Configuring Sorry Server over multiple 115xx's

[reswob001]

We have a series of 6 11503's over 3 sites. They each have a cacheflow server connected on the backside and the network is configured such that when a client requests a DNS resolution for a web page the 11503's respond with the A record of its IP address. Each 11503 utilizes a http get keep alive to check to see if the web page is available. So if cacheflow#1 behind CSS#1 is down (or can't access the actual web server), CSS#1 automatically rolls over to CSS#2 and it responds with its IP address back to the client because CSS#2 can still retrieve the web page from cacheflow#2 (and so on if cacheflow#2 is down). The problem comes in when the actual web server itself is down. Then none of the cacheflow boxes can access the web server and all of them respond back to the CSS boxes with an error and the none of the CSS boxes reply to the DNS request when the http get keep alive does not get a positive response. The client gets a DNS ERROR in his browser and thinks he has a DNS problem and contacts his local help desk when in reality the problem lies at the actual web server delaying the actual diagnosis of the problem. Is there anyway to create a response more accurately using a sorry server? The example on the Cisco support page only showed one (1) CSS. Thanks.

[Gilles Dufour]

what you can try to do - not test so can't confirm this is a solution.

You can add the redirect services under a content rule as a normal service - not sorry server.

CSS first tries to use regular service before redirect services.

So it should use the local servers as long as they are actives.

Use the sorryserver to point to a webserver somewhere else in your network that displays a message that everything is temporarily down.

Like this you guarantee that the CSS will always answer the DNS request and when everything is down, people are shown a page with a correct error message.

Regards,

Gilles.