[Linux-cluster] HEARTBEAT and WAN

[Michael Will]

A floating ip address makes only sense if it is the same on both systems since it is used to hide the different identities of the servers implementing the service, making the failover transparent to the clients that can just reconnect to the same ip address after a failure.

If you do need that feature you could achieve that with a virtual private network vpn that layers over both subnets but then your clients need to be on that private network as well.

You can do without that that if your clients can be programmed to try both ip addresses before giving up.

Michael Will / SE Technical Lead
www.penguincomputing.com

 -----Original Message-----
From: 	lionel.hay [mailto:lionel.hay at agriculture.gouv.fr]
Sent:	Tue Jan 30 03:27:53 2007
To:	linux-cluster at redhat.com
Subject:	[Linux-cluster] HEARTBEAT and WAN

I'am looking for some exemples of files (ha.ch, haresources ..) to 
create a cluster between two LAN (over a WAN)

- How do I find the rights values for ha.cf ?
Do I use bcast,ucast, mcast ?
- Each machine have it's own IP and SUBMASK
FIRST 10.64.1.2/255.255.128.0 on eth0 with 10.64.0.254 for routing
SECOND 10.64.128.76/255.255.252.0 on eth0 with 10.64.128.254 for routing
My tracert command, give me 5 devices between my two LAN (including my 
own devices)
How do I choose the floating IP ? (on second subnet where my primary is, 
or second subnet where is my second)
How can I prevent false takeover in case of overloading ?

Thanks for your help



-- 
Lionel HAY
DDAF Pyrenees Atlantiques
64000 PAU
FRANCE
Tel : 0559021254

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20070130/ff19c454/attachment.htm>