Eric Sandeen wrote:
> Yep, I've got to look into this, the XFS installer has been plagued with
> mkinitrd problems for a while...
This doesn't appear to be an installer problem but rather a kernel problem.
For some reason the loop module either isn't loading or is refusing to
and more devices.
Since the install images is mounted via loopfs the modules does appear to
loading but mkinitrd is failing.
I need to find a box that actually fails to make an initrd and try and
The installer could be modified to call mkinitrd with the -x option and
dump the output to syslog or a console... that might give us an idea
of where the failure occurs.
> Although - I'm not surprised that there is no initrd in your first case,
> since IDE, EXT2, and XFS are all built into the kernel, and no initrd is
> But unless I'm missing something (only one cup of coffee so far today),
> I don't see why the second case built an initrd - can you take a look at
> what's in the initrd image?
> So in this case, you got an extra initrd you didn't need - as opposed to
> the unlucky folks who _didn't_ get one that they _did_ need. :)
> Bryan-TheBS-Smith wrote:
> > Test2: Sometimes she initrd's, sometimes she doesn't ...
> > I made two "test installs" on the same system. One was with
> > partitions:
> > hda2 /boot Ext2
> > hda3 swap
> > hda7 / XFS
> > hda8 /tmp XFS
> > hda9 /var XFS
> > hda10 /usr XFS
> > And one with partitions:
> > hda2 / Ext2
> > hda3 swap
> > hda7 /tmp XFS
> > hda8 /var XFS
> > hda9 /usr XFS
> > In the first case, with the separate /boot, it did NOT install and
> > enable an "initrd.img" file. *BUT*, in the second case, it *DID*!
> > Interesting ...
Digital Elves inc. -- Currently on loan to SGI
Linux XFS core developer.