Starting API Tracing Activity

Once you have trace data sets available for use, you can invoke the ATBTRACE REXX exec to start tracing activity for a particular TP. To collect all the pertinent trace data for this TP, you have to know more than just the TP name. Your installation's APPC configuration, and your TP's partners and processing, determine what values you supply on the ATBTRACE exec, and how many times, and where, you might have to issue ATBTRACE START requests. At a minimum, you must supply an LU name and TP name to uniquely identify the conversations to be traced. This LU/TP combination is required because APPC/MVS allows multiple TP profile files, each of which might define a TP with the same name, for a particular z/OS system.

Before you begin issuing ATBTRACE START requests to activate tracing, you need to answer the following questions to decide what values to supply on those START requests:
  • Does your TP initiate conversations with partners, or do the partners initiate the conversations?
  • Through which APPC/MVS LUs do inbound Allocate requests arrive for the TP? How are the LUs configured; for example, are they members of a VTAM generic resource group?
  • Where do the TPs partners reside: On the same z/OS system? On a different z/OS system? On an operating system other than MVS?
Once you have answered these questions, continue to the next topics, which cover:
  • How APPC/MVS handles an ATBTRACE START request, which will help you understand how many START requests you need to issue, and with what values
  • Suggestions that might increase programmer productivity and efficient use of the API trace facility.
Parent topic: Diagnosing TP Conversation Errors with the API Trace Facility