Multi server (master and fallback) setup

[Christ]

Hello,

I started using VHCS.
But ever since VHCS "stopped" developing and ISPCP came out i started using ISPCP, wich is already a decent time.

But now i think its time to upgrade my server a bit.

What i want to do is have a primary server with everything on it.

But if that primary server fails i want everything to fall back to the secondary server.

But now comes the problem, i have never used a multi server system, so I was kinda hoping someone here could give me some pointers or tips.

What i want is have the primary server wich is constantly updating its changes (at least everything that concerns ISPCP (mail, web, settings, and so on)) to the secondary server.

And if for some reason the primary fails, i want it to fall back to the secondary. And if/when the primary comes back online, it syncs to the secondary and everything becomes back to normal.

I hope someone here knows a bit (or a lot) about these kind of setups and is able to help me get started.

My current system is a standard debian installation with ISPCP and some minor modifications, nothing special.

[motokochan]

That sounds like a high-availability cluster, which is outside the realm of this project. There is a lot of good documentation out there, and even an nice book published by No Starch Press (out of print, but the publisher sells an electronic edition).

[Christ]

It doen't have to be integrated in ISPCP i think.
As far as ISPCP is concernd it doesn't even have to know this is happing i think.

I always thought a HA claster was a bit different, but i could (and probebly am) wrong.
I'll look in to that, maybee it'll be my solution.

I just want te keep everything working if there is another power failure over here.
For some reason it's pretty difficult to keep the power going in a DC...

[kilburn]

I would go the DRBD way...

[motokochan]

DRBD or Linux-HA should work fine, but this is certainly out of scope for this project.