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
rawhide: systemd-254 drops ostree-prepare-root from transaction #1527
Comments
It seems to be something in |
The source for that is in:
Not really sure what would have changed here. Apparently reverting to the older version of systemd makes tests pass again so something is going on. |
Yeah, but I think somebody who groks |
actually looking at the log block you copy/pasted it was actually On a F38 system I see a bunch of mount units in the initrd:
but on rawhide with the new systemd I don't:
|
That's after If you have a shell in this environment, I'd try just invoking e.g. |
Looks like maybe that is the problem:
Yes, this seems to work:
|
Some notes from realtime debugging
We still believe there's likely some systemd bug here, but we're not sure what it is yet. |
With that patch we get:
|
I know from
|
Started on some work for this in coreos/fedora-coreos-config#2515 |
@dustymabe points out that the problem I'm seeing in openQA with the DVD package repo not being used on DVD installs may well be related (as they're both about mounts not working) - https://bugzilla.redhat.com/show_bug.cgi?id=2223795 . |
I played around with the ISO image and the /run/install/* mounts are there in initramfs (when inspecting the state after |
upstream report: systemd/systemd#28452 |
Fixed in systemd/systemd@2bfe726 which is in v254 and v254-rc3 so we should be able to unpin this because rawhide is up to systemd-254~rc3-1.fc39 @aaradhak can you take care of that? |
sure , will unpin systemd in rawhide |
The feature that caused us to pin systemd has been fixed and is up to https://koji.fedoraproject.org/koji/buildinfo?buildID=2259554. The fix : systemd/systemd@2bfe726 systemd v254: https://github.com/systemd/systemd/releases/tag/v254 systemd v254-rc3: https://github.com/systemd/systemd/releases/tag/v254-rc3 Resolves: coreos/fedora-coreos-tracker#1527
The feature that caused us to pin systemd has been fixed and is up to https://koji.fedoraproject.org/koji/buildinfo?buildID=2259554. The fix : systemd/systemd@2bfe726 systemd v254: https://github.com/systemd/systemd/releases/tag/v254 systemd v254-rc3: https://github.com/systemd/systemd/releases/tag/v254-rc3 Ref: coreos/fedora-coreos-tracker#1527
I'm still seeing the system get dropped into emergency.target so I don't think the "fix" is sufficient for our uses yet. |
FWIW, this did fix the "DVD installs do not use repo from DVD" problem in the installer, so it's not like it did nothing. |
Hmpf, that's strange, I'll try to look into that. |
@mrc0mmand - if you'd like to screenshare we can poke around on a running system. Find me in |
Some more logging shows a lot of complaining that
This is with |
Ultimately for Fedora CoreOS what we're doing in the initramfs is just extremely dynamic...we support reprovisioning the root filesystem from the initramfs, setting up LUKS, doing networking etc. So for us, Adding to the complexity is the way the reprovisioning works, we today mount the filesystem, then copy it to RAM, etc. But for that part in theory I think because we're using a mount namespace systemd shouldn't see a I do think coreos/fedora-coreos-config#2515 is probably the right direction - basically, it may make sense to have a |
I built an FCOS image with the Rawhide configuration (sans the pinned systemd packages), and it ends up in emergency shell during the first boot. The main difference I see, so far, is that nothing pulls in Rawhide +
Rawhide +
Leading to:
Hopefully a bisect will pinpoint the culprit. |
Ah, latest Rawhide:
vs the working systemd build:
Which points to systemd/systemd@89e9df1 introduced in systemd/systemd#27852. |
The feature that caused us to pin systemd has been fixed and is up to https://koji.fedoraproject.org/koji/buildinfo?buildID=2259554. The fix: systemd/systemd@2bfe726 systemdv254: https://github.com/systemd/systemd/releases/tag/v254 systemdv254-rc3: https://github.com/systemd/systemd/releases/tag/v254-rc3 Ref: coreos/fedora-coreos-tracker#1527
The feature that caused us to pin systemd has been fixed and is up to https://koji.fedoraproject.org/koji/buildinfo?buildID=2259554. The fix: systemd/systemd@2bfe726 systemdv254: https://github.com/systemd/systemd/releases/tag/v254 systemdv254-rc3: https://github.com/systemd/systemd/releases/tag/v254-rc3 Ref: coreos/fedora-coreos-tracker#1527
In e8eb4cb we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in xxx This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
In 45396d8 we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in ignition-ostree-mount-sysroot.sh. This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
Thanks @mrc0mmand that investigation really helped us figure out the change was a desired change and we needed to adjust. I opened a new PR that builds on top of Colin's PR to address this: coreos/fedora-coreos-config#2543 |
In 45396d8 we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in ignition-ostree-mount-sysroot.sh. This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
In 45396d8 we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in ignition-ostree-mount-sysroot.sh. This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
This should be fixed with coreos/fedora-coreos-config#2543 |
In 45396d8 we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in ignition-ostree-mount-sysroot.sh. This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
In 45396d8 we started creating a static sysroot.mount to appease systemd 254+ (see [1]), but it missed the `coreos-rootflags` handling that was present in ignition-ostree-mount-sysroot.sh. This PR adds back in the dynamic rootflags by using an EnvironmentFile generated by a coreos-rootflags.service to get the same behavior. [1] coreos/fedora-coreos-tracker#1527 (comment)
Describe the bug
The kola basic tests are failing on the latest rawhide x86_64 builds since build 39.20230716.91.0
Observed on checking the console logs that these tests began failing after the systemd version upgrade from
systemd 253.5-6.fc39 -> systemd 254~rc2-1.fc39
console.log output related to systemd:
This is the kola basic output from one of the latest rawhide builds:
kola-basic-7146e04d.zip
Reproduction steps
Expected behavior
kola basic test to pass with upgraded systemd-254~rc2-1.fc39
Actual behavior
kola basic test fails with systemd-254~rc2-1.fc39
System details
Fedora CoreOS latest x86_64 Rawhide Version - https://builds.coreos.fedoraproject.org/browser?stream=rawhide&arch=x86_64
Butane or Ignition config
No response
Additional information
No response
The text was updated successfully, but these errors were encountered: