You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
During install of IoT Provisioner to a mmcblk device, coreos-installer-growfs.service will fail (without log output). I believe this applies equally to nvme devices. I think the root cause is an issue in coreos-installer-dracut (coreos/coreos-installer-dracut#35) which got fixed upstream, but never packaged for Fedora.
To Reproduce
Start the Fedora IoT via UEFI HTTP boot with a grub entry like linuxefi /images/pxeboot/vmlinuz rd.neednet=1 coreos.inst.crypt_root=1 coreos.inst.install_dev=/dev/mmcblk0 coreos.inst.image_url=http://<SERVER_HOST>/image.raw.xz coreos.inst.insecure quiet coreos.inst.append=ignition.config.url=http://<SERVER_HOST>/dell.ign coreos.inst.append=rd.neednet=1
Wait for install to finish
Install never finishes.
I've tried figuring out a repro for libvirt, but mmcblk seems to be unsupported and nvme proved annoying to set up.
Expected behavior
Install should finish successfully with a shell/powered off/rebooted depending on cmdline.
Screenshots
Log output (best-effort transcription)
[TIMESTAMP] coreos-installer-service[PID]: Modifying kernel arguments
[TIMESTAMP] coreos-installer-service[PID]: Install complete.
Failed to start coreos-installer-growfs.service - Grow root filesystem.
Dependency failed for initrd.target - Initrd Default Target.
[TIMESTAMP] coreos-installer-service[PID]: coreos-installer install <...>
[TIMESTAMP] coreos-installer-service[PID]: Downloading image from [SERVER_HOST]
...
[TIMESTAMP] coreos-installer-service[PID]: Partitions in use on /dev/mmcblk0
[TIMESTAMP] coreos-installer-service[PID]: /dev/mmcblk0p3 in use by /dev/dm-0
this continues to loop (forever?).
OS version:
Fedora-IoT-provisioner-41-20241027.0.x86_64
Additional context
Originally this bug was reported in RH #2232406 for RHEL Edge and got fixed (in the RH package repo?). Apparently this fix didn't make it into the Fedora packages. I created https://bugzilla.redhat.com/show_bug.cgi?id=2328530 for Fedora specifically.
The text was updated successfully, but these errors were encountered:
Sebmaster
changed the title
IoT Provisioner infinitely loops during install to /dev/mmcblk0
IoT Provisioner infinitely loops during install to /dev/mmcblk devices
Dec 4, 2024
It's a sad state of affairs that the last change to coreos-installer-dracut that landed in Fedora is from April 2023 (coreos/coreos-installer-dracut@a4be31d)
Describe the bug
During install of IoT Provisioner to a mmcblk device, coreos-installer-growfs.service will fail (without log output). I believe this applies equally to nvme devices. I think the root cause is an issue in coreos-installer-dracut (coreos/coreos-installer-dracut#35) which got fixed upstream, but never packaged for Fedora.
To Reproduce
linuxefi /images/pxeboot/vmlinuz rd.neednet=1 coreos.inst.crypt_root=1 coreos.inst.install_dev=/dev/mmcblk0 coreos.inst.image_url=http://<SERVER_HOST>/image.raw.xz coreos.inst.insecure quiet coreos.inst.append=ignition.config.url=http://<SERVER_HOST>/dell.ign coreos.inst.append=rd.neednet=1
I've tried figuring out a repro for libvirt, but mmcblk seems to be unsupported and nvme proved annoying to set up.
Expected behavior
Install should finish successfully with a shell/powered off/rebooted depending on cmdline.
Screenshots
Log output (best-effort transcription)
this continues to loop (forever?).
OS version:
Fedora-IoT-provisioner-41-20241027.0.x86_64
Additional context
Originally this bug was reported in RH #2232406 for RHEL Edge and got fixed (in the RH package repo?). Apparently this fix didn't make it into the Fedora packages. I created https://bugzilla.redhat.com/show_bug.cgi?id=2328530 for Fedora specifically.
The text was updated successfully, but these errors were encountered: