I made a diskless image against Fedora15, during the boot I found it displayed
the following error message and went into emergency mode.

================================
The error message:
Starting Relabel all filesystems, if necessary ^[[1;31maborted^[[0m because a
dependency failed.
[ 107.607155] systemd[1]: Job fedora-autorelabel.service/start failed with
result 'dependency'.
Starting Mark the need to relabel after reboot ^[[1;31maborted^[[0m because a
dependency failed.
[ 107.625156] systemd[1]: Job fedora-autorelabel-mark.service/start failed with
result 'dependency'.
[ 107.634580] systemd[1]: Job local-fs.target/start failed with result
'dependency'.
[ 107.642615] systemd[1]: Triggering OnFailure= dependencies of
local-fs.target.
[ 107.650738] systemd[1]: Job var-tmp.mount/start failed with result
'dependency'.
[ 107.658580] systemd[1]: Job fsck/start failed with result
'dependency'.
[ 107.666876] systemd[1]: Job tmpfs.device/start failed with result 'timeout'.
Welcome to emergency mode. Use "systemctl default" or ^D to activate default
mode.
================================

Then I could log on the node. But when I run the command 'service sshd start'
to start the sshd, it displayed the semilar error message:

================================
[ 8545.884404] udev[2872]: starting version 167
[ 8635.560123] systemd[1]: Job tmpfs.device/start timed out.
[ 8635.565867] systemd[1]: Job fedora-autorelabel.service/start failed with
result 'dependency'.
[ 8635.574812] systemd[1]: Job fedora-autorelabel-mark.service/start failed
with result 'dependency'.
[ 8635.584217] systemd[1]: Job local-fs.target/start failed with result
'dependency'.
[ 8635.592203] systemd[1]: Triggering OnFailure= dependencies of
local-fs.target.
[ 8635.600311] systemd[1]: Job var-tmp.mount/start failed with result
'dependency'.
[ 8635.608102] systemd[1]: Job fsck/start failed with result
'dependency'.
[ 8635.616355] systemd[1]: Job tmpfs.device/start failed with result 'timeout'.
Welcome to emergency mode. Use "systemctl default" or ^D to activate default
mode.
================================

Then I tried to start the tmpfs.device

================================
# systemctl start tmpfs.device
[432039.543362] systemd[1]: Job tmpfs.device/start timed out.
[432039.549335] systemdJob timed out.
================================

But from the output of 'systemctl start tmpfs.device', it has been loaded:

================================
# systemctl status tmpfs.device
tmpfs.device
          Loaded: loaded
          Active: inactive (dead)
================================

I also tried to start the tmpfs.device on the Fedora 15 diskfull node, it also
said timed out:

================================
# systemctl start tmpfs.device
Job timed out.
================================

But the tmpfs looked work well. 

On the diskless node I found a way to workaround the starting of sshd.
I commented out the '#. /etc/rc.d/init.d/functions' from the sshd control file:
/etc/init.d/sshd. Then sshd was started successfully.