On 11/03/14 19:17, Francis SOUYRI wrote: > Hello, > > I am running a Fedora 3.16.6-203.fc20.x86_64, after some updates > (do not know what updates) now systemd can not start rtkit-daemon, ntpd, > cups when PrivateTmp=yes. > > I see a same problem when /var/tmp is a symlink of /tmp or when /tmp is > a tmpfs but this is not my case. > > # df -k /tmp /var/tmp > Filesystem 1K-blocks Used Available Use% Mounted on > /dev/mapper/vgroot-lvtmp 4022592 434892 3363652 12% /tmp > /dev/mapper/vgroot-lvvar 4022592 1862340 1936204 50% /var > > # ls -ld /var/tmp /tmp > drwxrwxrwt. 89 root root 53248 Nov 3 12:01 /tmp > drwxrwxrwt. 16 root root 12288 Nov 3 12:00 /var/tmp > > l have some other systems the a 32 bits FC20 without this problem. > > I set PrivateTmp=no systemd work fine, but this is only a bypass. > > Since this seems like a standard F20 system you're running you may want to bring this up on the "users" list? But, I will say I'm not seeing this on a fully updated F20 system. I checked this both with /tmp being on "real" disk as well as tmpfs. [egreshko@f20f ~]$ grep PrivateTmp /lib/systemd/system/cups.service PrivateTmp=true [egreshko@f20f ~]$ ps -eaf | grep cupsd root 603 1 0 20:48 ? 00:00:00 /usr/sbin/cupsd -f What means "can not start"? Error messages in log files? Error messages displayed with "systemctl status"?

On 11/03/14 21:05, Francis SOUYRI wrote: > Sorry for the mailing list error, I have NAMESPACE error messages: > > rtkit-daemon.service - RealtimeKit Scheduling Policy Service > Loaded: loaded (/usr/lib/systemd/system/rtkit-daemon.service; disabled) > Active: failed (Result: exit-code) since Mon 2014-11-03 11:42:53 CET; 16s ago > Process: 5585 ExecStart=/usr/libexec/rtkit-daemon (code=exited, status=226/NAMESPACE) > Main PID: 5585 (code=exited, status=226/NAMESPACE) > CGroup: /system.slice/rtkit-daemon.service > > > > ntpd.service - Network Time Service > Loaded: loaded (/usr/lib/systemd/system/ntpd.service; enabled) > Active: failed (Result: exit-code) since Mon 2014-11-03 11:45:44 CET; 36min ago > > Nov 03 11:45:44 francis.apec.fr systemd[5799]: Failed at step NAMESPACE spawning /usr/sbin/ntpd: Operation not permitted > Nov 03 11:45:44 francis.apec.fr systemd[1]: ntpd.service: control process exited, code=exited status=226 > Nov 03 11:45:44 francis.apec.fr systemd[1]: Failed to start Network Time Service. > Nov 03 11:45:44 francis.apec.fr systemd[1]: Unit ntpd.service entered failed state. I don't think I'm going to be much help. Other than to point out that this is a known condition when /var/tmp is a symlink of /tmp. Currently running my test VM .... [egreshko@f20f ~]$ ls -ld /var/tmp /tmp drwxrwxrwt. 22 root root 4096 Nov 3 21:01 /tmp drwxrwxrwt. 18 root root 4096 Nov 3 20:48 /var/tmp [egreshko@f20f ~]$ df -k /tmp /var/tmp Filesystem 1K-blocks Used Available Use% Mounted on /dev/mapper/fedora_f20f-root 29775932 13066196 15174148 47% / /dev/mapper/fedora_f20f-root 29775932 13066196 15174148 47% / but was previously running /tmp on tmpfs without error. FWIW, systemd was just updated on my systems today. You may want to try downgrading to see if that is what caused the change to your system.

On 11/03/14 23:55, Francis SOUYRI wrote: > With and without the lastest systemd update I have the problem. I would compare how your file system is laid out with your other working systems. Since this isn't a common problem, it hasn't been mentioned on the user's list, I would take the question to the wider audience on the "users" list as this list is more devoted to testing of the upcoming release of F21.