> Additionally, as I see it, there seem to be 3 ways to go right
> now (not in any
> order of preference):
>
> 1. Use Pirahna
> 2. Use UltraMonkey
> 3. Cobble the pieces together yourself
>
> I'd like to be able to fully understand what's needed for option
> 3. It is
> forseeable that there would be people who want to do things the
> hard way, and
> I'd like to be able to fully document this option.
Cobbled setups:
1. Mon, Heartbeat, ipvsadm
(to install mon, you have to find and install several other
files: fping, Convert-BER, Period, Time-HiRes, Mon, mon)
The last two are not the same. one is the daemon, the other
is a collection of clients to connect to the daemon.
Also note that to compile mon on a linux box you have to
comment out line 222. Mention that in you howto, it will
probably save someone a lot of time.
2. ldirectord, heartbeat, ipvsadm
Heartbeat has Fake incorporated in it, so if you have Heartbeat,
you don't need fake.
mon is very flexible, and easy to customize. In addition to watching
the RealServers, you can use mon to verify that the nic's are working on
the director, and ping the public ip address of the cluster, and even
alert you if the web pages are too slow.
ldirectord is a bash script used to monitor the RealServers.
for examples, look at
http://www.linuxvirtualserver.org/HighAvailability.html
hope that helped.
Ian
|