Event triggered trace

You can configure a list of events to activate the trace instead of the running the trace all of the time.

This function can be used to enable a full trace for small parts of the process. It is useful in performance and scalability testing, where tracing adversely impacts the performance results, but you need to collect details about a small part of the process that is malfunctioning.

The trace is configured as normal, but if a list of events is set, the trace is dormant until one of the events occurs. The trace then activates for the relevant unit of work until that unit of work is committed. The trace also activates if an UnexpectedError causes the unit of work to roll back and is retried. That is, the second attempt activates the trace.

Note: Because less trace is captured, you might need to wait longer after the test completes for the trace to be written. Or, you can stop the broker to force a flush of the trace at the end of the test.

The following table shows the format that is used to set the list of events as an entry in the VALUE table.

Table 1. Trace events list in the VALUE table
Category	Key	Value
CORE	PROFILING_EVENTS%	Comma-separated list of events.

Examples of an event triggered trace configuration are shown in the following table.

Table 2. Examples of an event triggered trace configuration
Category	Key	Value
CORE	PROFILING_EVENTS	E_ValPass,E_ValFail
CORE	PROFILING_EVENTS_001	E_AckRecvd,E_OutTxnSent