kleini | jrosser_: that was easy but still there are a lot of packages outdated in newly created containers. how can I upgrade them in the container base image? | 08:34 |
---|---|---|
jrosser_ | kleini: it depends what release you are on I think | 08:40 |
kleini | still bionic | 08:40 |
kleini | W on bionic | 08:40 |
jrosser_ | there is a script which prepares the container roots in the lxc_hosts role | 08:40 |
jrosser_ | hmmm W | 08:41 |
jrosser_ | kleini: i'm not sure how things are being out of date, because on W the rootfs should be built from scratch with this command https://github.com/openstack/openstack-ansible-lxc_hosts/blob/stable/wallaby/vars/ubuntu-18.04.yml#L16 | 08:46 |
jrosser_ | in previous releases a tarball was downloaded so that could easily be out of date | 08:47 |
jrosser_ | would be worth checking that the new way of building with debootstrap is actually working for you | 08:47 |
kleini | I saw the debootstrap process during lxc-hosts with cache refresh: true | 08:48 |
kleini | so it works | 08:48 |
jrosser_ | i wonder if it doesnt have the right apt repos configured inside the chroot? | 08:48 |
kleini | but especially with debootstrap it depends on the repository URLs, how actual the newly created image is. normally updates and security fixes are not part of debootstrap | 08:49 |
jrosser_ | i have to go now for today but hopefully this gives a starting point | 08:50 |
kleini | _ /bin/sh /usr/sbin/debootstrap --variant minbase bionic /var/lib/machines/ubuntu-bionic-amd64 | 08:50 |
kleini | jrosser_: thanks, that's a good starting point. will look on from there | 08:51 |
opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build master: Fix wheel build alert condition https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/823199 | 11:28 |
noonedeadpunk | jrosser_: jamesdenton really silly mistake has been made ^ | 11:29 |
opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build stable/xena: Verify that wheels are built while running against multiple hosts https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/823137 | 11:33 |
opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build stable/xena: Verify that wheels are built while running against multiple hosts https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/823137 | 11:33 |
opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build stable/wallaby: Verify that wheels are built while running against multiple hosts https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/823138 | 11:33 |
opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build stable/victoria: Verify that wheels are built while running against multiple hosts https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/823139 | 11:33 |
jamesdenton | noonedeadpunk whoops! | 13:37 |
jrosser_ | noonedeadpunk: what about https://opendev.org/openstack/openstack-ansible-os_tempest/src/branch/master/tasks/tempest_install.yml#L43 | 13:52 |
jrosser_ | feels like the wrong condition somehow | 13:52 |
noonedeadpunk | it shouldn't hit :D https://opendev.org/openstack/openstack-ansible/src/branch/master/playbooks/os-tempest-install.yml#L23 | 13:53 |
jrosser_ | maybe replicating the venv_build_host condition to there is better | 13:56 |
noonedeadpunk | um, not sure I got what you mean | 13:58 |
noonedeadpunk | check if no build host is present? | 13:59 |
noonedeadpunk | So what I was trying to do is to fail when wheels are not used when role is run across multiple hosts - like computes upgrade. Whatever why - build hosts not present or os-nova-install.yml launched with -e venv_wheel_build_enable=false | 14:01 |
noonedeadpunk | at least until we sort out the way how to properly have multiple os repo containers... | 14:02 |
noonedeadpunk | (probably with haproxy ACLs) | 14:02 |
jrosser_ | yep ok | 14:03 |
noonedeadpunk | as now it's really easier to just disable wheels while upgrading OS rather then sort out way with repo container :( and I bet ppl don't think to what consequences that might lead (ie opendev dos) | 14:05 |
jrosser_ | I will think about how we can do this better | 14:09 |
noonedeadpunk | The only idea I had were really ACLs... As we have specific path per distro, and we know what backend what version is running... | 14:10 |
noonedeadpunk | it would be still mess with lsyncd though... | 14:10 |
opendevreview | James Denton proposed openstack/openstack-ansible-os_neutron master: Change os_region to region_name https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/823169 | 23:42 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!