-
Notifications
You must be signed in to change notification settings - Fork 157
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
Reapply "manifests: enable cliwrap on Fedora 40+" #2887
Conversation
This reverts commit 3789b06. The reported issue that motivated the revert[[1]] is due to an rpm-ostree bug[[2]]. We can trivially work around this bug until the rpm-ostree fix lands in FCOS, so let's do that and re-enable cliwrap. [1]: coreos/fedora-coreos-tracker#1679 [2]: coreos/rpm-ostree#4848
Doing this means we are knowingly shipping a bug to anything on F40+? Normally this might be fine, but we just entered beta freeze which means we potentially ship this in `next`. I’m not super comfortable with that.
I think it would be better to get the fix back ported to rpm-ostree in Fedora and then fast track it to all branches that are on F40+?
…On Wed, Feb 28, 2024, at 21:33, Jonathan Lebon wrote:
This reverts commit3789b06.
The reported issue that motivated the revert[1 <coreos/fedora-coreos-tracker#1679>] is due to an rpm-ostree bug[2 <coreos/rpm-ostree#4848>].
We can trivially work around this bug until the rpm-ostree fix lands in FCOS, so let's do that and re-enable cliwrap.
You can view, comment on, or merge this pull request online at:
#2887
Commit Summary
• 3ce2f77 <3ce2f77> Reapply "manifests: enable cliwrap on Fedora 40+"
File Changes
(4 files <https://github.com/coreos/fedora-coreos-config/pull/2887/files>)
• *A* manifests/cliwrap.yaml <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-58c8610e2193f25aa642ccdd8553d6dc9351b66629fabbc13ce85eb8d652a502> (2)
• *M* manifests/fedora-coreos.yaml <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-7a0901e1ea4a3c8647549674fea7574437d9242945ff65ca49cd73819761f9de> (2)
• *M* tests/kola/extensions/package <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-898decc50ad2b84e074c61ca72c987c37327d8ac904f2953ccc85991d5754409> (6)
• *M* tests/kola/rpm-ostree/kernel-replace <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-ffd616bc9607b8abd660319a3679b1afa2d949dc673e752ffd58a6160eaa7a1e> (3)
Patch Links:
• https://github.com/coreos/fedora-coreos-config/pull/2887.patch
• https://github.com/coreos/fedora-coreos-config/pull/2887.diff
—
Reply to this email directly, view it on GitHub <#2887>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCR63VM7HRUGOKYVOKESLTYV7SHXAVCNFSM6AAAAABD7D7K62VHI2DSMVQWIX3LMV43ASLTON2WKOZSGE3DAMRTGM2DSOI>.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
|
An option here would be to just not workaround the issue in the test and just denylist that test on branched and rawhide. This would protect against us shipping a bug in our prod streams but would set a timer on getting rpm-ostree fixed.
…On Thu, Feb 29, 2024, at 06:22, Dusty Mabe wrote:
Doing this means we are knowingly shipping a bug to anything on F40+? Normally this might be fine, but we just entered beta freeze which means we potentially ship this in `next`. I’m not super comfortable with that.
I think it would be better to get the fix back ported to rpm-ostree in Fedora and then fast track it to all branches that are on F40+?
On Wed, Feb 28, 2024, at 21:33, Jonathan Lebon wrote:
>
>
> This reverts commit3789b06.
>
>
> The reported issue that motivated the revert[1 <coreos/fedora-coreos-tracker#1679>] is due to an rpm-ostree bug[2 <coreos/rpm-ostree#4848>].
>
> We can trivially work around this bug until the rpm-ostree fix lands in FCOS, so let's do that and re-enable cliwrap.
>
>
> You can view, comment on, or merge this pull request online at:
>
> #2887
>
> Commit Summary
>
> • 3ce2f77 <3ce2f77> Reapply "manifests: enable cliwrap on Fedora 40+"
> File Changes
>
> (4 files <https://github.com/coreos/fedora-coreos-config/pull/2887/files>)
>
> • *A* manifests/cliwrap.yaml <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-58c8610e2193f25aa642ccdd8553d6dc9351b66629fabbc13ce85eb8d652a502> (2)
> • *M* manifests/fedora-coreos.yaml <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-7a0901e1ea4a3c8647549674fea7574437d9242945ff65ca49cd73819761f9de> (2)
> • *M* tests/kola/extensions/package <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-898decc50ad2b84e074c61ca72c987c37327d8ac904f2953ccc85991d5754409> (6)
> • *M* tests/kola/rpm-ostree/kernel-replace <https://github.com/coreos/fedora-coreos-config/pull/2887/files#diff-ffd616bc9607b8abd660319a3679b1afa2d949dc673e752ffd58a6160eaa7a1e> (3)
> Patch Links:
>
> • https://github.com/coreos/fedora-coreos-config/pull/2887.patch
> • https://github.com/coreos/fedora-coreos-config/pull/2887.diff
>
>
> —
> Reply to this email directly, view it on GitHub <#2887>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCR63VM7HRUGOKYVOKESLTYV7SHXAVCNFSM6AAAAABD7D7K62VHI2DSMVQWIX3LMV43ASLTON2WKOZSGE3DAMRTGM2DSOI>.
> You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
>
>
|
Yes, this needs to be fixed in RHCOS too. I wasn't planning on immediately backporting everywhere in case more bugs show up. But I did plan to eventually make sure it's fixed before f40 GA, whether by backporting or just because we've cut a new release. (Notice I didn't mark this PR as closing coreos/fedora-coreos-tracker#1679 :) ). Not too concerned about leaving this in Beta for now. Guess we could denylist, yeah... Though kernel replacement in the layering flow is a important feature. I don't want us to lose coverage of that in f40+ in the meantime. |
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.
LGTM
FWIW (as the person who pushed for cliwrap and wrote a lot of the code) I now consider it a technological dead end and also a giant social mistake (effectively agreeing with initial feedback!) The big technology problem here is that upgrading a package in a container build (whether via rpm or dnf or rpm-ostree if it ever supported |
Yikes, yes that's not great. I'm still also not a big fan of the approach overall. The main thing I'm interested in really is trying to make |
Hmm but actually, the confusing thing here is that we definitely do at least need the I don't see a kernel replacement example in https://gitlab.com/bootc-org/examples and https://github.com/coreos/layering-examples/blob/main/replace-kernel/Containerfile still includes an explicit |
yeah cc coreos/rpm-ostree#4726
Yeah, would probably make sense except do note it comes from systemd and is hence pretty likely to get replaced when upgrading systemd |
All we really want to cliwrap is `dnf` so that a `dnf install ...` will work. Again, as mentioned in the commit that introduced this, ideally we can actually ship dnf itself once it's ready and then we can back this out. Also wrap `kernel-install` for now to make kernel replacements smoother until we can get it to work seamlessly (this matches [[1]]). [1]: CentOS/centos-bootc#377
OK, I changed the proposal here to only wrap
Done in CentOS/centos-bootc#377. |
This reverts commit ff686c6. With cliwrap, rpm-ostree runs rpm with less privileges. We special-case `--query`, so e.g. `rpm -qf` still runs with dropped privileges. But anyway, we actually want to reduce this to wrapping *just* `dnf` for now (see discussions in coreos/fedora-coreos-config#2887). Resolves: https://issues.redhat.com/browse/OCPBUGS-30149
Retracting this. Will open a tracker issue with another proposal. |
This reverts commit ff686c6. With cliwrap, rpm-ostree runs rpm with less privileges. We special-case `--query`, so e.g. `rpm -qf` still runs with dropped privileges. But anyway, we actually want to reduce this to wrapping *just* `dnf` for now (see discussions in coreos/fedora-coreos-config#2887). Resolves: https://issues.redhat.com/browse/OCPBUGS-30149
This reverts commit 3789b06.
The reported issue that motivated the revert[1] is due to an rpm-ostree bug[2].
We can trivially work around this bug until the rpm-ostree fix lands in FCOS, so let's do that and re-enable cliwrap.