|
Hi, I'm still struggling to get LVS+Keepalived going.
In brief, this is my setup:
Director 1, master:
Internal interface, eth0, 192.168.x.1
External interfaces, eth1, 16 VIPs bound.
Default route(s) through eth1; different gateways depending on the
subnet.
Director 2, slave:
Internal interface, eth0, 192.168.x.2
External interface, eth1, no IPs bound, down.
Default route through eth0, gateway 192.168.x.1
Real servers:
Internal interface, eth0, 192.168.x.x; default route is 192.168.x.1.
What I need to do is have the slave take over /every/ IP from the
master if anything on the master fails at all. E.g. if .1 fails,
traffic can come in but not get out; if one of the eth1 interfaces
fails, traffic can get out but not come in.
I got the internal interface to fail over by adding another IP to it
on D1 (192.168.x.3), which seems to be necessary for keepalived to
tell D2 that it's come back up. Things stopped working when I tried
to set up one of the external interfaces; do I need a similar setup
for those interfaces (1 real IP per director, 1 shared VIP = 3 total/
subnet), or can keepalived communicate through the internal interfaces?
The other issue I had was that when D2 took over .1, it didn't take
over any other interfaces or routes. It's default gateway being .1
means that no traffic could get out. I'm not sure if this would be
fixed by getting just the VIPs failing over as well, or if I'd need
to have real IPs on each of the subnets for the external interfaces,
with the gateways set up the same as they are on D1.
Thoughts?
--
Ian Eure
Developer,
eNotes.com LLC
|
