/dev/hda7 gone ? Long (and sad) message

[Guest scrat]

Hi everybody, I would greatly appreciate some advice on how to deal with a major filesystem problem. I don't know if this is the right forum though; if it isn't please tell me.

 

This is a long message, I know. Sorry. The story is long (and painful).

 

Well, it happened this way. Two days ago, I restarted my laptop running 9.0, using

shutdown -h now

(I had to restart it because of a bug in XFree86 which sometimes displays the cursor some 20 pixels off its logical position; it has been this way since XFree3.x , and it still is with the latest available.Restarting the Xserver won't do, I have to reboot.This is old stuff and never caused major problems.)

 

But, when I rebooted, I received the following message (more or less)

 

"/dev/hda7 Unexpected Inconsistency; run fsck manually (i.e., without -a or -p options)

failed to check filesystem.Do you want to repair the errors? (Y/N)"

 

to which I answered Y.Then an apparently endless process began, where first

it said that every inode had become disconnected and it would reattach it

to /lost+found; then that the inode count was wrong and it would correct it.

Anyhow, when it was finished, the compuer would not respond, so I restarted

it with CTRL+ALT+DEL. And there I go anytime I start it, the same procedure

as above over and over again.

 

If I answer N, it "drops me to a shell" to perform maintenance... which I

would suppose should be fsck /dev/hda7, but exactly the same happens, only I

am asked:

 

Pass 1: Checking inodes, blocks and sizes

Pass 2: Checking directory structure

Pass3: Checking directory connectivity

'..' in /(2) is <The NULL inode> (0), should be (2)

Fix <y> it? yes

Unconnected directory inode 12 (.../#12)

Connect to /lost+found<y>?

...

and so on and so forth. When it's over (after a very long while spent typing

``y'') I am able to mount read only /dev/hda7; it seems to me that

everything is listed in my home, and I have been able to copy the most

critical work-in-progress to the /dev/hda1 FAT partition. Nevertheless,

if I shutdown -h now, and then reboot, the filesystem seems _not_ to be

repaired, and the process starts over again. Also, the lost+found directory

in /dev/hda7 is

filled with number-named files like #12345 , but I can't give ``file #12345'' to try and find out what a file is, it just complains there's no such file.

 

/dev/hda7 is ...err, was :-( my home part. I did have some valuable data in

it, only a few files though (latex writing), and those I have been able to

rescue as I said. Of course I would be glad to recover the rest too

(some music, books, not work-critical stuff), but

if that's not possible or exceedingly hard, I can afford formatting

/dev/hda7...

What would you wise guys suggest ?

 

Does all this mean there is some bug in the 9.0 stock kernel handling of ext2 ?

 

Any hints ?

 

Thanks ! Scrat

[theYinYeti]

I'm sorry for you... Unfortunately, I don't know what you should do.

 

Anyway, sooner or later (if possible after you have recovered the data you've lost), there will be nothing left to do with /dev/hda7.

Then I suggest you reformat the partition, and use a journalized filesystem (JFS, XFS or ReiserFS). That way, those FSCK will be only old (and painful) memory :)

 

Good luck.

 

Yves.

[Cannonfodder]

Also, when rebooting because you have too, are the partitions being unmounted first? If not, then your file systems get damaged. Switching to reiserfs as suggested is a good way to go, but I would also fix that need to reboot..

[bvc]

Yuk!

 

You could try parted and see if it finds/fixes anything.

http://www.mandrakeusers.org/viewtopic.php...ighlight=parted

[Guest scrat]

Thanks...

 

I'll go and see what parted can do for me.

 

Regarding the journalized filesystem, it already was; it was ext3. I had converted it from ext2 under old 8.1.The journalling seemed to work, that is, if for some reason the system was shut down "uncleanly", during boot it would say "recovering journal" and it wouldn't go to fsck. I had chosen ext3 because I could convert my home part from ext2 preserving its content, of course. But maybe ext3 isn't good enough after all.

 

Regarding the shutdown procedure... well, I am not sure, but I thought that

shutdown -h now

was the correct way to send the computer to sleep.It should send to init 0, which entails umounting everything...err, I think. Isn't it ?

 

Well, thanks for your interest. When I'll have time, I will try some rescue procedure I have googled. Not yet sure if anything applies to my case, but I'll do my homework. And after all, my home part is already sc... messed up, I can't fear to do much harm.

If anything works, I'll post back.

Scrat