RE: [cobalt-users] redundancy how too....

[Rodolfo Paiz <rpaiz@xxxxxxxxxxxxxx>]

> we actually have our first web host client

Congratulations!

> and they have asked about the second nameserver info...

Uh-oh... just kidding.

Second nameserver is really easy. For the first couple of days, just get
your RaQ to listen on a second IP address and point ns2.yourdomain.com
to it; and voilà! you have a second nameserver.

However, you don't want to do this permanently. If no DNS at all is
available, then browsers will get a "Host does not exist" error, which
makes you look really bad to customers. If DNS is available and
webserver is down, then worst case you get "Page not available," which
is a lot less embarrassing.

The best way is to setup a second server. I use a Qube for my secondary;
works like a charm. Once you have that server working, make sure it uses
the ns2.yourdomain.com name and IP address and the customer won't even
know you've made the change.

> how do we set up a second redundant web server? how do we get
> the files to be always the same on each of the web servers?

That's another story. Two levels here:

Set up a second web server (on your secondary DNS machine, maybe?) and
synchronize it using any method you like (I suggest rsync under Linux,
or the mirror script, but the way to go depends on your level of
knowledge). Now you have a second "redundant" web server, since
redundant means something like "having more than necessary in order to
cover for potential failures." So just owning two cars gives you
redundancy, as does the spare tire in your car.

You can even share the load between them if you have things worked out
well, by assigning two IP addresses to the name www.yourdomain.com. DNS
will then hand out a different address to each new client that comes in,
roughly giving half the load to each server. Now you have a redundant
server.

What you actually want, however, is fault-tolerance, meaning that you
don't stop serving pages just because one component failed (like the
entire server). Making your second server take over when (not if, when)
the first one fails can entail:

* Clustering software/hardware
* Load-balancing software/hardware
* Short DNS TTL times and quick manual/automatic modification
  of DNS info if server dies
* Automatic/manual changing of the second server's IP address
  when the first goes down

There are a *lot* of things you can do. All of them take some learning,
and I can't give you more information because (a) you need to
investigate it, (b) I'm still investigating, and (c) I still don't know
much.

This is more a topic than a question, and you'll have to do some
research. Let's see what other list-members come up with...

--
Rodolfo J. Paiz
rpaiz@xxxxxxxxxxxxxx <mailto:rpaiz@xxxxxxxxxxxxxx>