The installer aborts with a popup indicating that 'script' failed for both systemd-255.4-1.mga10.x86_64 and lib64cloudproviders0-0.3.5-1.mga10.x86_64 (twice). However, at least in the case of systemd, urpmi on a running cauldron system gets the error but ignores it and continues. Redoing the install of systemd on the running system gives: [root@ftglap ~]# urpmi --replacepkgs systemd installing systemd-255.4-1.mga10.x86_64.rpm from /mnt/cauldron/x86_64/media/core/release Preparing... ############################################# 1/1: systemd ############################################# 1/1: removing systemd-255.4-1.mga10.x86_64 ############################################# /var/tmp/rpm-tmp.sWX7CE: line 24: /etc/hostname: Operation not permitted warning: %posttrans(systemd-255.4-1.mga10.x86_64) scriptlet failed, exit status 1 ERROR: 'script' failed for systemd-255.4-1.mga10.x86_64 You should restart your computer for systemd The "operation not permitted" on /etc/hostname is because that file is set to immutable to prevent systemd from messing with it, but that obviously is not the case for an install. The install is a network install with all repositories enabled and all package categories selected. VT3 shows no errors (or else they've scrolled off). [root@ftglap ~]#
Thank you for the report. Can you be more precise about the libcloudproviders failure? Since this looks more like a systemd install problem than the Installer itself, assigning to BaseSystem maintainers. CC'ing DavidG for the libcloudproviders issue.
CC: (none) => geiger.david68210Assignee: bugsquad => basesystemSource RPM: (none) => systemd, libcloudproviders
There isn't any post/preun/posttrans script in libcloudproviders.
I tried a net install in a vb guest using a freshly downloaded Mageia-Cauldron-netinstall-nonfree-x86_64.iso I also encountered an error getting systemd installed. Snippet from /root/drakx/install.log ... util-linux-2.39.3-1.mga10.x86_64 /var/tmp/rpm-tmp.mxpzOX: 1: /usr/lib/rpm/sysusers.sh: not found %sysusers(systemd-255.4-1.mga10.x86_64) scriptlet failed, exit status 127 systemd-255.4-1.mga10.x86_64 systemd-255.4-1.mga10.x86_64: install failed I'm thinking /usr/lib/rpm/sysusers.sh needs to be added or the package containing it required prior to util-linux/systemd being installed. This stops the installation. Attempts to restart it failed.
CC: (none) => davidwhodginsSeverity: major => critical
The thing that caught my attention was that the error stopped the installation cold, but didn't stop urpmi from a running system command line. Lewis, if we get past the systemd issue I'll retest and see if the other package error goes away.
File /usr/lib/rpm/sysusers.sh is provided by rpm, does rpm not be installed by default even using a netinstall?
Created attachment 14449 [details] Compressed ddebug.log showing /usr/lib/rpm/sysusers.sh: not found Attaching the full ddebug.log (compressed using xz). I think the problem is that it's trying to install systemd before installing rpm. Note that there are other messages that may or may not be related. $ grep 'no packages match group' ddebug.log|sort -u * no packages match group(colord) (it is either in skip.list or already rejected) * no packages match group(lightdm) (it is either in skip.list or already rejected) * no packages match group(man) (it is either in skip.list or already rejected) * no packages match group(rpm) (it is either in skip.list or already rejected) * no packages match group(wbpriv) (it is either in skip.list or already rejected)
Looking back at an m8 install [root@i9v drakx]# grep -e rpm-4 -e systemd-2 ddebug.log |grep trans|head -n 2 * trans: scheduling update of systemd-252.4-1.mga9.i586 (id=2619, file=/tmp/image/media/core/systemd-252.4-1.mga9.i586.rpm) * trans: scheduling update of rpm-4.18.0-4.mga9.i586 (id=1096, file=/tmp/image/media/core/rpm-4.18.0-4.mga9.i586.rpm) So the problem in the m9 install is that systemd now requires rpm to already be installed before it is installed, but systemd does not have a requires on rpm, so the order is not guaranteed.
Adding Stig-Ørjan to the cc list as the last person who made changes to the spec for systemd.
CC: (none) => smelror
Dare one hope that this is a relatively simple fix ?