Home | Previous Page | Next Page   Fault Tolerance > Consistency Checking >

Collecting Diagnostic Information

Several ONCONFIG parameters affect the way in which the database server collects diagnostic information. Because an assertion failure is generally an indication of an unforeseen problem, notify IBM Technical Support whenever one occurs. The diagnostic information collected is intended for the use of IBM technical staff. The contents and use of af.xxx files, shared memory dumps, core files are not further documented.

To determine the cause of the problem that triggered the assertion failure, it is critically important that you not destroy diagnostic information until IBM Technical Support indicates that you can do so. Send email with the af.xxx file to IBM Technical Support at tsmail@us.ibm.com. This file often contains information that they need to resolve the problem.

Several ONCONFIG parameters direct the database server to preserve diagnostic information whenever an assertion failure is detected or whenever the database server enters into an abort sequence:

For more information about the configuration parameters, see the IBM Informix: Administrator's Reference.

You decide whether to set these parameters. Diagnostic output can consume a large amount of disk space. (The exact content depends on the environment variables set and your operating system.) The elements of the output could include a copy of shared memory and a core dump.

Tip:
A core dump is an image of a process in memory at the time that the assertion failed. On some systems, core dumps include a copy of shared memory. Core dumps are useful only if this is the case.

Database server administrators with disk-space constraints might prefer to write a script that detects the presence of diagnostic output in a specified directory and sends the output to tape. This approach preserves the diagnostic information and minimizes the amount of disk space used.

Home | [ Top of Page | Previous Page | Next Page | Contents | Index ]