Stack Traces in Case of Critical Error

A diagnostic feature is available for certain supported operating systems that provides a stack trace showing calls for all threads. This is performed automatically when a fatal error occurs.

This feature redirects output to the file pstack<server_pid>_<sernum>.log, where <server_pid> is the process ID of the c-treeACE Server process and <sernum> is a serial number maintained by the server to ensure unique log names. The server also writes a message to its status log, CTSTATUS.FCS, indicating that a process stack trace was dumped. For example, the following message in CTSTATUS.FCS refers to the file pstack_454_01.log:

Dumped stack for server process 454, log=1, loc=73, rc=0

Please contact your vendor immediately if you encounter a c-treeACE crash and be prepared to supply these stack dumpa. This enables FairCom engineers to quickly pinpoint critical errors and enable fast response times for corrected servers. In some cases, FairCom has been able to provide a turnaround time in less than a day!

Solaris Operating Systems

For Servers running on Solaris, the implementation invokes the pstack utility.

Microsoft Windows Operating Systems

This feature requires an external DLL, dbghelp.dll, which is part of the Windows installation, and is dynamically loaded at stack dump time. It also requires that this DLL exports the function MiniDumpWriteDump() which may not be the case with older versions of this DLL. In the case where the DLL cannot be found or it does not contain the function, the stack dump fails and a message is logged in CTSTATUS.FCS without any other consequence.

It is suggested to enable Dr. Watson on the Windows system of interest when attempting to generate a dump file with information needed to trace a continuing server stack dump. Visual Studio 7.0 or greater is required to inspect the file for Windows stack dumps.

IBM AIX Operating Systems

The server uses the AIX procstack utility to log the stack trace.