Re: FreeBSD semi-automatic deployment
phn_at_icke-reklam.ipsec.nu
Date: 10/28/03
- Next message: Lee Harr: "Re: 19.5 hours to build KDE... kernel at fault?"
- Previous message: Robert Downes: "19.5 hours to build KDE... kernel at fault?"
- In reply to: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Next in thread: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Reply: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 28 Oct 2003 21:31:02 +0000 (UTC)
Tim Daneliuk <tundra@tundraware.com> wrote:
> phn@icke-reklam.ipsec.nu wrote:
> <SNIP>
>>># After you have a working system up and running that is custom suited,
>>>what is the best way to "save" this setup and replicate it in future
>>>installations.
>>
>>
>> Ghost will du just fine. Replacing unused diskspace with constants
>> will make it possible for ghost to only copy the used space.
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Can you say a bit more about what you mean here. I have considered
> using Ghost as a backup for FreeBSD partititon "images", but could
> not figure out a way to avoid backing up the blank space...
Shure. It's simple.
Consider the task Ghost phases: Reading all used blocks, compressing
them and write them to an "archive". The problem ghost has
with FFS is that it cannot interprete the structure and
can therefore not skip the unused disk space.
What can be done is a simple trick; make shure all unused
space is _very compressible_, zero-filling would do.
How then zero-fill all unused blocks ?
Thanks to unix architecture and well-done design the answer
is simple. A one-liner !!
dd if=/dev/zero of=<filesystem/dummy> ; rm <filesystem/dummy>
( repeat for all filesystems, don't bother with swap partition)
The above will create a file, filled with null-bytes until
the filesystem is full, then removal of the same file will
remove the inode and directory information, but all datablocks
are still filled with zeroes.
The remaining problem with ghost is that you will have to fill
either a complete disk with an image of another complete disk, or
you can fill a partition with a "partition-image". Both has it's
pro's and con's, filling a disk will create boot-blocks and stuff,
but your disks will all be equal size as your source. Partition
in this context is what FreeBSD-er calls "slice" and other people
might refer to as "DOS-partitions". Ghost does not understands
FreeBSD "partitions", it will only use "slices".
Using a partition will fill a partition of the same size as
the sorce disk, leaving boot-blocks and other areas unaffected.
This is probably the most general way, but you will gave to
create boot-information.( a bootable diskette could be
created to do this)
A final advice, if using dissimular disks, partitions that
does not start at exact same position will make the FFS system
unavailable. There seems to be references to absolute
disk-addresses in the superblock ( not relative to partition
start but to disk start). The only "safe" partition seems to be
the first ( starting at block 64), this seems to work
across various disks, and seems to work for FreeBSD and OpenBSD.
( my last use was FreeBSD 3.x systems, but i _think_ there is
simular behaviour with later BSD's)
> TIA,
> ----------------------------------------------------------------------------
> Tim Daneliuk tundra@tundraware.com
> PGP Key: http://www.tundraware.com/PGP/
--
Peter Håkanson
IPSec Sverige ( At Gothenburg Riverside )
Sorry about my e-mail address, but i'm trying to keep spam out,
remove "icke-reklam" if you feel for mailing me. Thanx.
- Next message: Lee Harr: "Re: 19.5 hours to build KDE... kernel at fault?"
- Previous message: Robert Downes: "19.5 hours to build KDE... kernel at fault?"
- In reply to: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Next in thread: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Reply: Tim Daneliuk: "Re: FreeBSD semi-automatic deployment"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|