/t/openstack/build/86a170c3e142470cb343e20ecac59f2a/log/job-output.txt#20458-20504 | 13:41 |
| jrosser | and how is this possible https://zuul.opendev.org/t/openstack/build/86a170c3e142470cb343e20ecac59f2a/log/job-output.txt#20973-21119 | 13:42 |
| jrosser | _systemd_networkd_distro_packages is a list with values, but systemd_networkd_distro_packages somehow ends up as [] | 13:43 |
| jrosser | aaaaaaahhhhhhhh https://opendev.org/openstack/openstack-ansible-lxc_container_create/src/branch/master/tasks/lxc_container_network.yml#L22 | 13:45 |
| jrosser | noonedeadpunk: so we either need to remove that ^ or make the service restart somehow conditional | 13:46 |
| noonedeadpunk | I'm really not sure why we have that | 13:56 |
| noonedeadpunk | oh | 13:56 |
| noonedeadpunk | I know.... | 13:57 |
| noonedeadpunk | these should be part of the image | 13:57 |
| noonedeadpunk | not to spend time on installation during the runtime | 13:57 |
| noonedeadpunk | so pretty much we need to extend this list then: https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/vars/debian.yml#L48 | 13:57 |
| noonedeadpunk | jrosser: ^ | 13:58 |
| jrosser | ok - maybe we do both then? extend the list and remove that line? | 13:58 |
| jrosser | becasue the behaviour is very counterintuitive with that list set externally to [] | 13:59 |
| noonedeadpunk | the thing is that we skip whole this block for all containers: https://opendev.org/openstack/ansible-role-systemd_networkd/src/branch/master/tasks/main.yml#L32-L100 | 14:00 |
| noonedeadpunk | even though it will be mostly skipped individually (except package), it still should be faster to skip the block | 14:01 |
| noonedeadpunk | and given the multiplier.... | 14:01 |
| jrosser | so but the issue is that the restart handler still tries to restart all the services defined in vars/<os>.yml | 14:02 |
| jrosser | and adjusting the base image package list is somewhat complicated because of systemd_resolved being optional | 14:02 |
| opendevreview | Merged openstack/ansible-hardening master: Make sysctl configuration path configurable Defaults to /etc/sysctl.conf to retain current behavior https://review.opendev.org/c/openstack/ansible-hardening/+/939600 | 14:03 |
| jrosser | and even if we did that successfully it would not fix these upgrade jobs | 14:03 |
| jrosser | as the contaienr is not rebuilt from a new image during the upgrade | 14:03 |
| noonedeadpunk | we have some mapping already? https://opendev.org/openstack/openstack-ansible-lxc_hosts/src/branch/master/vars/debian.yml#L63 | 14:03 |
| jrosser | ah ok, sire | 14:04 |
| jrosser | *sure | 14:04 |
| noonedeadpunk | I wonder if we should rebuild the container image in run_upgrade | 14:04 |
| jrosser | we would need to do a rolling destroy/create too | 14:04 |
| noonedeadpunk | doh | 14:05 |
| noonedeadpunk | true | 14:05 |
| jrosser | so yes absolutely we can update the base image packages, that makes total sense | 14:05 |
| jrosser | but thats not the same as for an in-place upgrade | 14:05 |
| noonedeadpunk | and actually jobs are not failing for non-upgrade, do they? | 14:05 |
| jrosser | those work, its just an upgrade that fails | 14:06 |
| noonedeadpunk | so I'm looking now at 2024.1 and I hardly see a difference? | 14:06 |
| jrosser | theres an example here with my added debug https://review.opendev.org/c/openstack/openstack-ansible-haproxy_server/+/939601?tab=change-view-tab-header-zuul-results-summary | 14:06 |
| noonedeadpunk | ah, udev is missing | 14:07 |
| jrosser | yeah, becasue its not in the base image (in the 'upgrade from' branch) and thr package list for the networkd role is [] | 14:07 |
| jrosser | maybe we need to split the handler into two parts | 14:08 |
| jrosser | one to restart networkd | 14:08 |
| noonedeadpunk | but well. how does then networkd works there at all | 14:08 |
| jrosser | and the other part to restart other things, but only if systemd_networkd_distro_packages is not [] | 14:08 |
| noonedeadpunk | as systemd-udevd was kinda crucial to set an IP to interfaces | 14:08 |
| jrosser | in docker ^ | 14:09 |
| noonedeadpunk | yes | 14:09 |
| jrosser | i think that we have made a fix for the molecule case, which would be working anyway in lxc | 14:09 |
| noonedeadpunk | though I thought it would be working for LXC , as LXC does have a service running anyway | 14:10 |
| noonedeadpunk | which is apparently not true | 14:10 |
| jrosser | wierd isnt it | 14:10 |
| noonedeadpunk | I'm just trying to think if it is this, which should be conditional: https://opendev.org/openstack/ansible-role-systemd_networkd/src/branch/master/handlers/main.yml#L16-L25 | 14:12 |
| jrosser | i wonder if we have some difference between udev and systemd-udevd | 14:13 |
| jrosser | becasue we install udev, but the thing that fails is restarting systemd-udevd | 14:13 |
| noonedeadpunk | um, yes, but also depends on debian/ubuntu version | 14:13 |
| noonedeadpunk | as jammy and debian 11 had systemd provided with a single package | 14:13 |
| noonedeadpunk | and 24.04/12 have it split into pieces | 14:14 |
| jrosser | i was wondering if the serivce name needs to depend on the os version | 14:14 |
| noonedeadpunk | as starting udev would fail/block systemd-udevd eecution at least in docker until reboot | 14:14 |
| noonedeadpunk | the problem is - they're valid in non-upgrade (because it's already in the image) and in docker... | 14:15 |
| noonedeadpunk | so could be systemd was not relying on udev that much before... | 14:16 |
| noonedeadpunk | dunno | 14:16 |
| jrosser | i'm just looking at a jammy deploy | 14:16 |
| noonedeadpunk | just pretty much we need to handle an upgrade scenario | 14:16 |
| jrosser | theres no udev installed at all | 14:16 |
| jrosser | in my keystone container, as example | 14:16 |
| noonedeadpunk | yeah, so it could be not needed | 14:16 |
| noonedeadpunk | and then it's the handler who should be conditional | 14:17 |
| noonedeadpunk | or at least move list of services to vars | 14:17 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-role-systemd_networkd master: Only restart non-networkd services when the role is configured to install them https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/939640 | 14:27 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-haproxy_server master: Make sysctl configuration path configurable Defaults to /etc/sysctl.conf to retain current behavior https://review.opendev.org/c/openstack/openstack-ansible-haproxy_server/+/939601 | 14:28 |
| jrosser | that feels like the simplest change and also matching the behaviour of testing the distro package length elsewhere | 14:29 |
| noonedeadpunk | actually it's true.... | 14:34 |
| noonedeadpunk | despite being counter-intuitive as well. as if services installed/running we don't need to start them | 14:35 |
| noonedeadpunk | and likely we can drop that after 2025.1 at all | 14:35 |
| opendevreview | Jonathan Rosser proposed openstack/ansible-role-systemd_networkd master: Add test coverage for upgrades https://review.opendev.org/c/openstack/ansible-role-systemd_networkd/+/939641 | 14:43 |
| opendevreview | Merged openstack/openstack-ansible stable/2024.2: Use the same version of ansible for linters as is used for deployments. https://review.opendev.org/c/openstack/openstack-ansible/+/939356 | 15:26 |
| opendevreview | Merged openstack/openstack-ansible master: Add httpd role to a-r-r https://review.opendev.org/c/openstack/openstack-ansible/+/938271 | 16:00 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_horizon master: Remove use of async when compressing static files https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/927953 | 16:40 |
| opendevreview | Merged openstack/openstack-ansible master: Ensure repo_server proper username https://review.opendev.org/c/openstack/openstack-ansible/+/938275 | 18:59 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-repo_server master: Use standalone httpd role https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/938270 | 19:02 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!