Re: HOWTO example for ReadOnly Rootfs using multistrap
Am 18.01.2011 23:45, schrieb Hector Oron:
Surely something to look at it. Thanks.
Hi Hector, many thanks for the feedback.
I have found some bugs which could be filed against the BTS
(excerpt of TODO:
- fsck.ubifs dummy package similar to fsck.nfs (fsck.ubifs is not
implemented)
From UBI documentation:
UBIFS is tolerant of unclean reboots and power-cuts. It does not need
stuff like fsck.ext2.
Yes. I know. That is what I found, too. But anyway the debian standard
setup tries to check every filesystem (but nfs). Since there is no
fsck.ubifs this produces alienate messages on startup. Debian ignores
fsck for NFS - so it should ignore ubifs, too. The fsck.nfs thing was
solved with a dummy package. To keep the pattern, a similar dummy
package for fsck.ubifs should be created.
This bug ticket patches some skript:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=571241
Anyway, I prefere to keep the pattern and create a dummy package similar
to fsck.nfs.
- noauth must be true even if Emdebian archive keyring installed on host
system
I thought that was already fixed.
I did not try the newest version from sid. My host system is a debian
squeeze amd64 without "sid-backports".
- mksquashfs 4.0 from Squeeze did not work for me on my arm target, 4.1 from
Sid does)
In which way failed for you?
The mksquashfs (4.0) command itself does not show any error. But when I
try to mount the generated image with mount -o loop,ro my.sqfs
/mnt/sqtest it breaks with some weird trace messages viewable via dmesg.
I tried the same mksquashfs command (4.1) and it could be mounted via
mount -o loop,ro without problems.
(I tried to put the sq-image on several media - usbstick, 2,5"usbdrive
and internal nand flash - that was NOT the reason - the 4.0 image was
NOT mountable from anywhere [armelbox#mksquashfs /var /.var.sqfs.new
-noI -noD -noF])
Any comments on RO solution with aufs-overlay?
Thanks again for the feedback.
Regards,
Marcus
Reply to: