Troubleshooting Guide

Trap Files

Under severe error conditions, the DB2 system issues a signal (for UNIX-based platforms) or exception (for Intel-based platforms) to itself. Other names for this occurrence are segment violations or traps, depending on the platform.

All signals or exceptions initiated by DB2 are reported in a trap file, which contains a function flow of the last steps that were executed before the system stopped. Trap files are required by DB2 Customer Support. They are located in the directory specified by the DIAGPATH database manager configuration parameter. The first letter in their names is "t", followed by a process identifier (pid). The file extension is 000 for single processors, or the node number for multi-node systems with a db2nodes.cfg file.

Examples:

• t56772.000 is a trap file for the process with pid 56772.
• t56772.010 is a trap file for the process with pid 56772. It was created by a DB2 process running on node 10.

Gathering Stack Traceback Information on UNIX-Based Systems

On UNIX-based systems, DB2 can generate a stack traceback for when you intentionally stop a DB2 procedure. This diagnostic information may be required by DB2 Customer Support when it is not possible to generate a trap file through a signal or exception.

A stack traceback file lists active internal functions in the order that they were called by a process, and shows the last steps of the function flow before the process was stopped. Each file reports the single signal or exception that was issued by the DB2 engine to stop or interrupt the process.

To activate a stack traceback:

1. Issue the db2_call_stack command to activate stack traceback.
2. Stop your DB2 instance.

The stack traceback information is located in trap files. There is one file for each process. For DB2 Enterprise - Extended Edition systems, there are files for each process on each node.

(Stack trace back information is not available for the SINIX operating system.)

Notes for DB2 Enterprise - Extended Edition:

• For systems with multiple nodes, use db2_all db2_call_stack to activate stack traceback on every node.
• The db2_call_stack command should be run by the instance owner only. Before running the command, ensure that a /tmp/$DB2INSTANCE directory exists at each node. The instance owner can create this directory by issuing the following command:

  db2_all "mkdir /tmp/$DB2INSTANCE"