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

reiser on /, fsck and ro: bug?


I have set up a box that uses reiserfs as the root filesystem

I have noticed 3 facts that seem to be bugs (but I could not tell for sure)

I use a stock kernel by Herbert XU, which uses initrd; when initrd's /sbin/init is run, eventually it mounts the root from the hard disk, and it always mounts it read-only, without checking if the kernel option 'ro' was given: is this a bug? what package's bug?

Then /sbin/init is executed from the hard disk, and this calls all /etc/rcS.d/* that do an fsck on /

When this happens, though, fsck.reiserfs says: "filesystem is mounted read-only: skipping journal replay": so it seems that the journal will never be replyed, even if the filesystem is dirty (I am not sure that this is the case, but I am not willing of doing extensive testing on this issue). This is the opposite of what fsck.ext3 would do: AFAIK it does a journal replay and a fsck ONLY IF the root is mounted read-only.

Then, fsck.reiserfs does a fsck of the disk. Regardless of it being dirty or not (that is, ignoring the absence of the '-f' flag). This is another difference with fsck.ext{2,3}. This is a minor bug, but annoying (it wastes time).


problem 2 is the most worrisome.
What is the right boot configuration to avoid it? should I mount / as read-write?

But (if this is the case) then this cannot be used as a general solution : ext2 and ext3 cannot be fsck-ed if rw

the problem is also worrisome since debian-installer will be able to install Debian with a reiserfs partition


Reply to: