If you need to report an error to an IBM representative, you may be asked to switch on tracing so that information can be written to a file that can be used for locating the error. Use the trace facility only as directed by an IBM Support Center representative, or by your technical support representative.
System performance is affected when tracing is switched on, so use it only when error conditions are occurring.
To turn tracing on, enter:
txtrace on options
The syntax, and lists of the events and components are given in TXTRACE. Other options are also described there.
You can filter the trace by specifying a "mask" which causes the trace to accept or reject each trace record on the basis of its ID. The default is to trace everything.
A mask has four parts separated by periods, for example: 2.2-6.1,3.* where:
You can exclude system errors below a certain severity, and you can specify, if the trace buffer becomes full, whether to keep the first or the last records.
To reproduce the error and write the trace information in binary to a dump file, enter:
txtrace dump dump-filename
After you have written the trace information to a dump file, turn tracing off:
txtrace off
To produce a formatted version of the dump file, enter:
txtrace format dump-filename formatted-filename
You can also write the trace information directly from shared memory to a formatted file while tracing is switched on:
txtrace format > formatted-file