Bug#785525: systemd fails to boot if non-essential block device is missing

Thibaut Varène varenet at debian.org
Sun May 17 13:59:34 BST 2015


On 17 mai 2015, at 14:57, Martin Pitt <mpitt at debian.org> wrote:

> Control: tag -1 wontfix
> 
> Hello Thibaut,
> 
> Thibaut VARENE [2015-05-17 14:41 +0200]:
>> Just upgraded my system from wheezy to jessie. Boot stopped after initial reboot, with prompt to enter root password. After looking at the logs, it seems that systemd choked on this line of my fstab:
>> 
>> UUID="8633-12F1"	/media/WDX360	vfat	user,auto,utf8=no,iocharset=iso8859-1	0	2
>> 
>> That's a line I added for convenience whenever I plug a specific USB
>> drive to the machine. Granted, it doesn't have the nofail flag
> 
> Indeed, that *and* it's marked as "auto".
> 
>> and it does have a 2 pass_no (instead of 0), but this never caused
>> any problem in wheezy, so the new behavior, and the fact that it
>> entirely halts the boot process (the machine isn't even remotely
>> accessible) is quite a huge change from previous behavior...
> 
> It's a change indeed, but IMHO a correct one. Aside from "auto" and
> "nofail" there is no other indication whether a file system in fstab
> should be considered "essential" (in your terms) or not. That's
> precisely what these flags are for. sysvinit might have not complained
> about this situation, but that doesn't mean that I'd like to
> proliferate that bug forever.
> 
> Hence I consider this a "wontfix".

A footnote in the release/upgrade instructions would have been nice. If that machine had been remote without OOB access, I'd have been screwed over, because something that never broke previous upgrades.

T.




More information about the Pkg-systemd-maintainers mailing list