Determining Data Set Characteristics

APPC/MVS requires API trace data sets to be pre-allocated and sequential, with a block size of 4096 or larger. (The block size requirement applies only if you specify the BLKSIZE operand.) If you already have data sets that meet those requirements, you might be able to use them without having to allocate others. Whether you are using existing data sets, or setting up new data sets, keep the following points in mind:
  • APPC/MVS uses only primary space for trace data sets. If tracing activity is not stopped before APPC/MVS reaches the end of the primary space, and the data set resides on DASD, APPC/MVS begins wrapping the trace data by returning to the beginning of the data set, and overwriting older trace entries with more recent entries. APPC/MVS does not indicate how many times wrapping occurs.
  • APPC/MVS uses variable blocked record format, regardless of the format specified during allocation, but does use the block size, if you specify the BLKSIZE operand. Depending on the specified block size, and I/O contention, APPC/MVS might collect trace data faster than it can write that data to the data set. When this backlog of collected data becomes too large, APPC/MVS temporarily stops collecting trace data until it can finish writing the backlog to the data set. APPC/MVS indicates any suspension of tracing activity through an entry in the trace data.
Both of these situations result in the loss of trace data, which might hamper your ability to diagnose problems with the TPs being traced. They also might make reading the trace data more difficult. Fortunately, several techniques can help you reduce the possibility of losing trace data in either situation, and reduce the relative complexity of interpreting trace data.