is this possible on CSS??? someone

[followurself]

can someone pls help

no server load balancing within farm but dependency

farm 1

e.g. server A- probe -port 80

      server B-probe-port80/636

request from user should only go to server A  but if port 80 on any server fails then that sfarm should be out of loadbalancer.

farm 2

e.g server C -probe -port 80

     server D - probe -port 80/636

farm 2 to be used as backup serverfarm

also is it possible to load balance traffic between farms. so the traffic gets load balanced between server A and server C but the dependency of ports above remains between farms

thanks

pls reply

[paul.matthews · Accepted solution]

I am taking a guess here - Sever B is some backend that

A uses, and you want to test that A has a backend?

If that is the case, the way I would prefer to do that would be to get a URL on A that would test the application to the backend. If that is not practical, you will be into writing keepalive scripts - https://www.cisco.com/en/US/docs/app_ntwk_services/data_center_app_services/css11500series/v8.20/configuration/administration/guide/Scripts.html - look at the examples, http list and ping list

[followurself]

thanks Paul for the reply

if i use the inbuilt http script, do i have to modify the script?

for e.g

i want to poll both the servers on http port, where within the script i can mention the ip address of both servers. because witin the content rule i want to use server A as all user request should go only to server A. but if sever B port 80 goes down or server A port 80 goes down then server A should be out of service

[busterswt]

I recently had a customer implement something very similar to what you're wanting to do. They had server A perform some kind of check on server B, and modify the contents of it's healthcheck page accordingly. If server B was accessible on port 80, server A changed it's healthcheck file to read 'UP'. If server B was not accessible, server A changed the contents to read 'DOWN'.

On the CSS we implemented a service similar to the following:

service 10.1.1.1-serverA

ip address 10.1.1.1

keepalive type http

keepalive port 80 (default)

keepalive method get

keepalive uri '/healthcheck.asp'

(optional) keepalive hash xxxxxxxxxxxxxx

active

To sum it up, server A was responsible for determining whether or not server B was accessible and updated its own healthcheck page accordingly. Server A could be up and responding on port 80, but because it failed it's keepalive (hash mismatch) the service was taken out of rotation on the LB. If server A were to really become inaccessible it would also keepalive due to port 80 not responding.

I don't really know how to implement this in a custom keepalive script, but you'd likely introduce unnecessary cpu usage to process that script continuously.

James