2.4.2 ext2 filesystem corruption ? (was 2.4.2: What happened ? (No such file or directory))

[Frédéric L. W. Meunier <0@pervalidus.net>]

Hi. After a reboot I had to manually run fsck (sulogin from
sysinit script) since there were failures.

In my second (and problematic) boot with 2.4.2 I used the
option mount --bind in my sysinit script to mount the old /dev
in /dev-old before devfs was mounted, so I could get rid of all
entries that were still there (I removed most before building a
Kernel with devfs support).

For some reason I couldn't remove /dev-old/hdd2. It reported
can't state file. Note that I never used /dev/hdd*, since I
only use hda and hdc, but am sure it was OK with 2.4.0 (mc
reported an error when I accessed /dev-old, what never happened
before), the last time I used a Kernel without devfs support.

If you read my old thread, you should notice various
applications couldn't access (or rename ?) files. It happened
after ~8h of idle time. It was OK at 5:58, when I last ran cvs
and killed pppd, but failed at ~14:30, when multilog (from
daemontools) had to do something to a full dnscache log file (I
was online).

I'm not sure 2.4.2 is the culprit. I just hope it's the last
time. There were no errors when I first booted with this Kernel
(I was using 2.4.1), and my first uptime was ~6 days (~23 with
2.4.1). Also there were no errors when I booted 2.4.2 for the
second time.

BTW, /lost+found contains hdd2:

brw-r-----    1 root     disk      22,  66 May  8  1995 #518878

The other partitions (/home/ftp/pub and /usr/local/src) have no
problems.

-- 
0@pervalidus.{net, {dyndns.}org} Tel: 55-21-717-2399 (Niterói-RJ BR)