LinuxQuestions.org - [SOLVED] systemd startup and init, rc.d interactions vs sysvinit (migrated from sysvinit to system)

- Linux From Scratch (https://www.linuxquestions.org/questions/linux-from-scratch-13/)

- - systemd startup and init, rc.d interactions vs sysvinit (migrated from sysvinit to system) (https://www.linuxquestions.org/questions/linux-from-scratch-13/systemd-startup-and-init-rc-d-interactions-vs-sysvinit-migrated-from-sysvinit-to-system-4175633458/)

systemd startup and init, rc.d interactions vs sysvinit (migrated from sysvinit to system)

Hi everyone,

The backstory is when originally building my first LFS I thought I was following along with the 8.2 systemd version but after was all said and done I realized I had been using the 8.2 sysvinit version (derrr)

What I ended up doing instead of starting over is first moved:

init.d > init.d.bak
inittab > inittab.bak
rc.d > rc.d.bak

Then I installed systemd and dbus, reconfigured the system according to the systemd instructions and recompiled the kernel.

Everything appears to be working except a couple things.

During boot I receive a few error messages:

Code:

[152]: failed to execute '/etc/rc.d/init.d/setclock' '/etc/rc.d/init.d/setclock start': No such file or directory



systemd-udevd[135]: Process '/etc/rc.d/init.d/setclock start' failed with exit code 2.

systemd[1]: System is tainted: local-hwclock

I'm assuming some sysvinit processes trying to run during boot but I don't know why or how. I feel like this might be an easy fix I'm just a noob though so I'm a little befuddled. I don't want to just reintroduce the rc.d directory for fear of creating more boot conflicts between the new systemd and old sysvinit.

I'm also getting an error when running:

Code:

timedatedctl set-local-rtc 1



Failed to create bus connection: no such file or directory.

Are both errors related?

In addition journalctl tells me:

Code:

systemd-udevd[133]: Specified group 'kvm' unknown

What is the proper way to set up the kvm group? (i've enabled kvm in my kernel).

Thanks for the suggestions.

Spencer

I tracked down the bus connection problem after a long while. The problem was dbus was being compiled without systemd support. I found this out after I realized I was missing the dbus.socket unit file. When dbus compiles it automatically enables systemd support so I tradidn't think anything of it. What was happening was my PKG_CONFIG_PATH environment variable was not set properly so dbus though that systemd wasn't there. I've had this happen on other compiles during LFS but the compile usually fails so I am able to fix it right away. dbus does not report any errors though. It was only when I explicitly "--enable-systemd" didnt it throw the error after configure that said I didn't have systemd installed.

Just need to trackdown the hwclock error and I think I have a shiny new system to drive off the lot.

hwclock error comes from a udev rule installed by lfs-bootscripts. Do a grep -r "hwclock" /lib/udev/rules.d, and remove the file

awesome. thanks for the info!