-
Notifications
You must be signed in to change notification settings - Fork 107
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
manifest: adapt for rpm-ostree modularity support #603
Conversation
Skipping CI for Draft Pull Request. |
Need to hold this until new rpm-ostree release with coreos/rpm-ostree#2760 hits cosa. |
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks really good! Awesome work!
Updated this, but now requires rpm-software-management/libdnf#1328. |
I'd like to get openshift/os#603 merged in a controlled fashion otherwise without it RHCOS builds will just break whenever rpm-ostree hits Fedora stable.
I'd like to get openshift/os#603 merged in a controlled fashion otherwise without it RHCOS builds will just break whenever rpm-ostree hits Fedora stable.
New cosa with rpm-software-management/libdnf#1328 is now built, so this is ready to go! |
/lgtm |
I think this is fine, but in the future we probably should have made this easier to ratchet in by adding something like |
Yeah... or I guess opt into the new behaviour via |
Hmm, CI failing with
so I'm guessing it's some stale image issue again. |
Yep I like that even more given it's in theory experimental. |
Yeah we mirror in Prow right now:
|
OK right, this is super confusing. The merge of coreos/coreos-assembler#2390 triggered both cosa-buildroot and coreos-assembler builds. But the latter uses |
/retest |
1 similar comment
/retest |
Hmm OK, so this is still using old rpm-ostree. Yet the |
/retest |
I think there's a further replication lag between the main CI cluster and the |
/retest-required Please review the full test history for this PR and help us cut down flakes. |
With the latest rpm-ostree support for modules, rpm-ostree will no longer pull in modular packages by default, which is something we've been relying on so far. So if we don't adapt, composes would start failing. This is a good thing, because it forces us to be more explicit wrt modularity: 1. enable the `container-tools:rhel8` module for the podman stack 2. enable the `virt:rhel` module for `qemu-guest-agent` While we're here, use the new `repo-packages` key to clarify some more of our intentions: 1. always pull in the `kernel` from BaseOS 2. always pull in `nss-altfiles` from AppStream 3. always pull in `toolbox` from RHAOS until we move over to the `container-tools:rhel8` one 4. always pull in `cri-o` and `conmon` from RHAOS to ensure they're synced All of the choices above should just be a reflection of our current stance. I've derived them from the various `exclude=` hacks we have in our repo files internally, and analyzing where we're currently getting what from. So this shouldn't significantly change our composes here. Though I will note that we now drop quite a few `Perl`-related modular packages, which I don't think we were pulling in intentionally. This then also allows us to drop all the `exclude=` hacks as well. Closes: coreos/rpm-ostree#3035
Ahhh
I was so focused on composes that I forgot about extensions entirely. Anyway, coreos/rpm-ostree#3095 adds support for it. Should be able to just backport that patch and tag it into the continuous tag. Or cool too if you'd rather implement the |
/retest |
/lgtm |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: cgwalters, jlebon, travier The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
NOTE: After this merges we'll require an updated coreos-assembler with newer rpm-ostree. |
…021.10 The internal RHCOS pipeline is rebuilding cosa due to multiarch, and there was a subtle difference that crept in between the two `Dockerfile`s; notably the main one wasn't configuring the `-continuous` repo where we'd bypassed the manual Bodhi stuff to fast track a new `rpm-ostree`. We need that for openshift/os#603 Sync the two `Dockerfile`s and also require the updated version.
…021.10 The internal RHCOS pipeline is rebuilding cosa due to multiarch, and there was a subtle difference that crept in between the two `Dockerfile`s; notably the main one wasn't configuring the `-continuous` repo where we'd bypassed the manual Bodhi stuff to fast track a new `rpm-ostree`. We need that for openshift/os#603 Sync the two `Dockerfile`s and also require the updated version.
…021.10 The internal RHCOS pipeline is rebuilding cosa due to multiarch, and there was a subtle difference that crept in between the two `Dockerfile`s; notably the main one wasn't configuring the `-continuous` repo where we'd bypassed the manual Bodhi stuff to fast track a new `rpm-ostree`. We need that for openshift/os#603 Sync the two `Dockerfile`s and also require the updated version.
With the latest rpm-ostree support for modules, rpm-ostree will no
longer pull in modular packages by default, which is something we've
been relying on so far. So if we don't adapt, composes would start
failing. This is a good thing, because it forces us to be more explicit
wrt modularity:
container-tools:rhel8
module for the podman stackvirt:rhel
module forqemu-guest-agent
While we're here, use the new
repo-packages
key to clarify some moreof our intentions:
kernel
from BaseOSnss-altfiles
from AppStreamtoolbox
from RHAOS until we move over to thecontainer-tools:rhel8
onecri-o
andconmon
from RHAOS to ensure they'resynced
All of the choices above should just be a reflection of our current
stance. I've derived them from the various
exclude=
hacks we have inour repo files internally, and analyzing where we're currently getting
what from. So this shouldn't significantly change our composes here.
Though I will note that we now drop quite a few
Perl
-related modularpackages, which I don't think we were pulling in intentionally.
This then also allows us to drop all the
exclude=
hacks as well.Closes: coreos/rpm-ostree#3035