|
On Tue, 18 May 1999, Peter Kese wrote:
> The Virtual Server patch development for the 2.2 kernel is
> continuing.
Well, we gave the new patch a shot tonight and I still can't seem to get
it working. I'm using my desk machine as a load balancer that tries to
connect to one of our webservers.
This is my current configuration:
On amadeus (running 2.2.9+vs0.2):
ipchains -F
ipchains -M -S 30 30 30
ifconfig eth0:1 209.183.99.145 netmask 255.255.255.255 broadcast \
209.183.99.145 up
route add -host 209.183.99.145 dev eth0:1
./ipvsadm -C
./ipvsadm -A -t 209.183.99.145:80 -s rr
./ipvsadm -a -t 209.183.99.145:80 -r 209.183.99.15:80 -g
On galileo (running 2.0.36):
ifconfig lo:0 209.183.99.145 netmask 255.255.255.255 broadcast\
209.183.99.145 up
route add -host 209.183.99.145 dev lo:0
When I telnet to 209.183.99.145 80 from dillinger, I just get a
connection refused. When I do a `ipchains -M -L` on amadeus, it shows
masqueraded connections that appear to be going from galileo back to
dillinger. I was under the impression that the return path using direct
routing would go straight from the real server to the client w/o returning
to the load balancer at all.
[root@amadeus /usr/src/ipvs-0.2-2.2.9] ipchains -M -L
IP masquerading entries
prot expire source destination ports
TCP 00:21.56 galileo.aureate.com dillinger.aureate.com http (80) -> 1497
TCP 00:20.97 galileo.aureate.com dillinger.aureate.com http (80) -> 1496
TCP 00:19.92 galileo.aureate.com dillinger.aureate.com http (80) -> 1495
Did I miss a step when I set the VS up? If I change -g to -m, then I get
a connection but it just sits there accepting input. The connection does
not appear in galileo's logs.
- cls
|
