Changing FTP- port on NAS ix4-300d

TheDutchman · 04-23-2013, 08:03 AM

Hello,

As my ISP blocks all ports below 1024 and I do not have a router for NAT I would like to change the proftpd ftp-server settings on my ix4-300d.

I have found some dpkg strings on the filesystem so I suppose it is debian based. I found the configuration file but it is read-only. It is located at

Code:

./mnt/apps/usr/local/cfg/proftpd.xml

When I try to chmod it to rw I get a file system read-only error. When I try to umount and remount I get a busy error with 20 processes which I'm not sure that are safe to kill. Here is an overview of the mount:

Code:

root@iom-server:/initrd# mount
rootfs on / type rootfs (rw)
/dev/root.old on /initrd type ext2 (rw,relatime,errors=continue)
none on / type tmpfs (rw,relatime,size=51200k,nr_inodes=63950)
ubi0:boot on /boot type ubifs (rw,sync,noatime)
/dev/md0_vg/BFDlv on /mnt/boot type ext2 (rw,noatime,errors=continue)
/dev/loop0 on /mnt/apps type ext2 (ro,relatime)
/dev/loop1 on /mnt/etc type ext2 (rw,sync,noatime)
none on /etc type unionfs (rw,sync,noatime,dirs=/mnt/etc=rw:/mnt/apps/etc=ro)
/dev/loop2 on /oem type squashfs (ro,relatime)
proc on /proc type proc (rw,relatime)
none on /proc/bus/usb type usbfs (rw,relatime)
none on /proc/fs/nfsd type nfsd (rw,relatime)
none on /sys type sysfs (rw,relatime)
tmpfs on /mnt/apps/lib/init/rw type tmpfs (rw,nosuid,relatime,mode=755)
tmpfs on /dev type tmpfs (rw,relatime,size=10240k,mode=755)
tmpfs on /dev/shm type tmpfs (rw,nosuid,nodev,relatime)
devpts on /dev/pts type devpts (rw,nosuid,noexec,relatime,gid=5,mode=620)
/dev/mapper/md0_vg-vol1 on /mnt/system type ext4 (rw,noatime,user_xattr,barrier=                                                        1,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /mnt/pools/A/A0 type ext4 (rw,noatime,noui                                                        d32,user_xattr,barrier=0,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /nfs/Software type ext4 (rw,noatime,nouid3                                                        2,user_xattr,barrier=0,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /nfs/Films type ext4 (rw,noatime,nouid32,u                                                        ser_xattr,barrier=0,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /nfs/Torrents type ext4 (rw,noatime,nouid3                                                        2,user_xattr,barrier=0,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /nfs/Backup type ext4 (rw,noatime,nouid32,                                                        user_xattr,barrier=0,data=ordered)
/dev/mapper/2f28cb78_vg-lv3d7cb3b1 on /nfs/QuikTransfer type ext4 (rw,noatime,no                                                        uid32,user_xattr,barrier=0,data=ordered)
rpc_pipefs on /var/lib/nfs/rpc_pipefs type rpc_pipefs (rw,relatime)

The most important line is:

Code:

/dev/loop0 on /mnt/apps type ext2 (ro,relatime)

As killing all the processes is not a very good option I went to look where the loop device was mounted as read-only. I found this in the ~/initrd/common.sh file (which is writable):

Code:

###
# Search for and execute (if found) preimage script.
###
run_preimage_script()
{
	preimage_script=/usb_drive/emctools/h4c_images/preimage.sh

	echo -n "Checking for pre-image script... "
	if [ -x $preimage_script ]
	then
		echo "found."
		$preimage_script
		return $?
	else
		echo "none."
		return 0
	fi
}

###
# Extract a platform imager upgrade file. Validate each invidual image
# via md5 checksum.
###
extract_validate_image()
{
	IMAGES="apps config initrd zImage"
	USB_IMAGER_IMAGE=/usb_drive/emctools/h4c_images/*.tgz
	NoExtractionFile=/usb_drive/emctools/h4c_images/noextraction
	tempFolder=/usb_drive/emctools/h4c_images/temp

	if [ -f $USB_IMAGER_IMAGE ]
	then
		echo "Image tar file found: ${USB_IMAGER_IMAGE}"
	else
		echo "Image tar file not found: ${USB_IMAGER_IMAGE}"
		return 1
	fi

	tmp=`pwd`

	# extract
	if [ -f $NoExtractionFile ] && [ -d $tempFolder ] 
	then
		echo "No need to extract Image tar file."
		cd $tempFolder
	else
		echo "Extract Image tar file."
		mkdir -p $tempFolder
		cd $tempFolder
		tar -xzvf $USB_IMAGER_IMAGE
	fi

	# Validate the images
	for IMAGE in $IMAGES; do
		if [ -f $IMAGE ]
		then
			local_md5=`md5sum $IMAGE |awk -F ' ' '{print $1}'`
			md5_from_tar=`cat $IMAGE".md5"`
			if [ "$local_md5" == "$md5_from_tar" ]
			then
				echo "$IMAGE is valid"
			else
				echo "$IMAGE is not valid"
				cd $tmp
				rm -fr /usb_drive/emctools/h4c_images/temp/
				return 3
			fi
		else
			echo "$IMAGE Not found in $USB_IMAGER_IMAGE tar file"
			cd $tmp
			rm -fr /usb_drive/emctools/h4c_images/temp/
			return 2
		fi
	done

	cd $tmp
	return 0
}

setup_rootfs_to_shutdown()
{
	# set up the rootfs just enough to finish the boot
	# and shutdown
	mkdir /sysroot/bin
	cp -r -d /bin/* /sysroot/bin/
	mkdir /sysroot/sbin
	cp -r -d /sbin/* /sysroot/sbin/
	cp -r -d /imager_contents/etc /sysroot/
	mkdir -p /sysroot/dev
	cp -r -d /dev/* /sysroot/dev/
	mkdir -p /sysroot/sys
	mkdir -p /sysroot/proc
	mkdir -p /sysroot/usr
	mkdir -p /sysroot/usr/local
	mkdir -p /sysroot/usr/local/cfg/
	cp -r -d /imager_contents/initrd_bootstrap.sh /sysroot/usr/local/cfg/

	# mount the boot filesystem
	mkdir -p sysroot/boot
	mkdir -p /sysroot/sysroot
	sleep 5
}


enable_network()
{
	echo "Enable Network" $1
	ethno=$1

	ifconfig $ethno up
 	if [ $? -eq 0 ]
 	then
 		sleep 5
		udhcpc -t 10 -n -s /etc/network/udhcpc.script -p /var/run/udhcpc.$ethno.pid -i $ethno
 		sleep 2
 	fi
 	ifconfig $ethno 

 	sleep 2
}
 
###
# Invoke the main USB imaging process.
###
do_usb_imaging()
{
	# invoke the actual imager script
	/imager_contents/emc_imager.sh usb_imaging $1 $2 $3 $4
	setup_rootfs_to_shutdown

	echo "Imaging Successful. Please remove the USB drive before the next power on"
}

###
# Normal, platform independent lifeline boot. Mount images
# and set up the rootfs.
###
do_normal_boot()
{
	echo "Found Valid EMC Image in disk..."
	MNTOPTS=loop
 	NFSDEBUGMOD=
 	NFS_SERVER=
 	NFS_ROOT=/home/soho/soho_root

	echo "Mounting apps image..."
	mkdir -p sysroot/mnt/apps
 	if [ "$NFSDEBUGMOD" = "" ] 
 	then	
		mount -o loop,ro $APPS_IMAGE sysroot/mnt/apps
 	else 
	 	#for NFS debugging
 		enable_network eth0
 		echo "mount -t nfs -onolock $NFS_SERVER:$NFS_ROOT sysroot/mnt/apps"
 		mount -t nfs -onolock $NFS_SERVER:$NFS_ROOT sysroot/mnt/apps
 		sleep 2
 	fi

	# mount config
	echo "Mounting config image..."
	losetup /dev/loop1 $CONFIG_IMAGE

	# we temporarily symlink /lib and /usr/lib to the apps image, so we can run
	# the dynamically linked fsck executable from that location in initrd context
	echo "Check config filesystem"
	ln -s /sysroot/mnt/apps/lib /lib
	mkdir -p /usr
	ln -s /sysroot/mnt/apps/usr/lib /usr/lib
	e2fsck -y /dev/loop1 > etc/fsck
	rm -rf /lib /usr/lib /usr

	# mount the writeable image
	mkdir -p sysroot/mnt/etc
	mount /dev/loop1 sysroot/mnt/etc -o noatime,sync
	# mount a unionfs of the RO content and RW image
	mkdir -p sysroot/etc
	mount -t unionfs none /sysroot/etc -o noatime,sync,dirs=/sysroot/mnt/etc=rw:/sysroot/mnt/apps/etc=ro

	# remove old System.map files
	rm /sysroot/boot/System.map-*

	# copy the System.map file to /boot
	echo "Copying System.map..."
	SYSTEM_MAP_SRC_NAME="/sysroot/mnt/apps/var/System.map-*"
	SYSTEM_MAP_DEST_NAME="/sysroot/boot/"
	if [ -f $SYSTEM_MAP_SRC_NAME ]; then
		cp -f $SYSTEM_MAP_SRC_NAME $SYSTEM_MAP_DEST_NAME
	fi

	# mount the branding image if it exists
	if [ -f $OEM_IMAGE ]; then
		mkdir -p sysroot/oem
		mount -o $MNTOPTS sysroot/boot/images/oem sysroot/oem
	fi
}

###
# Update an image in NAND flash. Locate the mtd partition by name,
# erase, and reflash.
###
update_flash_image()
{
	# search for the image dev by name
	mtdname=`grep $1 /proc/mtd | awk -F : '{ print $1 }'`
	if [ ! $mtdname ];
	then
		echo "No Image"
		return
	fi

	# get the major:minor for this mtd dev
	major=`cat /sys/class/mtd/$mtdname/dev | awk -F : '{ print $1 }'`
	minor=`cat /sys/class/mtd/$mtdname/dev | awk -F : '{ print $2 }'`

	# create the dev node, update the image
	mknod /dev/$mtdname c $major $minor
	flash_eraseall /dev/$mtdname
	nandwrite -p /dev/$mtdname $2
}

###
# Copy images from the USB drive to the image
# location on the device.
###
copy_image()
{
	IMAGES_DIR=/sysroot/boot/images
	SDKAPPS_DIR=/sysroot/boot/SDKApps
	SDKAPPS_SOURCE=/usb_drive/emctools/h4c_images/SDKApps
	SDK_ETC_DIR=/sysroot/etc
	SDK_ETC_SOURCE=/usb_drive/emctools/h4c_images/etc
	CONFIG_IMAGE=/sysroot/boot/images/config
	mkdir $IMAGES_DIR

	cp $APPS_UPGRADE_IMAGE $IMAGES_DIR
	cp $CONFIG_UPGRADE_IMAGE $IMAGES_DIR
	cp $OEM_UPGRADE_IMAGE $IMAGES_DIR
	echo "Copied apps config and oem image"

	if [ -f $KERNEL_IMAGE ]
	then
		cp $KERNEL_IMAGE $IMAGES_DIR
	fi

	if [ -f $RAMDISK_IMAGE ]
	then
		cp $RAMDISK_IMAGE $IMAGES_DIR
	fi

	if [ -d "$SDKAPPS_SOURCE" ] 
	then
		#Copy the SDK Apps
		mkdir -p $SDKAPPS_DIR
		echo "Copying SDK Apps to $SDKAPPS_DIR"
		cp -f $SDKAPPS_SOURCE/* $SDKAPPS_DIR 
	fi

	if [ -d "$SDK_ETC_SOURCE" ]
	then
		#Copy the SDK ETC
		losetup /dev/loop1 $CONFIG_IMAGE
		mkdir -p $SDK_ETC_DIR
		mount /dev/loop1 $SDK_ETC_DIR -o noatime,sync
		echo "Copying SDK Etc to $SDK_ETC_DIR"
		cp -f $SDK_ETC_SOURCE/* $SDK_ETC_DIR
		sleep 4
		umount $SDK_ETC_DIR
	fi
}

Here the important part is:

Code:

if [ "$NFSDEBUGMOD" = "" ] 
 	then	
		mount -o loop,ro $APPS_IMAGE sysroot/mnt/apps
 	else 
	 	#for NFS debugging
 		enable_network eth0
 		echo "mount -t nfs -onolock $NFS_SERVER:$NFS_ROOT sysroot/mnt/apps"
 		mount -t nfs -onolock $NFS_SERVER:$NFS_ROOT sysroot/mnt/apps
 		sleep 2
 	fi

This is when I try to see what the variable contains:

Code:

root@iom-server:/# echo $NFSDEBUGMOD

root@iom-server:/#

I tried changing

Code:

mount -o loop,ro $APPS_IMAGE sysroot/mnt/apps

to

Code:

mount -o loop,rw $APPS_IMAGE sysroot/mnt/apps

and

Code:

mount -o loop,$APPS_IMAGE sysroot/mnt/apps

followed by a reboot, but the common.sh still somehow gets replaced by the original file and the loop device is always mounted read-only..

Who can help me with this ?

Thanks already

mostlyharmless · 04-24-2013, 01:34 PM

I have a ix2-200, similar setup originally. There is a trick on these to getting on root, mounting the image file rw and changing it so that it doesn't get over-written. You might need serial console access if you want to be safe. I'll see if I can find my notes; I got most of them originally over at NAS-Central http://forum.nas-central.org

Unfortunately I got tired of hacking it and ended up installing ARMEDSlack on it, which works much better and is easier to modify (like installing SANE for a network scanner) or to maintain. Of course that's a much bigger and somewhat more drastic step.

mostlyharmless · 04-24-2013, 03:26 PM

WARNING, following the steps below may brick your device.

I haven't found my notes yet, but was thinking about what I did in the past. You didn't say how you were accessing your disks/system. If you take your disks out and connect them another machine to explore them, you should find a software RAID setup, which you should be able to assemble and mount. On that RAID system you should find a directory with the ext2 images of the config and apps directories, which you should be able to mount with -loop as you describe above, rw, and then make the modifications to the files. Unmount the loop devices, unmount the RAID, remove the drives and replace them in your NAS, and reboot. See http://www.krausam.de/?p=33 for some details, though the page only talks about enabling ssh...

For 2 disk RAID 1:
If you want to be safe(r) but don't want to/can't get serial port access to your device, remove the RAID and only modify one of the drives, keeping the other one aside. (You'll have to assemble the RAID with the "missing" parameter) Put the modified disk in and verify it works. If it does, wipe the other disk and rebuild the RAID as though it were a disk replacement. If it doesn't, you should be able to use the disk you put aside. I assume though that the ix4 has 4 disk RAID, setup as RAID-5, which might make this actually less safe than assembling the entire RAID.

You should look at this:
https://github.com/basilfx/ix4-300d/...er/TUTORIAL.md
which will show you how to get a serial console to rescue your system if it bricks.