Showing results for 
Search instead for 
Did you mean: 


CSS Keepalive to look at state of application not just state of IIS server

I have a pair of CSS11501s load sharing to some servers running the Livelink application. I am able to configure a keepalive to the uri "/" on the servers to check that the IIS server is working but I have not been able to successfully check that the Livelink application is working.

I don't think I can use an HTTP GET and use the hash to check that the service is working because all pages that I am able to test against include the date in the response so I assume the hash will change every day. Am I correct that this is enough to change the hash? I identified that there are some uris I can use from my browser that allow me to bring up a page without the date in them but that these were all over the 64 character uri limit.

By using Wireshark and my browser I have identified a uri that gives me a 302 Redirect when the app is up and 200 OK when the application is down. This is because when the application is up the server tries to redirect me to authenticate but when it is down the IIS server just gives me a message saying it can't connect to the server. I configured an http non-persistent keepalive with an http-rspcode of 302 but that always stays down. When I use llama and do an icp probe to this server/uri from the CSS, however, it always reports 200 OK, even though the application is down. From the wireshark trace of my browser I see that the first response from the server is 302 but I also get a 200 OK from the page I am being redirected to. Is this second response confusing the CSS?

Can you give me any other suggestions of what I can try?


Gilles Dufour
Cisco Employee

Why don't you do a HEAD instead of a GET.

HEAD just ask the server to return the response header and not the entire page.

No hash will be computed.