e2fsck consistently finds corrupt inodes...

e2fsck consistently finds corrupt inodes...

Post by Jac » Thu, 14 Apr 1994 00:37:10



I posted this a while back, but never got any replies. The problem still
occurs, and with what appears to be greater ferocity. I have looked briefly
at the code for e2fsck v0.4, but it ain't obvious whats happening. I am
vaguely concerned about the integrity of my hard disk.

***

...Replies via email please...

I've just installed 99PL13 and I'm having a problem with e2fsck.
I have two separate bootdisks:
- The original one, from when setup was first run to create a bootdisk, and
  it boots, mounting / as readonly, and doesn't actually run e2fsck (as far
  as I can see)
- A new bootdisk created by remaking zdisk (wanted to get rid of generic
  stuff like the mitsumi driver) which doesn't mount / as readonly and e2fsck
  runs, finding lots of apparently corrupt inodes. (NB I never recompiled the
  kernel). I also notice it now does an init 1 before adding swap.

Not sure why the original bootdisk didn't complain, they both use the same
kernel configs, remaking zdisk didn't actually recompile anything but it still
got rid of the generic stuff I didn't want. The filesystem was installed as
type ext2 which uses i_links_count, unlike other types (they use i_nlinks).
I never had the problem with the last patchlevel I had installed. NB The
Slackware distribution allowed me to blast away the previous installation and
created a new ext2 partition (without any problems).

Not really understanding what was going on, I tried a simple test, took
the system down to single user and ran e2fsck until the corruption was clear,
then rebooted (init 6)... sure enough, more corruption had appeared. I did
this a couple of times, and every time a couple of inodes links_count were
wrong.

I know I shouldn't run fsck on mounted (even though it was quiescent)
filesystems, but I was unable to unmount it at single user (question
is, do I somehow need to boot a miniroot so I can fsck the root
filesystem)... and anyhow, I still get the corruption when I boot up
(the system always shuts down clean and unmounts everything).

Will the problem go away with time (obviously I don't like it keep having
to do fsck at bootup) or is there a remedy, and I don't wanna haveto
reformat the partition with new parameters!?

I'd appreciate reply's by email...

Ta,
Jack

ORIGINAL BOOTUP SEQUENCE:
Partition check:
  hda: hda1 hda2 hda3
[EXT II FS 0.3c, 93/08/05, bs=1024, fs=1024, gc=23, bpg=8192, ipg=2008]
VFS: Mounted root (ext2 filesystem) readonly.
Adding Swap: 16380k swap-space

NEW BOOTUP SEQUENCE:
Partition check:
  hda: hda1 hda2 hda3
[EXT II FS 0.3c, 93/08/05, bs=1024, fs=1024, gc=23, bpg=8192, ipg=2008]
VFS: Mounted root (ext2 filesystem).

Feb 21 23:37:50 init[1]: version 2.4 booting
Adding Swap: 16380k swap-space
e2fsck 0.4, 93/11/01 for EXT2 FS 0.4, 93/11/01
Inode 32130 not used with links_count not null. Repair? yes
Inode 45503 not used with links_count not null. Repair? yes
----------------------------
FILE SYSTEM HAS BEEN CHANGED
----------------------------
Remounting root device with read-write enabled.
...
Feb 21 23:39:22 init[1]: Entering runlevel: 6
--

Illud Latine dici non potest
You can't say that in Latin

 
 
 

1. e2fsck consistently finds corrupt inodes...

...Replies via email please...

I've just installed 99PL13 and I'm having a problem with e2fsck.
I have two separate bootdisks:
- The original one, from when setup was first run to create a bootdisk, and
  it boots, mounting / as readonly, and doesn't actually run e2fsck (as far
  as I can see)
- A new bootdisk created by remaking zdisk (wanted to get rid of generic
  stuff like the mitsumi driver) which doesn't mount / as readonly and e2fsck
  runs, finding lots of apparently corrupt inodes. (NB I never recompiled the
  kernel). I also notice it now does an init 1 before adding swap.

Not sure why the original bootdisk didn't complain, they both use the same
kernel configs, remaking zdisk didn't actually recompile anything but it still
got rid of the generic stuff I didn't want. The filesystem was installed as
type ext2 which uses i_links_count, unlike other types (they use i_nlinks).
I never had the problem with the last patchlevel I had installed. NB The
Slackware distribution allowed me to blast away the previous installation and
created a new ext2 partition (without any problems).

Not really understanding what was going on, I tried a simple test, took
the system down to single user and ran e2fsck until the corruption was clear,
then rebooted (init 6)... sure enough, more corruption had appeared. I did
this a couple of times, and every time a couple of inodes links_count were
wrong.

I know I shouldn't run fsck on mounted (even though it was quiescent)
filesystems, but I was unable to unmount it at single user (question
is, do I somehow need to boot a miniroot so I can fsck the root
filesystem)... and anyhow, I still get the corruption when I boot up
(the system always shuts down clean and unmounts everything).

Will the problem go away with time (obviously I don't like it keep having
to do fsck at bootup) or is there a remedy, and I don't wanna haveto
reformat the partition with new parameters!?

I'd appreciate reply's by email...

Ta,
Jack

ORIGINAL BOOTUP SEQUENCE:
Partition check:
  hda: hda1 hda2 hda3
[EXT II FS 0.3c, 93/08/05, bs=1024, fs=1024, gc=23, bpg=8192, ipg=2008]
VFS: Mounted root (ext2 filesystem) readonly.
Adding Swap: 16380k swap-space

NEW BOOTUP SEQUENCE:
Partition check:
  hda: hda1 hda2 hda3
[EXT II FS 0.3c, 93/08/05, bs=1024, fs=1024, gc=23, bpg=8192, ipg=2008]
VFS: Mounted root (ext2 filesystem).

Feb 21 23:37:50 init[1]: version 2.4 booting
Adding Swap: 16380k swap-space
e2fsck 0.4, 93/11/01 for EXT2 FS 0.4, 93/11/01
Inode 32130 not used with links_count not null. Repair? yes
Inode 45503 not used with links_count not null. Repair? yes
----------------------------
FILE SYSTEM HAS BEEN CHANGED
----------------------------
Remounting root device with read-write enabled.
...
Feb 21 23:39:22 init[1]: Entering runlevel: 6
--

Dic! Itane esti?
You don't say!

2. REQUEST: for Linux 0.12

3. e2fsck: root inode isn't a directory???

4. Help w/ backing up large amount of data on HP-UX platform

5. e2fsck inode problem

6. VI refresh problem

7. e2fsck: root inode isn't a directory

8. How can I recover this partition

9. Admin tips pl. : why e2fsck complains about inode 2

10. e2fsck inode count buffer error.

11. Any way to cure crashed inodes which e2fsck relinquished[failed]?

12. e2fsck: root inode not a directory [HELP!]

13. e2fsck 0.4 11/93 -- What's the deal with all these Inode messages?