[Gllug] repeat soft noise from hard drive

John Hearns hearnsj at googlemail.com
Wed Dec 10 14:11:35 UTC 2008


2008/12/10 Jose Luis Martinez <jjllmmss at googlemail.com>

>
>
> The dd technique was (is?) widely used with Sun's stuff (as you  may
> or may not know, they normally support a very precise range of disks
> so support tasks are predictable).  I personally installed hundreds of
> machines like that, many moons ago I have to add :-)
>
> There is nothing stopping this in Linux, but normally what happens is
> that, since there is no nanny company dictating disk sizes supported,
> by the time you need to replace a disk the ones available are much
> bigger and you are unlikely to find one identical to the one close to
> collapsing, snd most likely you won't want to anyway.
>

Yes indeed, I agree with what you say. Actually, having a limited set of
hardware is good from a support viewpoint.

Moving on to Linux, using 'dd' would be a great technique to deploy Beowulf
clusters - just boot into RAMdisk,
and dd the disk image from a 'golden image' you keep on the cluster head
node.
The problem is coping with, as you say, a heterogenous set of drive types or
replacement drives. Which is why it is common to use a boot/partition
disk/rsync process.

However, having said that, I recently installed a cluster where the guy I
was working with cloned the nodes using 'dd' over a netcat socket - boot the
nodes in RAMdisk, set up netcat, dd - Bob's your Uncle.
The pair of us even cooked up a scheme to run netcat via multicast - sadly
we didn't have enough time to get this to fly.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/gllug/attachments/20081210/18648d21/attachment.html>
-------------- next part --------------
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug


More information about the GLLUG mailing list