On 2023-01-29 07:24, Zbigniew Jędrzejewski-Szmek wrote:
On Sun, Jan 29, 2023 at 01:50:21AM -0800, Luya Tshimbalanga wrote:
Hello team,
When building a package i.e. Blender as an exampleon Rawhide repository, the following message displayed:
DEBUG util.py:445: /proc/ is not mounted, but required for successful operation of systemd-tmpfiles. Please mount /proc/. Alternatively, consider using the --root= or --image= switches. DEBUG util.py:445: ⚠️ /proc/ is not mounted. This is not a supported mode of operation. Please fix DEBUG util.py:445: your invocation environment to mount /proc/ and /sys/ properly. Proceeding anyway. DEBUG util.py:445: Your mileage may vary. DEBUG util.py:445: ⚠️ /proc/ is not mounted. This is not a supported mode of operation. Please fix DEBUG util.py:445: your invocation environment to mount /proc/ and /sys/ properly. Proceeding anyway. DEBUG util.py:445: Your mileage may vary. DEBUG util.py:445: /usr/lib/sysusers.d/basic.conf:9: Conflict with earlier configuration for user 'root', ignoring line.
It seems like a bug related to the new systemd 253rc which can lead to failure. Here is the link of root.log:
https://kojipkgs.fedoraproject.org//work/tasks/79/96810079/root.log
And the affected build: https://koji.fedoraproject.org/koji/taskinfo?taskID=96810014
The issue as in the message. It seems that some commands are called in a partial chroot where /proc is not mounted. procfs is an "api filesystem" that is effectively part of the linux userspace api and is not optional. For example, /proc/self/fd/ for a program to get information about its own file descriptors. Thus, this is a problem with how the caller has set up the chroot.
Thanks for the info. That setup impact some subcommand on both make and ninja accordingly, hence the last line:
ninja: build stopped: subcommand failed.
See build.log for details: https://kojipkgs.fedoraproject.org//work/tasks/79/96810079/build.log