On Tue, 2005-05-31 at 23:01, Ashley M. Kirchner wrote:
I already have backups on the MX and web servers, however those
do
me no good when the main users server goes down. THAT's the one I need
to figure out how to build a redundant setup for. Users can't get to
their accounts/files/e-mail if that main server goes down. I don't know
if there's any way of building some kind of quiet failover system that
would allow users to continue using their shell and get to their files
and e-mail when one server goes down (and the other takes up the
slack.) Somehow when the main server comes back up some things would
have to switch back and all of that.
You'll find some projects here:
http://www.linux-ha.org/ about
doing real-time redundancy. My opinion is that for most things
the additional complexity is about as likely to cause failure
as to avoid it, expecially for a multipurpose server.
A simple-minded approach is to use servers with swappable drive
carriers, mirror all the partitions, and keep a spare server
around. The most likely failure would be a disk drive and
RAID1 lets you continue without even a slowdown. If you have
hot swap drives you can replace and rebuild without stopping
too. In the less likely event of a motherboard component failure
you can pull the drives, move to your spare box, reboot and
be back with minimal downtime. That's not particularly elegant
but it's pretty foolproof and simple enough that you can talk
someone else through doing it over the phone. You do still need
backups to cover the case where a software or operator error
destroys the data on the live disks.
--
Les Mikesell
lesmikesell(a)gmail.com