Creating Mirror Back-up Hot Swap Server

[jfb130]

I am currently running Debian on a system with 3 scsi hard drives with Raid and have an identical back up server. I want to schedule a daily "mirror" copy of the server operating system and data so that if the primary server drops I can move the network plug to the next server and be back in business. Can someone let me know what would be the best solutions for this? I am new to linux so this may be an obvious question so if it is I applogize but I have been searching through the forums and have not found an adeqate solution.

[wszachau]

That's probably because there is no easy solution to this, especially if you are new to Linux.
If your goal is to achieve a situation where you can simply move the cable from one box to the next, then this is actually quite hard to do. Classical backup systems, e.g. bacula or amanda, won't work for this. For starters, you will need to have a network connection between the two servers that is independent from the main access to them, i.e. a private little network which will be used to transfer the data. The secondary server has to be on the same primary IP address as the primary server, otherwise the "change the cable" thing won't work. In order to get a complete mirror copy, you will need to have most of your applications and data on partitions that are separate from the partition containing the software needed for the transfer. This ***must*** include the /etc/ folder, since that is where most of the configuration is stored. I would suggest to use a separate /boot partition for this purpose. You then only need to keep that partition manually synchronized to the master.
Then you need to setup a transfer of entire partitions using dd (read the man page for that). In order to use dd, the partitions have to be unmounted (that's where the fun starts). And, of course, while the partitions are unmounted on the primary server, it won't do very much. Which is pretty much why this is so hard, and hardly anybody ever tries it.
You are much better off using a normal backup solution and focusing on file based replication to the slave. That probably won't give you a "just move the cable" kind of standby, but it is a lot more manageable and your primary server doesn't have any downtime while the backups are taken.