Because these are RAID boxes I think I would rather do the partitoning manually 
and copy over the network.
The main problem with this is time, because it's very boring at the colo ;). 
Also if I make several partitions on the destination and mount them (/usr /var 
/home etc.) will rsync or cp have a problem with doing it all at one shot,
or am I better off doing each partition separately
 
On Mon, 8 Aug 2005 22:17:28 -0700
David Thomas <[EMAIL PROTECTED]> wrote:

> I'm with Carl on this.  Particularly if you're on a Red Hat-distributed
> kernel (and it's associated modules), you should have very little
> trouble.
> 
> Are the disks the same size too, or are the new ones a tiny bit larger?
> If you want them to be truly identical, by far the easiest thing is to
> follow this procedure:
> 
> Mount both hard drives on the same system.  Boot from a live CD (is
> that what Knoppix does?  I dunno, I know Gentoo will work like a charm).
> 
> Don't mount either disk.  Then run:
> dd if=[src disk] of=[dest disk] bs=1024
> ex:
> dd if=/dev/hda of=/dev/hdb bs=1024
> 
> This will copy every byte of data on the src to the dest.  Don't think
> you even need to set the active / bootable flag on the destination boot
> partition since MBR and all are the same.
> 
> In this scenarion, _EVERYTHING_ is mirrored, even inode access 
> timestamps are preserved, this can be useful for forensics.
> 
> The only limitation is that the partitions will not match the natural 
> cylinder boundary.  If you're highly dependent on disk performance and
> the original disk is carefully laid out according to disk zones and 
> their application (nobody I know does this), you might notice.
> 
> Lots of other approaches, lemme know if you want me to elaborate, you
> might even get when they're root mounted.  :
> *  Create your existing partition sizes using fdisk and use software
>    mirroring to make a duplicate, then break the mirror.  Not sure if
>    you want to do this if you got an old linux.  This is handy if you
>    want to ultimately end up with mirrored disks, also the more disks
>    you need to copy, the duplication time shrinks exponentially.
> *  The traditional thing to do is to mount everything then 'cp -ax' or
>    'rsync -av'. 
> 
> On Mon, Aug 08, 2005 at 09:05:56PM -0700, Carl Lowenstein ([EMAIL PROTECTED]) 
> wrote:
> Subject: Re: cloning servers
> Message ID: <[EMAIL PROTECTED]>
> 
> > On 8/8/05, Todd Walton <[EMAIL PROTECTED]> wrote:
> > > On 8/8/05, Dovber Shapiro <[EMAIL PROTECTED]> wrote:
> > > > I have a few RedHat servers that I want to clone to new machines.
> > > 
> > > Are they identical machines?  Does each have the same exact hardware?
> > > The installed OS assumes certain hardware, and if you transplant it to
> > > a new machine with different hardware, it may go schizo.  I'm not sure
> > > how well udev/kudzu would handle this.
> > 
> > I think you will find that Linux is much more tolerant of hardware
> > differences than some other unmentionable x86-biased OS's.  I have
> > transplanted a disk from Pentium II to AMD K6-3 with no trouble,
> > vintage RedHat 7.2.  Also more recently from the same K6-3 to Athlon
> > with no trouble, vintage Fedora 3.  Have not tried K6-3 to Pentium 4.
> > 
> > Much of the hardware discovery process is done at boot time, rather
> > than being built in when the software is first installed.
> > 
> >     carl
> > -- 
> >     carl lowenstein         marine physical lab     u.c. san diego
> >                                                  [EMAIL PROTECTED]
> > 
> > 
> > -- 
> > [email protected]
> > http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
> 
> 
> -- 
> [email protected]
> http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to