Bug#760848: Similar Issue under different circumstances

[Rod Rodolico]

I am including this in but 760848 since I believe it is the same root cause.

I have set up a new install of Jessie, then modified fstab to place
several things into tmpfs

tmpfs /var/tmp tmpfs rw,size=64M 0 0

I made a mistake (did not include the third parameter). Instead of
booting in with a warning, systemd went into rescue mode and refused to
complete the boot. When I pressed Control-D to continue, it looped back
to the same message and I was forced to log into the rescue screen and
fix the non-critical error before my server would boot.

I then added an old hard drive to bring up, and the old hard drive had
an LVM2 signature on it. systemd again found the error, went to the
rescue prompt, and would refuse to boot into the system even after
Control-D was pressed. Again, I had to log into the rescue mode and
repair the error before the system was usable again.

Both errors were non-critical. Under wheezy, they would have (and did)
continue with warnings, then were repairable while the server was up and
running.

systemd should not block boot on non-critical file issues. That may be
fine on a workstation, but on a server which may be miles away, it is
not acceptable. The server should boot, even in a degraded state, with
warnings so it can begin serving while repairs are undertaken.

I assume this would also occur if an NFS mount or an iSCSI mount was not
available on reboot, neither of which should cause system failure.

This action decreases the robustness and maintainability of remote servers.

-- 
Rod Rodolico
Daily Data, Inc.
POB 140465
Dallas TX 75214-0465
214.827.2170
http://www.dailydata.net