Fsck Errors OS 11.2

English Applications
#1

Hello,

Here are the basics of my systems that might be important for this problem:

Server:
openSUSE 11.2 64bit, 3ware RAID 5 + RAID 1 (/var/ + swap), all FS’s are Ext4

Development:
openSUSE 11.2 64bit, 3ware RAID 5, all FS’s are Ext4

When I run “fsck -Asn” I get this on my server:

fsck from util-linux-ng 2.16
e2fsck 1.41.9 (22-Aug-2009)
Warning!  /dev/mapper/Fast--read-system is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
/dev/mapper/Fast--read-system contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Free blocks count wrong (20110710, counted=20110694).
Fix? no

Free inodes count wrong (5676962, counted=5676964).
Fix? no

/dev/mapper/Fast--read-system: ********** WARNING: Filesystem still has errors **********

/dev/mapper/Fast--read-system: 876638/6553600 files (0.1% non-contiguous), 6103690/26214400 blocks

And on my development computer I get:

fsck from util-linux-ng 2.16
e2fsck 1.41.9 (22-Aug-2009)
Warning!  /dev/mapper/Fast--read-system is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
System contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Free inodes count wrong (6394637, counted=6394638).
Fix? no

System: ********** WARNING: Filesystem still has errors **********

System: 158963/6553600 files (0.1% non-contiguous), 1605700/26214400 blocks

In each case I see errors. My first thought when I saw this on my server was that it must be that write operations are happening so it just gets a bit confused.

However, even on my development computer, when doing nothing and not seeing any disk activity, I get the same error (admittedly there could be writes that I just can’t see happening.)

On the server I already issued a fsck on reboot with “touch /forcefsck” and the error is still there after the reboot which causes me to believe that the file system must be okay.

So can anyone confirm that this is okay or should I be concerned and tracking down what’s going on?

#2

You should not do fschk on mounted file systems.

#3

Reg gie wrote:
> I get the same error (admittedly there could
> be writes that I just can’t see happening.)

writes to disk that you can’t see? hmmmmm…in my experience fsck is
best run on unmounted disks only (but, i’m more careful/conservative
than some)…

however, you may still have the same errors if the partition is
unmounted…i don’t know, have you tried?

and, if we assume all error are fixed when you force fsck during
boot (does it report them fixed) and THEN see the errors again when
run with the partitions mounted, it might be ok, or might not…that
is to say: i’m not able to answer your “can anyone confirm that this
is okay or should I be concerned”

–
DenverD
When it comes to chocolate, resistance is futile.
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]

#4

You are getting an inode count mismatch and the mismatch is by 1 on your development machine and by 2 on your server. This is quite possible in a mounted file system.

#5

My understanding is that this applies to correcting the file system. The “n” parameter tells it to just check the file system which will not hurt anything (read only).

#6

Thanks, that makes sense now that you mention it. I’ve been reading up on the internals of Linux.

#7

Well, I did do the fsck on reboot but I got side tracked and the system finished and started up before I could see what it was printing out. I could check the logs and see if it left information there.

Regarding doing this on unmounted drives. My concern here is that you can’t unmount all your drives because then you wouldn’t have access to anything to run so really the only way to do a complete check is through fsck on reboot, or… is there another piece of Linux magic that I am about to learn?

#8

Reg gie wrote:
> is there another piece of Linux magic that I am about to learn?

thankfully (or unfortunately, depending on how much you wanna learn)
there is always more MAGIC to learn…if not today, then just wait
and download in the morning what ‘they’ did over night…

yes, boot from a live CD and nothing on the machine is mounted! yipee!!
(or, anything on the machine may be mounted, if you wish…see?)

–
DenverD
When it comes to chocolate, resistance is futile.
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]

#9

On 2010-10-15 09:36, Reg gie wrote:
>
> syampillai;2238207 Wrote:
>> You should not do fschk on mounted file systems.
>
> My understanding is that this applies to correcting the file system.
> The “n” parameter tells it to just check the file system which will not
> hurt anything (read only).

Maybe, but it isn’t reliable either.

You can check during boot, from a rescue/live cd, or from another bootable partition. It is not a
bad idea to have another bootable partition in a linux computer. Or more than one, some times.

–
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)