Error messages when loading persistant image

[moon]

I created a persistant disk image on the hard disk using the menu "Knoppix->Configure->Create a persistent Knoppix disk image". After I create it, I reboot and when I get the persistent image dialog, I answer yes and the image loads with no errors. The second time I reboot and every time thereafter, I get the list of error messages shown at the end of this post. I have a newer laptop using Knoppix 5.1.1 DVD and its hard drive is a SATA drive and is recognized as sda. I have an older desktop using the Knoppix 5.1.1 CD and its hard drive is recognized as hda. The same thing happens on both computers. I googled this and found a few people that have had this problem in the past but no solutions were given. I have tried formatting the partition holding the persistent image as both ext2 and ext3 and the same thing happens in both cases. Even though I get these error messages, my settings seem to be getting saved from one session to the next. Should I just ignore these messages?

note - I tried an experiment once and modified the exit script to output the error messages when the umount command was given and the message was something like "device busy" or "file system busy" - sorry, I didn't write it down

note2 - This sounds very similar to a problem that was just posted today by Alexey931 titled "USB Flash unmounted incorrectly on every boot".

Here are the messages:

************************************************** ******************************

e2fsck 1.40-WIP (14-Nov-2006)
/media/sda6/knoppix.img was not cleanly unmounted, check forced.
....

Deleted inode xx has zero dtime. Fix? yes
.....

Block bitmap differences: ......................
Fix? yes

Free block counts wrong for group #0 .....
Fix? yes

Free block counts wrong .............
Fix? yes

Inode bitmap differences: ..............
Fix? yes

Free inode count wrong for group #0 ..............
Fix? yes

Free inodes count wrong .........................
Fix? yes

[moon]

If anyone is able to load a persistent image and NOT get error messages, could you post answers to the following to help me troubleshoot?

1. What is the size of the partition you are using for the persistent image and how much of the partition is used? You can get these answers by running QTParted.

2. What is the file system of the partition? ext2, ext3, etc.. Also in QTParted (although sometimes I think running Partition Manager in Windows gives better answers).

3. When you shut down the system what is the list of directories that are unmounted? UNIONFS, KNOPPIX.IMG, etc..

Thanks in advance.
Bill

[moon]

Another question for those with no error messages. Sorry I didn't get it into the previous post.

When booting, what options are you selecting in the persistent image dialog? The default options are: home, system, init.

[johnrw]

I'll confirm this bug. I decided to try using the kernel's Emergency Remount Read-Only hotkey. On the next reboot... the volume was mounted without an error.

This is what I was getting on every bootup except the first one after I created the image, like you.
I also use the defaults offered in the mount dialog.

e2fsck 1.40-WIP (14-Nov-2006)
/media/hdb6/knoppix.img was not cleanly unmounted, check forced.

Pass 1: Checking inodes, blocks, and sizes
Deleted inode 102 has zero dtime.
Fix? yes
Deleted inode 103 has zero dtime.
Fix? yes
Deleted inode 104 has zero dtime.
Fix? yes

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

Block bitmap differences: -(8231--8373) -(12288--12299) -(14336--14765)
Fix? yes
Free blocks count wrong for group #0 (30624, counted=31209).
Fix? yes
Free blocks count wrong (547538, counted=548123).
Fix? yes
Inode bitmap differences: -(102--104)
Fix? yes
Free inodes count wrong for group #0 (15897, counted=15900).
Fix? yes
Free inodes count wrong (378704, counted=378707).
Fix? yes

/media/hdb6/knoppix.img: ***** FILE SYSTEM WAS MODIFIED *****
/media/hdb6/knoppix.img: 5293/384000 files (0.8% non-contiguous), 219877/768000 blocks
>> /home/knoppix mounted OK from /media/hdb6/knoppix.img.
>> Read-only CD/DVD system successfully merged with read-write /media/hdb6/knoppix.img.
Network device eth0 detected, DHCP broadcasting for IP. (Backgrounding)
INIT: Entering runlevel: 5
Starting Common Unix Printing System: cupsd.

root!tty1:/# cat /dev/vcs1 > /home/knoppix/Desktop/knoppix.img.error0

Then after I did the Emergency RO remount... and powered off... Then rebooted... and mounted the image without errors and then rebooted using the "reboot" command... I got this on the next startup.

e2fsck 1.40-WIP (14-Nov-2006)
/media/hdb6/knoppix.img was not cleanly unmounted, check forced.

Pass 1: Checking inodes, blocks, and sizes
Deleted inode 102 has zero dtime
Fix? yes
Deleted inode 103 has zero dtime
Fix? yes
Deleted inode 104 has zero dtime
Fix? yes
Deleted inode 105 has zero dtime
Fix? yes
Deleted inode 106 has zero dtime
Fix? yes
Deleted inode 107 has zero dtime
Fix? yes

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

Block bitmap differences: -(804--1122) -(1124--1309) -(26628--26779) -(30720--30743)
Fix? yes
Free blocks count wrong for group #0 (30528, counted=31209)
Fix? yes
Free blocks count wrong (541949, counted=542630).
Fix? yes
Inode bitmap differences: -(102--107)
Fix? yes
Free inodes count wrong for group #0 (15894, counted=15900).
Fix? yes
Free inodes count wrong (378605, counted=378611).
Fix? yes

/media/hdb6/knoppix.img: ***** FILE SYSTEM WAS MODIFIED *****
/media/hdb6/knoppix.img: 5389/384000 files (0.9% non-contiguous), 225370/768000 blocks
>> /home/knoppix mounted OK from /media/hdb6/knoppix.img.
>> Read-only CD/DVD system successfully merged with read-write /media/hdb6/knoppix.img.
Network device eth0 detected, DHCP broadcasting for IP. (Backgrounding)
INIT: Entering runlevel: 2
Starting Common Unix Printing System: cupsd.

root!tty1:/# cat /dev/vcs1 > /home/knoppix/Desktop/knoppix.img.error1

Now I am not one who knows enough about disk repairing in the linux world, yet... to be able to see what is being done here... but in case someone wants to jump on this bug with us... I hope that this helps.

Now I do use Gilles van Ruymbeke's minirt_511a.gz, when I am booting an knoppix.iso residing on a fixed disk. I did take a look at the files used and besides Gille's different loop.ko file... everything else... well they looked to be the same. Minirt_511a.gz also has a cloop.ko module, that I didn't see in the knoppix minirt.gz.

So there is something preventing a successful umount of the filesystem.
Here is the output of the mount command by itself.

/dev/root on / type ext2 (rw)
/ramdisk on /ramdisk type tmpfs (rw,size=826960k,mode=755)
/UNIONFS on /UNIONFS type aufs (rw,br:/ramdisk:/KNOPPIX:/KNOPPIX2)
/dev/sda2 on /cdrom type vfat (ro,nodev,fmask=0022,dmask=0022,codepage=cp437,ioc harset=iso8859-1)
/dev/cloop on /KNOPPIX type iso9660 (ro)
/dev/cloop2 on /KNOPPIX2 type iso9660 (ro)
/proc/bus/usb on /proc/bus/usb type usbfs (rw,devmode=0666)
/dev/pts on /dev/pts type devpts (rw)
/dev/hdb6 on /media/hdb6 type ext2 (rw,nosuid,nodev)
/media/hdb6/knoppix.img on /KNOPPIX.IMG type ext2 (rw,loop=/dev/loop0)
persistent on /UNIONFS type aufs (rw,br:/KNOPPIX.IMG:/KNOPPIX:/KNOPPIX2)

[johnrw]

So what is the shutdown script that suppresses error messages?
I usually use boot to a console... and then I will type "init 5" when I want to use X.
When I want to end the session I just type reboot or poweroff as the need arises.

echo $PATH
/home/knoppix/.dist/bin:/sbin:/bin:/usr/sbin:/usr/bin:/usr/X11R6/bin:/usr/local/sbin:/usr/local/bin:/usr/games:/usr/NX/bin:.


locate reboot
/etc/init.d/knoppix-reboot
/etc/init.d/reboot
/etc/rc6.d/S90knoppix-reboot
/sbin/grub-reboot
/sbin/reboot

and a locate poweroff yields...
locate poweroff
/lib/modules/2.6.19/kernel/drivers/char/ipmi/ipmi_poweroff.ko
/sbin/poweroff
/usr/lib/klibc/bin/poweroff
/usr/share/man/man8/poweroff.8.gz

So a poweroff command executes /sbin/poweroff, and a reboot command executes /sbin/reboot

I will look into those later today... but I think a good goal would be to enable errors to be reported on shutdown.

[moon]

Hi John,

The exit script I was referring to is in the file /etc/init.d/knoppix-halt. It is also in /etc/init.d/knoppix-reboot (These files might be identical). At line 204 are 3 umount commands, each followed by "2>/dev/null" (no quotes). The "2>/dev/null" is what eats any error messages. If you delete these 3 occurrences of "2>/dev/null" you will see the error messages. You might want to add some newlines to keep the messages from running into each other and wrapping. It also might not hurt to put in a read command to stop the output so you can look at it but I don't think that I did that.
I also just noticed a comment at line 7 of each of these files indicating that the author found it "difficult to unmount everything cleanly" all the time. Hmmm...

Regards,
Bill

[johnrw]

Hi Bill,

About knoppix-halt and knoppix-reboot. They are the same file... as one is just a link to the other. I forget which is the link.

Okay... I tried changing line 204 to:

umount -d "$mp" >>/home/knoppix/0.out 2>&1 || { umount -r "$mp" >>/home/knoppix/1.out 2>&1 ; umount -l "$mp" >>/home/knoppix/2.out 2>&1 ; }

Here is 0.out

umount: /UNIONFS: device is busy
umount: /UNIONFS: device is busy
umount: /UNIONFS: device is busy
umount: /UNIONFS: device is busy
umount: /KNOPPIX.IMG: device is busy
umount: /KNOPPIX.IMG: device is busy

Here is 1.out

umount: persistent busy - remounted read-only
umount: /KNOPPIX.IMG: device is busy
umount: /KNOPPIX.IMG: device is busy

and 2.out is zero bytes... but there.
I got some errors in the console when running shutdown about can't create the file. I suspect they were about 2.out.
It seems as though the image is unmounted though... there may still be some open files.
How can I tell fsck to save files like chkdsk .chk files?

[johnrw]

Wouldn't it be better if KNOPPIX.IMG was unmounted first?
How can UnionFS be unmounted when it contains mounted filesystems?

[johnrw]

Well there is this word persistent that is bugging me. In the output of mounts above... and in the knoppix-image menu options. IE. "Add as persistent, writable system area" Now I unchecked that... and mounted it. Upon reboot, and remount, no error.

So what does that persistent aspect/option mean?

I also tried something crazy... I threw in a fuser -v -m right before the line 204 in knoppix-reboot... and at the first invocation...
it mentioned 2 files were still in use... pump and init.

I also tried something else... saying no to all knoppix.img mounting options... and then doing a
umount /KNOPPIX.IMG, which was successful. Then a restart of knoppix-image mounts without a fsck check. So it is the way the system is being brought down.

[takayama]

If you kill the "pump" before the shutdown, then fsck will not happen.
(If you have a NFS mounted disk, unmount it before you kill the pump.)

This trouble of the persistent home seems to be related to
the debian bug report
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=261803
"pump stays around and stops /var/ and /usr/from being unmounted"

To remastering, /etc/init.d/knoppix-halt should be modified to stop pump properly.

Credit of this post: knoppix math developping mailing list in Japanese.