SAP Note 16665 - Method for troubleshooting, trace level

Symptom:

Errors difficult to reproduce, trace method

Cause and prerequisites

Customer/user assistance required

Solution

We require additional information to track down the described error.

1. How often does the error occur? Are there times when the error occurs particularly often?

2. Does the error occur often in a specific location, for example, in connection with a specific R/3 server, frontend, or user?

3. Are there entries in the SysLog (SM21) or in trace files (dev_... in the work directory of the R/3 server, in stderr, stdout)?

4. Can the error be reproduced, or are there circumstances under which the error occurs particularly often?

5. Make sure that each occurrence of the error is recorded, together with the time, user name, R/3 server, and preceeding action, in a central log.

If certain regularities are observed, you can increase the trace level (rdisp/TRACE=2 in the instance profile) and record internal operations in the trace files.
When the trace level is increased, large quantities of data may be written to the trace files. Therefore, you must monitor the file system in question regularly to make sure sufficient free space is available. Let us know when you have collected the above information.
Your knowledge of the UNIX and R/3 systems permitting and if you feel comfortable in doing so, you can accelerate the troubleshooting process by increasing or decreasing the trace level. Otherwise, we will log on to your system ourselves.
Background information about the trace level:
The trace level is set with parameter rdisp/TRACE=. It is effective for all processes after the next R/3 system start.
The trace level for work processes can be set in the work process overview SM50 and the trace level for users can be set in the user overview. In Transaction SM50 select the process desired and choose Process -> Trace -> Components. Here you can set the trace level and the components to be traced. If the files become too large without the error being logged, you can start the trace again in the same menu path with 'Reset files'.
The default trace level is 1: only errors are logged.
Trace level, recorded in trace file:
Level 0: Nothing recorded; trace file is closed
Level 1: Only errors are logged
Level 2: Internal activities are logged
Level 3: Internal activities and data are logged
For individual processes, the trace level can be changed at runtime using the UNIX command:
kill -USR2 ;increases trace level by 1
kill -USR1 ;decreases trace level by 1
Trace files with level 0 can be deleted, thus gaining space in the work directory. Deleting trace files with a level other than 0 is useless.