[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#923400: initramfs-tools: failure inside chroot: W: Couldn't identify type of root file system for fsck hook



Quoting Ben Hutchings (2019-02-27 19:37:41)
> Control: severity -1 wishlist
> Control: retitle -1 Add option to override filesystem type detection
> 
> On Wed, 2019-02-27 at 18:38 +0100, Jonas Smedegaard wrote:
> > Quoting Jonas Smedegaard (2019-02-27 17:45:33)
> > > Building a system image using multistrap,
> > > including generating an initial ramdisk,
> > > worked fine recently.
> > >
> > > Now it fails with this error message:
> > > 
> > >   W: Couldn't identify type of root file system for fsck hook
> > > 
> > > It seems to me that git commit a8ed874 intended to extend the code 
> > > operating on "auto" mounted filesystems to cover all _except_ root 
> > > disk, but that the logic is flipped around so that now it _only_ 
> > > extends that to include root disk:
> 
> This commit went into version 0.123, before stretch, so if your use
> case "worked fine recently" then this is not the change that broke it.

Heh.  I even looked briefly at the date thinking "hmm, committed in 
January, not February when released was made, but oh well...", and 
noticed that corresponding bug#767471 had a quite low number... :-)


> > Please ignore my guess above: I think I understand now that it was 
> > intentional to check root disk (I got confused by the comment 
> > talking about ignoring root and then processing root not skipping 
> > it).
> > 
> > Let me clarify my use case: I generate a system image on a fast 
> > amd64 system targeted a slower real device (that's the reason having 
> > initramfs generated is important).
> > 
> > fstab now unconditionally being distrusted for root disk makes it 
> > more difficult to build on a different host than intended for target 
> > boot.
> >
> > Would it perhaps make sense to support passing pre-resolved root 
> > filesystem fstype as an environment variable, taking precedence over 
> > probing?
> 
> I don't think this should be an environment variable but it does seem 
> like a useful option.

Thanks for considering.


 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private

Attachment: signature.asc
Description: signature


Reply to: