|
Greetings again,
I'm attempting to do the Ultramonkey High Capacity High Availability
topology described at
http://www.ultramonkey.org/2.0.1/topologies/hc-ha-lb-eg.html.
I have a simple (I thought) test setup with two directors and three real
servers (all Redhat 8) serving a one line index.html file (via apache
1.3.x).
The HA function works fine, and now I'm just working on the primary HA
box's ldirectord settings. It seems to (sort of) work -- I have been
able to hit all three real servers, but it doesn't look like real round
robin.
The problems I have are:
1. At my first attempt to hit the virtual server, I always get a "server
not found" error in my browser. As soon as I reload, I get one of the
three real servers' index.html pages as expected.
2. I tried to hit the virtual address from 5 different machines using
lynx (within 30 seconds of each other) and they all landed on the same
real server. I assume there's either a stickyness or timeout issue I
have to configure.
Here is the ldirectord.cf file.
-----------------------------------------------------
# Global Directives
checktimeout=10
checkinterval=2
fallback = 127.0.0.1:80
logfile = "/var/log/ldirectord.log"
logfile = "local0"
quiescent = yes
#
virtual=192.168.225.160:80
real=192.168.225.154:80 gate
real=192.168.225.152:80 gate
real=192.168.225.151:80 gate
service=http
request="index.html"
receive="I AM"
scheduler=rr
protocol=tcp
checktype=negotiate
-----------------------------------------------------
Thanks in advance for any suggestions!
Andy
|
