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

Oh, he is running RAID 1, so it it might be /dev/md<something> instead of /dev/sda4.  I'm not that familiar with RAID.

Compare yours to mine and it looks like you are missing sda4:

root@Toaster:~# fdisk -l /dev/sda

Disk /dev/sda: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x00000000

   Device Boot      Start         End      Blocks   Id  System
/dev/sda1               1          66      530113+  82  Linux swap / Solaris
/dev/sda2             131      121601   975715807+  83  Linux
/dev/sda4              67         130      514080   83  Linux

Partition table entries are not in disk order
root@Toaster:~#

Looks like you have 260 cyls at the end of your disk that are not used.  Maybe you can use fdisk to create a sda4 partition that is about 62 cyls.

Last edited by karlrado (2010-11-04 22:40:34)

nicko wrote:

Code:

/ # df -h /mnt/HD_a4/.systemfile
Filesystem                Size      Used Available Use% Mounted on
rootfs                    9.7M      7.5M      1.7M  82% /
/ #

So /mnt/HD_a4 is on the ramdisk, and the upnp server is storing files in there. I'm pretty sure that's wrong.

Run

fdisk /dev/sda

and then type m for help.

You could have a damaged partition table.  Might try using the v command to verify what you have.  Also type p to see the current table.

If you want to proceed, then use the n command to create a new partition.  Specify p for primary when prompted.  If you get a choice, select 4 for the partition number.  I can't help past this point because my disk is full and I can't run fdisk any farther.  But specify 64 for the number of cylinders.  I was wrong about 62.

When done defining the partition, you'll go back to the fdisk menu.  Type p to see what it looks like and if you like it, w to write the partition table and exit.  But make sure that the only change you see when you type p, compared to the first time you typed p is the added partition.  The info for the other partitions should not change.  If other partition info changed, I would use q to exit.

You won't get the partition table to match mine exactly because your a4 will be at the end of the disk instead of starting at 67.  That probably should not matter.

The reason why you need to have an sda4 is because the operating system is expecting it to be there.  It tries to mount it to /mnt/HD_a4.  The OS uses HD_a4 for workspace.  If sda4 is not there, the mount fails, and then the space referenced by /mnt/HD_a4 is really the rootfs.  The OS then copies a bunch of stuff to /mnt/HD_a4, filling up the rootfs.

bjby wrote:

I state the fordem mantra.

Before doing this, backup the files you REALLY dont want to lose.

You may need to temporary turn off the upnp (which fills your rootfs) to do this.

Hmmm....

/ # fdisk /dev/sda

The number of cylinders for this disk is set to 121601.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Segmentation fault
/ #

Getting messy.

The seg fault can be caused by at least three things:

1) The partition table is really screwed up (unlikely, since fdisk -l /dev/sda worked)
2) The lack of space in rootfs is making fdisk bomb, as bjby suggested
3) You are running a broken version of fdisk.

I actually never got the fdisk to work that came with the stock firmware as /usr/bin/fdisk.  I get the same result as you when I execute /usr/bin/fdisk, and my system is working.  The fdisk in my optware installation does work.  I never tracked down what is wrong with /usr/bin/fdisk.  There may be forum or wiki posts about it.

Anyway, you still have options:

1) You should have a backup (or two).  You could put only this disk in the right slot of the DNS and do a factory reset or whatever it takes to reformat the disk and then restore your files to it from your backup.

2) You could put the disk in a Linux box and run an fdisk-like utility there to create the partition, and then put the disk back in the DNS.  This would require some extra Linux-know-how.

3) Try to figure out how to fix the stock /usr/bin/fdisk on the DNS and create the partition using that.  As bjby said, you'd have to get some breathing room by shutting down the upnp software that is writing to HD_a4 and perhaps delete the big file in HD_a4 to create some space.  I don't know how to shut down the upnp stuff.  And, I don't know exactly how to fix /usr/bin/fdisk once you do that.  Maybe someone else can help.  Installing optware to get its fdisk is an option (lots of work).  Is there a fdisk in ffp?

The first option is a sure-shot, but will take a lot of time to do the restore.

Right, I installed the fixed fdisk and created the new partition as sda4 (took the default to use all the spare cylinders):

/mnt/HD_a2/ffp # cd tmp
/mnt/HD_a2/ffp/tmp # ./fdisk /dev/sda

The number of cylinders for this disk is set to 121601.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/sda: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x00000000

   Device Boot      Start         End      Blocks   Id  System
/dev/sda1               1          66      530113+  82  Linux swap / Solaris
/dev/sda2              67      121340   974133405   83  Linux
/dev/sda4          121341      121601     2096482+  83  Linux

Command (m for help):

I then rebooted the box, but no change - it still thinks that the rootfs has all the AV stuff on it... can I somehow unlink that? /dev/HD_a4 still seems to be on the rootfs...

Got a quick scare - when I re-logged in on the web gui, I was prompted to re-init the disk! Luckily, there is an option to skip that, and it seems the disk data is ok (for the moment).

I took the other disk (its RAID 1) out and am copying the data from that as I type...

bgravato wrote:

creating the partition isn't enough... you probably need to format it too...

something like:

mke2fs -j /dev/sda4

After partitioning, you need to:
1. Make /dev/sda1 a swap partition. Just to be on the safe side, I turned swap off and added the commands  to check swap space (free, cat):

free
cat /proc/swaps
swapoff /dev/sda1
free
cat /proc/swaps
mkswap /dev/sda1
swapon /dev/sda1
free
cat /proc/swaps

2. Regarding mke2fs, I'd suggest reading the man page (http://linux.die.net/man/8/mke2fs) and decide what switches suites your needs. I'd look into -j -T -c and maybe -m. I used:

mke2fs -j -m 0 -T largefile4 /dev/sda4

Last edited by scaramanga (2010-11-05 10:35:09)

nicko wrote:

.... /dev/sda1 is already a swap partition (see the fdisk output above).

That's not enough. You need to check the output from "cat /proc/swaps" to make sure the OS knows it should use this partition for swap.