Currently the file system check during startup (/etc/init.d/checkfs
+checkroot) _always_ runs reiserfsck against the root (/) file
system.
The check finds an error in the (free/used space) bitmap, which
should be fixable. But the program says:
--fix-fixable ignored
so nothing gets fixed. Nonetheless checkfs/checkroot says
"* File system repaired":-<<
I had tried reiserfsck earlier from the command line and didn't
succed either (see below, items 3ff).
***
How it all began...
1) While I ran an
emerge -uD system | tee logfile
after an rsync, the speaker started to beep continuously. I panicked
(CPU overheated?) and cut the power.
2) On the next boot, reiserfsck considered the filesystem as clean!
But the last 30 lines of "logfile" were wrong data!
3) Now I wanted to run the file system check manually. I forced the
system to single-user mode by "telinit 1" (maybe not the correct
way), re-mounted / read-only and ran
reiserfsck --check /dev/hda8
This reported 2 fixable errors in the "bitmap".
Note: --check would refuse to run on the writeable reiserfs.
4) Next command:
reiserfsck --fix-fixable /dev/hda8
--fix-fixable rejected for a read-only reiserfs (of course;-)
Re-mounted /dev/hda8 r/w, re-issued above command...
"Cannot run check on a file system with write permission".
Termination!
FYI: reiserfsck is quite current (3.6.8), reiserfs is v3.6.25
(vanilla-kernel 2.4.19), installed 10 months ago.
Note: I'm writing this under Windows; all data come from memory;-)
5) But reiserfsck must have turned some kind of "unclean" bit on (in
the superblock?), triggering a complete check at each startup.
==> What must I do to make reiserfsck _repair_ the file system?
(Just occured to me: since this is the root file system, could it be
necessary to free (unmount) all mount points?)
Best regards,
-Heribert
--
Heribert Slama
Muttenz, Switzerland
--
[EMAIL PROTECTED] mailing list
