There seems to be no mutually agreed upon way for configuring and deploying Linux on a massive scale.

I have to configure files beyond kickstart, add software (i.e rpm's, tar archives) and deploy this built to many machines.

So do you do this, or would do this?

I'd be curious to hear about different solutions.

Thanks,
Chris

I guess it depends a lot on what you want to do with the systems... are they all to run identical images, say for a cluster computing project, or are they all to have different softwares installed and used as machines for desktop use?

the quickest way i know is create 1 working system.. tar up the files (not the root '/' directory, remount your root partition to another fs like /mnt/root) and copy it to another fs. you can use a livecd to transfer the files within a network. another option is that you can also use 'dd' to create an image of your partition. basically it's like this:
You can add more options to dd like 'bs=2M' or 'bs=8M' to speed up the copy operation.

Whether you use gz or bz2, it's just optional. I prefer gzip than bzip2 when it comes to larger files.

As more and more machines get the copy of the system, more and more copies can be made at a time .. for example it starts at 1 then became 2 then became 4 .. you know what i mean..

The configuration part is different though. Perhaps before you copy a working system.. Make sure the common configurations are already setup.

There is a specialised piece of software that will do this for you called SystemImager. Ideally the boxes should be pretty much identical in terms of hardware for this to work effectively.