On Tue, Aug 29, 2023 at 6:15 PM stan via users users@lists.fedoraproject.org wrote:
Thanks, Stan. I guess the files are almost similar, since they have almost the same size:
I agree. Your issue isn't the issue I had. Debugging these early issues is hard, because it has to be done from the emergency console. But, I think I remember there being other consoles to switch to that had things like logs, and there are some commands available, if you do an ls /usr/bin or ls /usr/sbin you should be able to examine the system to some extent to find why it is failing. I remember using less and vi at least. Could you mount one of the partitions and save / cp or cat the logs to a file there, maybe in your home directory. Then you can examine them from a working system, and even attach them to a bugzilla or email.
# ls -n /boot/initramfs* -rw-------. 1 0 0 83396891 Jan 21 2020 /boot/initramfs-0-rescue-5cbe81aa795444b29a47ec1bf2b6dca1.img -rw-------. 1 0 0 37876860 Aug 14 11:45 /boot/initramfs-6.4.10-200.fc38.x86_64.img -rw-------. 1 0 0 37878341 Aug 26 12:28 /boot/initramfs-6.4.12-200.fc38.x86_64.img
Thanks, Stan. Finally, I was able to capture the log of
journalctl -xb
which is at:
https://bugzilla-attachments.redhat.com/attachment.cgi?id=1985687
So, this seems to be the error: Aug 28 19:05:39 localhost @ystemctl[626]: Failed to switch root: Specified switch root path '/sysroot' does not seem to be an OS tree. os-release file is missing.
And it looks like it was triggered by this: Aug 28 19:05:39 localhost systemd[1]: Received SIGRTMIN+20 from PID 569 (plymouthd).
After that message, systemd starts shutting down instead of starting up.
Do you have fedora-release for fc38 installed? I can't see how you wouldn't, but just confirming. It looks like there is a bug in the code that is determining whether os-release is present, but only for warm restarts, not for cold boots. That would be plymouthd. Or whatever is creating the warm restart isn't creating it properly, so that plymouthd can recognize it, I think that would be systemd.
Were either of those in the large batch of updates you installed? If so, you could try downgrading them from koji to see if it fixes the error. https://koji.fedoraproject.org/koji/
Thanks, Stan.
I do have:
# rpm -q fedora-release-38 fedora-release-38-36.noarch #
Moreover, both plytmouth and systemd that I have installed correspond to versions dated earlier than the date of the first occurrence of the reported issue.
With the older
kernel-6.4.10-200.fc38.x86_64
there is no issue.
Paul