On 8/4/06, Lamont R. Peterson <peregrine@openbrainstem.net> wrote:
Here is how I would do it: 1. Kickstart install of new box (sets up LVM, includes packages, etc.) 2. Run a script to install all of the needed configuration files for all the services the new server needs (could be run as part of the Kickstart). 3. Copy the data the services will be "serving" to the new machine. There are lots of ways to do step 3. Personally, I like to avoid as much of computational overhead as possible at this stage, so I like NFS mounting and "cp -a". Plain, simple, fast and comprehensive. YMMV.
Thank you for the letter, Lamont! Procedure you described requires a so much time to complete. Tomorrow I end up the following pipeline: 1. Create only one 4Gb plain root partition (no LVM) which holds a entire system and my application. It's used in a development and test environment. The actual stored data is about a 1.5 Gb. I am pretty sure the base system it never fills entire 4 Gb in my case. 2. To build a production server I am bytecopy the root partition to a new box (only a few minutes), boot from it, create LVM in a free space and moves application from root to LVM. Production database now have space to grow. Also I get rid of equal UUID problem. Additionally there is no need for kickstart process is VM environment - I just need to setup OS one time, and then when I need another box I just clone initial setup - it takes a few second. For testing and development VM overhead is ok. -- Stas Senotrusov _______________________________________________ linux-lvm mailing list linux-lvm@redhat.com https://www.redhat.com/mailman/listinfo/linux-lvm read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
