Sector Issues on Dump/Clone of Embedded FW

[daemonbak]

Hi all,

Have an issue here. I have 2 drives that are marketed as being the same size 32G. However, the source drive is slightly larger than the target drive for the dump.

Is there a way to make a change so that I can dump sd0 to sd1?

sd0 (system drive)
sd1 (target drive)

Process so far:

disklabel sd0 > /tmp/label.0
disklabel sd0 > /tmp/label.1
fdisk -i /dev/rsd1c
disklabel -Rr /dev/rsd1c /tmp/label.0 (ERROR - unknown option -r???)
removed -r from command
disklabel -R /dev/rsd1c /tmp/label.0 (ERROR - disklabel: partition k: partition extends past end of unit)

I took the c sector size from label.1 and replaced label.0 c sector size


disklabel -e sd0

/dev/rsd0c:
type: SCSI
disk: SCSI disk
label: Cruzer
duid: a32277e951b9534c
flags:
bytes/sector: 512
sectors/track: 32
tracks/cylinder: 255
sectors/cylinder: 8160
cylinders: 7663
total sectors: 62530624
boundstart: 32
boundend: 62530080
drivedata: 0

16 partitions:
#                size           offset  fstype [fsize bsize  cpg]
  a:          4202368               32  4.2BSD   2048 16384    1 # /
  b:          8608800          4202400    swap                   # none
  c:         62530624                0  unused
  d:          4202400         12811200  4.2BSD   2048 16384    1 # /tmp
  e:          8396640         17013600  4.2BSD   2048 16384    1 # /var
  f:          8396640         25410240  4.2BSD   2048 16384    1 # /usr
  g:          2105280         33806880  4.2BSD   2048 16384    1 # /usr/X11R6
  h:          6299520         35912160  4.2BSD   2048 16384    1 # /usr/local
  i:          8396640         42211680  4.2BSD   2048 16384    1 # /usr/src
  j:          6299520         50608320  4.2BSD   2048 16384    1 # /usr/obj
  k:          5622240         56907840  4.2BSD   2048 16384    1 # /home

# Notes:
# Up to 16 partitions are valid, named from 'a' to 'p'.  Partition 'a' is
# your root filesystem, 'b' is your swap, and 'c' should cover your whole
# disk. Any other partition is free for any use.  'size' and 'offset' are
# in 512-byte blocks. fstype should be '4.2BSD', 'swap', or 'none' or some
/tmp/EdDk.akDxsfJLpVV: unmodified: line 1

disklabel -e sd1

# /dev/rsd1c:
type: SCSI
disk: SCSI disk
label: Cruzer
duid: d8fd3d7968f58007
flags:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 255
sectors/cylinder: 16065
cylinders: 3800
total sectors: 61056064
boundstart: 64
boundend: 61047000
drivedata: 0

16 partitions:
#                size           offset  fstype [fsize bsize  cpg]
  c:         61056064                0  4.2BSD   2048 16384    1

# Notes:
# Up to 16 partitions are valid, named from 'a' to 'p'.  Partition 'a' is
# your root filesystem, 'b' is your swap, and 'c' should cover your whole
# disk. Any other partition is free for any use.  'size' and 'offset' are
# in 512-byte blocks. fstype should be '4.2BSD', 'swap', or 'none' or some
# other values.  fsize/bsize/cpg should typically be '2048 16384 16' for a
# 4.2BSD filesystem (or '512 4096 16' except on alpha, sun4, ...)

Any solutions where I can dump this on the new drive as a emergency clone. I'd like to have a fallback that i can pop in when this drive dies.

Thanks

[daemonbak]

I ended up using disklabel -E to manually partition the disk. All sizes are the same as the source drive... with a little stolen from home to adjust for the slightly smaller size.

I did notice that the offset default was 64 on the new drive, but the original source drive was 32 offset. I left it 64 to be safe.

Now what would be the best way to backup:
src drive > target drive?

Would this be dump | restore one partition at a time and then use dd for partition k (/home)

Thanks

New disklabel -e sd1

 /dev/rsd1c:
type: SCSI
disk: SCSI disk
label: Cruzer
duid: 4ba99503bda02123
flags:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 255
sectors/cylinder: 16065
cylinders: 3800
total sectors: 61056064
boundstart: 64
boundend: 61047000
drivedata: 0

16 partitions:
#                size           offset  fstype [fsize bsize  cpg]
  a:          4202368               64  4.2BSD   2048 16384    1
  b:          8608800          4202432    swap
  c:         61056064                0  unused
  d:          4202400         12811232  4.2BSD   2048 16384    1
  e:          8396640         17013632  4.2BSD   2048 16384    1
  f:          8396640         25410272  4.2BSD   2048 16384    1
  g:          2105280         33806912  4.2BSD   2048 16384    1
  h:          6299520         35912192  4.2BSD   2048 16384    1
  i:          8396640         42211712  4.2BSD   2048 16384    1
  j:          6299520         50608352  4.2BSD   2048 16384    1
  k:          4139104         56907872  4.2BSD   2048 16384    1

# Notes:
# Up to 16 partitions are valid, named from 'a' to 'p'.  Partition 'a' is
# your root filesystem, 'b' is your swap, and 'c' should cover your whole
# disk. Any other partition is free for any use.  'size' and 'offset' are
# in 512-byte blocks. fstype should be '4.2BSD', 'swap', or 'none' or some
# other values.  fsize/bsize/cpg should typically be '2048 16384 16' for a
# 4.2BSD filesystem (or '512 4096 16' except on alpha, sun4, ...)

[TronDD]

dump | restore works pretty well. That's my primary local backup mechanism. And I have had to do a recovery, both of individual files as well as whole partitions.

I would have taken the extra space out of /tmp or some other partition you wouldn't actually backup/restore. Pretty much anything outside of /home /etc /var or /root.

[daemonbak]

I am doing the following now:

newfs /dev/sd1a
mount /dev/sd1a /mnt
(dump -0f - / )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1b

newfs /dev/sd1d
mount /dev/sd1d /mnt
(dump -0f - /tmp )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1e
mount /dev/sd1e /mnt
(dump -0f - /var )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1f
mount /dev/sd1f /mnt
(dump -0f - /usr )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1g
mount /dev/sd1g /mnt
(dump -0f - /usr/X11R6 )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1h
mount /dev/sd1h /mnt
(dump -0f - /usr/local )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1i
mount /dev/sd1i /mnt
(dump -0f - /usr/src )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1j
mount /dev/sd1j /mnt
(dump -0f - /usr/obj )|(cd /mnt/; restore -rf -)
umount /mnt;

newfs /dev/sd1k
mount /dev/sd1k /mnt
(dump -0f - /home )|(cd /mnt/; restore -rf -)
umount /mnt;

The issue I am having now is that / has the whole filesystem, /usr has all the data from partitions g + h + i + j as well as what it should. So dumping / dumps the entire disk to that partition, not how it is setup in the original disk and /usr partition has the following in it as well (it should have just f not the rest):

f: /usr
g: /usr/X11R6
h: /usr/local
i: /usr/src
j: /usr/obj

[TronDD]

Hmm... Are you sure it's the data of the sub-partitions and not just the mount points?

[daemonbak]

ls -alR shows it is the contents as well. Solved by:
mount / to /mnt && d /mnt && rm -rf tmp var usr...

But there must be a more efficient way