DSM-G600, DNS-3xx and NSA-220 Hack Forum

Hi bjby - Thanks for the reply!

Yes I read it several times, but because of missing/incorrect info, I was afraid of trashing my
disks.  After spending about 3 days on this problem, I finally reached the point where I was
about to reformat the drives and reload from backup., Your reply inspired me to go for broke
since I really didn't have much to lose. 

The following describes how to clean up a RAID1 file system with the error:
The filesystem size (according to the superblock) is 488116951 blocks
The physical size of the device is 488116928 blocks
Either the superblock or the partition table is likely to be corrupt!
--- and/or general problems that can be corrected with e2fsck.

I've attempted to clarify/add to the documentation for the benefit of anyone else may find
themselves in the same situation.

There were a number of problems / outstanding questions can anyone please help fill in the
missing info / suggest how to fix the problems or make suggestions on how to improve the
clarity of the procedure?

Outstanding Questions / Problems / Documentation Errors

1. If you have been using SSH to access your DNS323 (as suggested in the installation),
what changes have to be made to the procedure so that you don't end up locking yourself
out?  (This was my biggest concern and why I didn't attempt sooner.)

(I'm not sure if the installation provided the necessary telnet setup, or if I installed it
when I was setting things up initially.  I ended up doing a full install of all the packages
and busybox3, utelnetd, and usb-storage.ko to do some other things.  This means I have
about 500MB of excess stuff I will likely never use.)

2. Does ./reload.sh establish a completely new telnet environment?  If someone had set
up a password for telnet would it survive or mess things up in any way?  Are there any
other potential traps to watch out for?

3. What needs to be done about the fan / temperature control (if anything)?
(On my system the fan seemed to run on slow mode.  I think the temperature might
have been a bit higher than normal, but I didn't fry anything checking 2 x 2TB in RAID1
with about 600GB of data - which took about 5-6 hours to the best of my recollection.)

The original documentation cites the following:

     echo 150 >/sys/class/i2c-adapter/i2c-0/0-003e/pwm1
  and read current speed with:
     cat /sys/class/i2c-adapter/i2c-0/0-003e/fan1_input
  (100 in pwm1 = 3150 rpm, 150 in pwm1 = 4650 rpm)
  You can read the current temperature with:
     cat /sys/class/i2c-adapter/i2c-0/0-0048/temp1_input

on my DNS323 under fun_plug v0.5 the path /sys/class doesn't exist

4. Does the command:

        mdadm -A /dev/md0 /dev/sd[ab]2

    need to be modified for any reason (i.e. What about JBOD / RAID0)? 

    (When I first saw the command: mdadm -A /dev/md0 /dev/sd[ab]2  I wasn't
    sure if [ab]2 should be entered as is, or if it was shorthand for something.
    For the benefit of the uninitiated, you can cut and paste in this
    command exactly as is if your system is running RAID1.

5.Can someone with 2 separate disks please confirm part works???
>
>If you have configured 'two separate disks', you can now check
>filesystems with:
>
>    e2fsck /dev/sda2
>    e2fsck /dev/sdb2
>

6. What happens to /lost+found ??  I pressed y to create it.  I assumed that I should
find a directory full of fragments (similar to what you get out of CHKDSK in windows/dos).
After rebooting, the only /lost+found I could see was under '/' and it existed before I ran
e2fsck and appeared to be empty.  Why did I get  "/lost+found not found.  Create<y>?"
answer Y to the question, and then not find a lost+found directory on /dev/md0?
Does /lost+found not survive reboot?

7. What changes need to be made to the script so that the DNS323 will restart? 
When I ran reboot as described in the procedure the system hung.  The only way I could
get control back was to pull the plug for a few seconds and then switch back on.

================== Procedure for File System Repair

[--- Tested with FW 1.09 on HW Rev C - fun_plug v0.5 on 2011/03/14 ---]

Setup fun_plug and telnet or SSH into your DNS323: 
(See: http://dns323.kood.org/dsmg600/howto:fun_plug )

---- Install
  cd /ffp                                         (or /mnt/HD_a2)
  rsync -av inreto.de::dns323/fsck .

---- Change environment 
  cd /ffp/fsck                          (or /mnt/HD_a2/fsck)
  ./reload.sh

This will stop all processes and initiate a reboot. Your SSH or telnet
session is aborted.  The web interface and SSH are now dead so 
telnet is the only way to access your DNS323.   You will also find 
that the Hard Drive lights don't come on - they are not supported 
in this maintenance environment.

You should be able establish a new telnet session after 10-20 seconds. 

Login as 'root', there's no password.

After boot, run:

    free

Look for the 'swap' line  to verify that swap has been correctly
enabled.  The 'total' column must be non-zero. This is essential, 
e2fsck will fail without sufficient  swap space.

---- Check / Repair File System

> [can someone with 2 separate disks please confirm this works???]  <<<<<<<<<<<
>
>If you have configured 'two separate disks', you can now check
>filesystems with:
>
>    e2fsck /dev/sda2
>    e2fsck /dev/sdb2
>

If you are using RAID, set it up before checking the filesystem:

    mdadm -A /dev/md0 /dev/sd[ab]2
    e2fsck /dev/md0

e2fsck may take some time, and ask for lost+found:

  /lost+found not found.  Create<y>?

Just press Enter. Answering 'n' is ok as well, but e2fsck will not repair
your file system, and you will get a message  that your
 'Filesystem still has errors' at the end.

If you get an error that reads some thing like:
The filesystem size (according to the superblock) is 488116951 blocks
The physical size of the device is 488116928 blocks
Either the superblock or the partition table is likely to be corrupt!

then you will need to use the commands:  

   e2fsck -f /dev/md0
   resize2fs /dev/md0
   e2fsck /dev/md0

the last e2fsck should produce output that looks something like:

e2fsck 1.41.0 (10-Jul-2008)
/dev/md0: clean, xxxxxx/xxxxxxxxxx files, xxxxxxxxxxx/xxxxxxxxxxxx blocks


== Rebooting ==

When done, run:

 reboot

(This causes the system to hang - the only way to get control back is disconnect the
power by pulling the plug for a few seconds and then power back on with the power 
switch on the front.)

---- ADDITIONAL INFO for ADVANCED USERS

By default, Linux 2.6.12.6 will be booted (check with 'uname -r'). 
More recent kernels are also available. To boot them, pass a
zImage file to reload.sh, e.g. 
   ./reload.sh zImage-2.6.25.1
Note that 2.6.25.1 and 2.6_orion will not work on CH3SNAS.

Manual pages for e2fsck can be found at:

  http://linux.die.net/man/8/e2fsck

For Parted documentation, see

  http://www.gnu.org/software/parted/

Hi bjby - thanks for the reply

Yes, I did run from telnet, and I know that the intent is only a minimal maintenance system--no problem there. 

RE: #7 I was wondering if there was a way to make return to normal more graceful after maintenance is done?
Having to pull the plug, restart, kill the script, and reboot again is a bit messy. It would be great if there was a
way to at least power right off, or force a proper reboot so the regular fun_plug starts!

RE:#3 Original docs said there was a way to manually set the fan, I would assume there is a way to do it, just
a different way.  I'm hoping somebody who knows can shed some light on it.

I'll check the other stuff in a day or two after the current copy operation is complete... loading about 1TB of data
onto the DNS323 which isn't speedy!!!

Thanks for the reply dhub.... your package looks cool and I'll check it out when I get a chance... don't think it will solve the problem though...

I'm wondering if the problem is that the RAID1 volume isn't mounted and a soft reboot using the reboot script doesn't clean things up?  Could that be it?

root@NAS3:/mnt# mount
rootfs on / type rootfs (rw)
proc on /proc type proc (rw,nodiratime)
usbfs on /proc/bus/usb type usbfs (rw)
sysfs on /sys type sysfs (rw)
tmpfs on /dev type tmpfs (rw)
devpts on /dev/pts type devpts (rw)

I tried reboot -f and that hung too....  I'm thinking the easiest way to solve the problem is to power down
the DNS323.  Is there a command to cause to DNS323 to power off????

Then when the power was turned back on the box would boot up into whatever fun_plug was there....
A lot easier than crawling under the desk to unplug and replug the DNS323!