I’m currently using kernel 2.4.18-6mdk, don’t ask why. One morning I boot my system and there are file system problems. When I run e2fsck, I get the following errors:

Inode XXX is in use, but has dtime set. Fix?
Inode XXX has imagic flag set. Clear?

This pair of errors repeat many of 1000’s of times, to the point where I terminated the e2fsck check.

(Just to note, the other errors sprinkled throughout the check include)

Inodes that were part of a corrupted orphan linked list found. Fix?

and

Inode XXX was part of the orphaned inode list.

and

Inode XXX was part of the orphaned inode list. IGNORED.

I know there are known FS corruption issues with the Mandrake 8.2 release but for some reason, I don’t think this is what I’m witnessing as those issues seem to manifest in a single file or couple of files. The many thousands of inodes with issues indicates to me something else happened. Here are my questions:

1) What is e2fsck trying to tell me, perhaps in lay terms?
2) From what I think I know, there is a copy of the magic number for each block group to help in recovery from FS corruption. How it is possible that I see so many of these imagic errors (thousands)?
3) Is it more likely that these errors were provoked by SW issues (perhaps turning the power off at inappropriate times, etc.) or HW (faulty HDD)?

Any direction re: this issue will be greatly appreciated.

"When I run e2fsck, I get the following errors:"

e2fsck must be run against an umounted file system. I suggest that you boot a rescue CD or live CD and run e2fsck against every Linux partition except swap. If your live CD mounts all of your disk partitions then umount them before running e2fsck.

"This pair of errors repeat many of 1000’s of times, to the point where I terminated the e2fsck check."

Use the -p option and e2fsck will not ask you for confirmations:
e2fsck -p /dev/hda1

"1) What is e2fsck trying to tell me, perhaps in lay terms?
Inode XXX is in use, but has dtime set. Fix?
Inode XXX has imagic flag set. Clear?"

These errors are minor and fixable.

"Inodes that were part of a corrupted orphan linked list found. Fix?
Inode XXX was part of the orphaned inode list."

e2fsck has found directoies, files, or file fragments that it doesn't know where to put them. e2fsck will put them in the lost+found directory under a number name. Later you will have to check lost+found by hand to see where these things belong.

"Inode XXX was part of the orphaned inode list. IGNORED."

Good news, e2fsck found out where an orphan actually belonged.

"to the point where I terminated the e2fsck check."

It is not a good idea to terminate e2fsck in the middle. You can create even more errors by doing so.

"3) Is it more likely that these errors were provoked by SW issues (perhaps turning the power off at inappropriate times, etc.) or HW (faulty HDD)?"

Most likely improper shutdown. Second most likely is bad hard drive.


___________________________________
Be prepared. Create a LifeBoat CD.
http://users.rcn.com/srstites/LifeBo...home.page.html

Steve Stites

"e2fsck must be run against an umounted file system. I suggest that you boot a rescue CD or live CD and run e2fsck against every Linux partition except swap. If your live CD mounts all of your disk partitions then umount them before running e2fsck."

Yup, did this, used tomsrtbt.

"Use the -p option and e2fsck will not ask you for confirmations:
e2fsck -p /dev/hda1"

Yup, I used e2fsck -n /dev/hda1

"These errors are minor and fixable."

Great! Is it not surpising that I'm witnessing these issues? Are these issues common? Am I just going through the normal motions of using a Linux system?

"Most likely improper shutdown. Second most likely is bad hard drive."

What is considered an improper shutdown? Not using 'halt' or 'shutdown'? There are many instances where I incorrectly shutdown my device, yet this is the first instance in which I've seen FS integrity issues like these.

Thanks very much for your input. I think I need to curl up with a ext3 FS book to understand the details a bit more.

The main issue I'm having is that I'm researching the feasibility of an old Mandrake distro for a future embedded product, again don't ask why . The potential for the user to flip the switch to turn of the power w/o properly shutting down the system is high. Having these FS integrity issues may not be acceptable for this configuration. If you are suggesting that these FS issues are "normal" then perhaps I can work e2fsck into the normal boot process of the device.

Your comments are of course welcome.

Thanks again.

"What is considered an improper shutdown? "

You must umount all devices before you shut down. umount writes all dirty buffers in the cache to the hard drive. Not doing so results in a corrupted file system.

"I think I need to curl up with a ext3 FS book to understand the details a bit more."

If you run fsck with the wrong filesystem type then fsck will try to repair a filesystem to the wrong specs and make matters worse. If you are checking an ext3 system then use:
fsck -t ext3 /dev/hda1

"There are many instances where I incorrectly shutdown my device, yet this is the first instance in which I've seen FS integrity issues like these."

Most likely the device sat idle long enough before shutdown to get all of the dirty cache buffers written out. You could issue a sync command or a umount command. In addition to syncing umount also marks the partition as cleanly umounted.

"Having these FS integrity issues may not be acceptable for this configuration."

In addition to sync and umount you might read:
man tune2fs
for ideas on how to manage this problem.

___________________________________
Be prepared. Create a LifeBoat CD.
http://users.rcn.com/srstites/LifeBo...home.page.html

Steve Stites