Description
===========
Job https://zuul.opendev.org/t/openstack/job/tripleo-ci-centos-9-scenario000-multinode-oooq-container-updates is suppose to be running containers updates on overcloud.
In tripleo-upgrade container check change https://review.opendev.org/c/openstack/tripleo-upgrade/+/872592/12 , which compares container images before overcloud ovn update run(before overcloud update) and before reboot(after overcloud update), we have discovered that containers are updated from the same tag to the same tag.
The content of files used in check can be seen below:
container-ps-before_reboot.txt
clustercheck aa83fb96012e 192.168.24.1:8787/tripleomastercentos9/openstack-mariadb:67316b1caba6cecebf7f78764dcc361c
2galera-bundle-podman-0 88a5a63ea012 192.168.24.1:8787/tripleomastercentos9/openstack-mariadb:pcmklatest
3haproxy-bundle-podman-0 854576d5a98a 192.168.24.1:8787/tripleomastercentos9/openstack-haproxy:pcmklatest
4keystone_cron b136f2daef99 192.168.24.1:8787/tripleomastercentos9/openstack-keystone:67316b1caba6cecebf7f78764dcc361c
5keystone e0d237d553a0 192.168.24.1:8787/tripleomastercentos9/openstack-keystone:67316b1caba6cecebf7f78764dcc361c
6memcached 481d3d388bfe 192.168.24.1:8787/tripleomastercentos9/openstack-memcached:67316b1caba6cecebf7f78764dcc361c
7rabbitmq-bundle-podman-0 b171403303bb 192.168.24.1:8787/tripleomastercentos9/openstack-rabbitmq:pcmklatest
container-ps-before_ovn_controller_update.txt
clustercheck aa83fb96012e 192.168.24.1:8787/tripleomastercentos9/openstack-mariadb:67316b1caba6cecebf7f78764dcc361c
2galera-bundle-podman-0 ebeaef1a4a3f 192.168.24.1:8787/tripleomastercentos9/openstack-mariadb:pcmklatest
3haproxy-bundle-podman-0 503f451f8723 192.168.24.1:8787/tripleomastercentos9/openstack-haproxy:pcmklatest
4keystone_cron b136f2daef99 192.168.24.1:8787/tripleomastercentos9/openstack-keystone:67316b1caba6cecebf7f78764dcc361c
5keystone e0d237d553a0 192.168.24.1:8787/tripleomastercentos9/openstack-keystone:67316b1caba6cecebf7f78764dcc361c
6memcached 481d3d388bfe 192.168.24.1:8787/tripleomastercentos9/openstack-memcached:67316b1caba6cecebf7f78764dcc361c
7rabbitmq-bundle-podman-0 166936af2d93 192.168.24.1:8787/tripleomastercentos9/openstack-rabbitmq:pcmklatest
In both files the same tag for container images is used: 67316b1caba6cecebf7f78764dcc361c. This indicates that container images update is skipped for some reason.
Expected result
===============
Containers on overcloud are updated from one tag to another.
Adding some more context on the CI side - agree this is no longer working as it was intended to.
This upstream update job was taking current-tripleo and updating to the latest current tag.
You can still see this mechanism from logs e.g.: from the emit_releases_file [1]
output releases: ... 'overcloud_ deploy_ hash': '514d9f3921ea13 5d3a304d382c08d 7e6', 'overcloud_ target_ release' : 'wallaby', 'overcloud_ target_ hash': '744609c1eade3e 921c5a4ee08d33d b37', /trunk. rdoproject. org/centos9- wallaby/ current- tripleo/ 51/4d/ current-tripleo and https:/ /trunk. rdoproject. org/centos9- wallaby/ current/ 74/46/ current
where https:/
After your ping today about the content providers it has become clearer that this probably fell through the cracks when we moved to the content provider model.
So one of the things we need is to adjust the content providers to provider the current build too.
[1] https:/ /9f8cf072dd722b 3de279- df315ac4bc3fab6 cd31bd9cee1db3c 46.ssl. cf1.rackcdn. com/879179/ 4/gate/ tripleo- ci-centos- 9-scenario000- multinode- oooq-container- updates/ 41177b0/ logs/quickstart _files/ emit_releases_ file.log