|
> Does anyone have director failover in a production setup, when
> directors have failed and changed over automatically and you've
> only found out about it when the failover mechanism notified
> you (rather than the customers notifying you)?
I am in a process to install a web cluster (Apache + PHP + Tomcat4) using
two directors. I'll be able to give you more information once it will be in
production (in a few weeks).
I plan on installing RedHat (lastest stable redhat kernel and lastest stable
IPVS) + Piranha on directors and RedHat (+ hidden patch) on realservers. We
may have some W2K boxes.
We will be using direct routing. And as I'm pretty satisfied by the local
node feature, the active director will also be used as a real server.
Unfortunately, using the inactive director as a RS is always a mess,
whatever network topology you choose.
--
Sébastien Bonnet
Centre de contacts - Experian France
|
