I've been toying with response based transactions and having some issues. I have a use case of an application that is running on websphere portal with encoded url's. In this use case they are using the Websphere Portal Field pack that will extract the portlets name from the transaction and then include the "nice name" as a part of the HTTP response. There is absolutely nothing in the request portion of the transaction that makes any sense to organize as a business transaction, but the response parameters are good, you can tell what the end user transaction was supposed to be.
The documentation around creating response based transactions is weak. I can find no where in the bookshelf (or here) where a very good explanation and example is provided. Hence, why I'm reaching out to the community.
In my testing, I have a single transaction.
1. I did a transaction discovery looking for the HTTP response parameter. No problem, found what I as looking for.
2. Promoted the transaction for monitoring, but keep in mind, outside of the http response parameter, there's nothing else to truly delineate this request transaction from any other request.
3. Extended the request based transaction using the link in the business service list using the "Create Response-Based Transaction"
4. Setup the matching for the response based transaction as expected.
Here's the problem. The request based portion of the transaction must be enabled, even though it has nothing redeeming to match upon. This results in that transaction picking up defects(partials, missing responses, etc) for virtually any transaction. The response based transaction that i created appears to work just fine, picking up on the response headers like i expect. But, i don't care about the request side of it...but must keep it enabled to get the response...sort of a catch 22.
What has been other peoples experiences with response based transaction monitoring? Has anyone come across any good documentation that fully explains how to use appropriately?