Hi Everybody! Ok, we have two windows 2003 servers here at the office. The windows administrator was able to set the two machines up as a cluster. So they are constantly replicating and if one goes off line, all network traffic will automatically go to the other machine.

I was wondering if there was a way to do something like that with Linux. I have a CentOS 4.1 web/e-mail server (Apache, MySQL, Sendmail). I just bought a second machine that has identical hardware specs to my server.

Could I possibly setup the second machine to be a live mirror of the web/mysql/mail server? Is there a feature or program in CentOS that would let me set something like this up.

I'm not a linux n00b, but I haven't got into any of this advanced networking before. If I can't do something like the Windows clustering I was just going to use RSYNC to sync the backup machine every couple of hours.

There is certainly clustering for Linux. The question is what exactly you want to do.

There are commercial products such as Veritas Cluster Services that do full clustering (from a brief search looks like only for RedHat and Suse on Linux - I've used it on both HP-UX and Solaris).

Oracle has a product called Real Application Cluster (known as Oracle RAC) that has built in clustering for databases (including either ASM for raw storage or ocfs for cooked storage). We are currently using that on RedHat AS 3.

Also there is the idea of "linux server farms" in which you can have multiple servers share the same storage and do load balancing between them. I'd worked with this at a prior job but we had just gottent into it. We were using NAS filers for storage. A friend of mine you to go on about how google (or maybe yahoo) had thousands of nodes in a farm so that if one went down they just put in another and never suffered any down time.