Background Info
I have an embedded system running Ubuntu 12.04 which I need to upgrade to patch some security vulnerabilities. The changes will include upgrading to ubuntu 14.04, changing a few config files located in /etc, /var, /media and also updating our application software. There are 1500 or so of these devices that the upgrades will need to be pushed out to so I am leaning towards a solution that does not require physical access to the device.
Disk Info:
root@HBOX-01003d:/dev# sudo fdisk -l
Disk /dev/mmcblk2: 7851 MB, 7851737088 bytes 4 heads, 16 sectors/track, 239616 cylinders, total 15335424 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000
Device Boot Start End Blocks Id System /dev/mmcblk2p1 * 2048 100351 49152 c W95 FAT32 (LBA) /dev/mmcblk2p2 100352 1329151 614400 83 Linux /dev/mmcblk2p3 1329152 15335423 7003136 83 Linux
Disk /dev/mmcblk2rpmb: 0 MB, 131072 bytes 4 heads, 16 sectors/track, 4 cylinders, total 256 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000
Disk /dev/mmcblk2rpmb doesn't contain a valid partition table
Disk /dev/mmcblk2boot1: 2 MB, 2097152 bytes 4 heads, 16 sectors/track, 64 cylinders, total 4096 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000
Disk /dev/mmcblk2boot1 doesn't contain a valid partition table
Disk /dev/mmcblk2boot0: 2 MB, 2097152 bytes 4 heads, 16 sectors/track, 64 cylinders, total 4096 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000
Disk /dev/mmcblk2boot0 doesn't contain a valid partition table root@HBOX-01003d:/dev# sudo parted -l Error: /dev/mmcblk2rpmb: unrecognised disk label
Error: /dev/mmcblk2boot0: unrecognised disk label
Error: /dev/mmcblk2boot1: unrecognised disk label
Model: MMC MMC08G (sd/mmc) Disk /dev/mmcblk2: 7852MB Sector size (logical/physical): 512B/512B Partition Table: msdos
Number Start End Size Type File system Flags 1 1049kB 51.4MB 50.3MB primary fat32 boot, lba 2 51.4MB 681MB 629MB primary ext3 3 681MB 7852MB 7171MB primary ext3
First Imaging Attempt
From this information, I gathered that the mmcblk2 drive was the primary drive used on this system and decided to image this drive. I used dd https://help.ubuntu.com/community/DriveImaging to create the image and the second answer on this post to help write the correct commands: https://unix.stackexchange.com/questions/132797/how-to-dd-a-remote-disk-using-ssh-on-local-machine-and-save-to-a-local-disk
I used the following command to copy, compress, and send the image to a remote server.
dd if=/dev/mmcblk2 | gzip -c --fast | ssh administrator@10.3.5.99 'dd of=/home/administrator/clone/ubuntu14.gz'
This was the output:
15335424+0 records in
15335424+0 records out
7851737088 bytes (7.9 GB) copied, 2038.58 s, 3.9 MB/s
6247083+1 records in
6247083+1 records out
3198506839 bytes (3.2 GB) copied, 2023.36 s, 1.6 MB/s
The first time around, I ran this command to log into the remote system, pull the image, and copy it to the box
ssh administrator@10.3.5.99 'dd if=/home/administrator/clone/hboxClone2.gz' | gunzip -1 - | dd of=/dev/mmcblk2
This obviously did not work as I was booted from the drive I was trying to write to and it bricked the device. Thus, I flashed the device using our existing process (requires a peedi box, serial connection, and a Firmware server) and started over.
Questions
So I have a couple low-level questions that I feel must be answered before I can even proceed with developing a solution.
1) Will imaging the mmcblk2 drive be sufficient for creating a replica of the system? It's hard to tell from the fdisk output whether or not everything is stored solely on the mmcblk2 disk.
2) Can partition 1 or 2 be utilized for the imaging? I am thinking I may be able to image just the first partition, write to one of the other two partitions, and swap which I boot from? I read somewhere online about swapping drives and I'm wondering if this same concept can be applied to the partitions. The other thing I'm concerned with is that there may be something written to those partitions (this is the reason I originally tried imaging the entire drive).
How to fix the UUID in Grub after restore from another machine?
Here is what I have in the fstab file:
# UNCONFIGURED FSTAB FOR BASE SYSTEM
#/dev/mmcblk0p2 / auto errors=remount-ro 0 1
#/dev/mmcblk0p1 /boot/uboot auto defaults 0 2
rootfs / auto defaults 1 1
proc /proc proc defaults 0 0
devpts /dev/pts devpts mode=0620,gid=5 0 0
usbdevfs /proc/bus/usb usbdevfs noauto 0 0
tmpfs /var/volatile tmpfs defaults 0 0
tmpfs /media/ram tmpfs defaults 0 0
/hibernateswap none swap sw
0 0
# uncomment this if your device has a SD/MMC/Transflash slot
#/dev/mmcblk0p1 /media/card auto defaults,sync,noauto 0 0
