|
Hiya. I'm trying to get a high availability cluster working as described
in http://www.linuxvirtualserver.org/HighAvailability.html using the
ldirectord+heartbeat option. It works great for a single virtual ip
address, but I want to provide load balancing and failover services for
one than one virtual ip address. I'm having difficulty figuring out what
the best route to follow when scaling would be. This configuration does
not work (where 192.168.1.x is the faked network and 10.0.0.0 is the
internal network)
/etc/ha.d/haresources:
director IPaddr::192.168.1.1 ldirectord::www
director IPaddr::192.168.1.2 ldirectord::www
/etc/ha.d/conf/www:
timeout = 10
checkinterval = 10
virtual = 192.168.1.1:80
service = http
protocol = tcp
scheduler = rr
request = "/testpage.txt"
receive = "test page"
real = 10.0.0.1:80 gate 5
real = 10.0.0.2:80 gate 5
virtual = 192.168.1.2:80
service = http
protocol = tcp
scheduler = rr
request = "/testpage.txt"
receive = "test page"
real = 10.0.0.1:80 gate 5
real = 10.0.0.2:80 gate 5
heartbeat is fine with this but ldirectord doesn't like it at all. i
suppose i could try this:
/etc/ha.d/haresources:
director IPaddr::192.168.1.1 ldirectord::www1
director IPaddr::192.168.1.2 ldirectord::www2
/etc/ha.d/conf/www1:
timeout = 10
checkinterval = 10
virtual = 192.168.1.1:80
service = http
protocol = tcp
scheduler = rr
request = "/testpage.txt"
receive = "test page"
real = 10.0.0.1:80 gate 5
real = 10.0.0.2:80 gate 5
(same idea for www2)
would this be a good way to tackle this? is there a better suggested
solution?
- donald
|
