Arch Linux

Are you running a custom kernel?

What's the filesystem and mode of /tmp?

Right now cups doesn't work even with the work around of the wiki and makes systemd eat a lot of cpu. Systemd and his journal with cups enabled use 70% of cpu on a 3GHz quadcore amd cpu!

That doesn't answer the question about your kernel and filesystem. Do you use anything custom and uncommon? You seem to be the only one with this issue.

I'm not the one who submitted the bug. Anyway I use the default archlinux kernel and I every core component is from the archlinux repos.

There was also another task open by another user which was closed as a duplicate of this one, he was reporting what I said in the previous post,a high cpu usage of systemd when cups is enabled.

I do have the same issue. Linux kernel is latest from stable core repo. /tmp is tmpfs with mode 1777. Will take a look at the mount options tomorrow.

/tmp is mounted with "defaults,size=6G", in reference to mount resulting in "rw,relatime,size=6291456k". So nothing unusual here I think.

Looks like this is a problem in systemd. All services using PrivateTmp fail:

# systemctl status ntpd.service
ntpd.service - Network Time Service
Loaded: loaded (/usr/lib/systemd/system/ntpd.service; enabled)
Active: failed (Result: exit-code) since Tue, 04 Sep 2012 08:15:46 +0200; 25min ago
Process: 1129 ExecStart=/usr/bin/ntpd -g -u ntp:ntp (code=exited, status=226/NAMESPACE)
CGroup: name=systemd:/system/ntpd.service

linux 3.5.3-1 (It's standard Arch Linux kernel)

tmpfs on /tmp type tmpfs (rw,nosuid,nodev,relatime)

drwxrwxrwt 10 root root 280 Sep 4 08:40 /tmp/
drwxrwxrwt 2 root root 4096 Sep 4 08:40 /var/tmp/

Regards

I disagree about ntpd:
# systemctl status ntpd.service
ntpd.service - Network Time Service
Loaded: loaded (/usr/lib/systemd/system/ntpd.service; enabled)
Active: active (running) since Tue, 04 Sep 2012 08:44:42 +0200; 3s ago
Process: 725 ExecStart=/usr/bin/ntpd -g -u ntp:ntp (code=exited, status=0/SUCCESS)
Main PID: 727 (ntpd)
CGroup: name=systemd:/system/ntpd.service
└ 727 /usr/bin/ntpd -g -u ntp:ntp
(...)

I found my problem: /var/tmp was a symlink to /tmp. systemd failed to bind mount the private namespace to /var/tmp/. After fixing this cupsd and ntpd are up and running for me.

Artur, how about you, is /var/tmp a symlink on your system? However I do not have any idea why cupsd and ntpd service files should have different results.

You are right. Now as You asked, I remember, that before (when I submit bug) my /var/tmp was a symlink to /tmp.. now I have it as separate dir, so cups is working again.
So problem is with symlink on /var/tmp.

# ls -dl /var/tmp/
drwxrwxrwt 5 root root 4096 Sep 4 09:23 /var/tmp/
# systemctl status cups.service
cups.service - CUPS Printing Service
Loaded: loaded (/usr/lib/systemd/system/cups.service; enabled)
Active: active (running) since Tue, 04 Sep 2012 09:23:27 +0200; 1min 56s ago
Main PID: 1045 (cupsd)
CGroup: name=systemd:/system/cups.service
└ 1045 /usr/sbin/cupsd -f

Sep 04 09:23:27 msi cupsd[1045]: Unknown directive DefaultAuthType on line 25.

and

# rm -rf /var/tmp && ln -s /tmp /var/tmp
# ls -ld /var/tmp
lrwxrwxrwx 1 root root 4 Sep 4 09:26 /var/tmp -> /tmp
# systemctl start cups.service
# systemctl status cups.service
cups.service - CUPS Printing Service
Loaded: loaded (/usr/lib/systemd/system/cups.service; enabled)
Active: failed (Result: exit-code) since Tue, 04 Sep 2012 09:26:50 +0200; 2s ago
Process: 1080 ExecStart=/usr/sbin/cupsd -f (code=exited, status=226/NAMESPACE)
CGroup: name=systemd:/system/cups.service

I suppose this breaks since "namespace: rework namespace support" (commit ac0930c892bc7979b4c9bc2a52e5e844650b025d).

This seems like a dupe of  FS#30229 . I see no reason to support /var/tmp as a symlink.

For me that is not the case


# ls -dl /var/tmp/
drwxrwxrwt 3 root root 143360 4 set 20.51 /var/tmp/


# systemctl start cups.service
# systemctl status cups.service
cups.service - CUPS Printing Service
Loaded: loaded (/usr/lib/systemd/system/cups.service; disabled)
Active: failed (Result: signal) since Tue, 04 Sep 2012 20:54:15 +0200; 8s ago
Process: 13373 ExecStart=/usr/sbin/cupsd -f (code=killed, signal=ABRT)
CGroup: name=systemd:/system/cups.service

zsh: segmentation fault systemctl status cups.service


I think is a slightly different bug as systemd use a lot of cpu.

Looks like reinstalling for the fourth time cups solve the problem!

Status?