On Windows, a crashed server may cause information to be written to the system’s drwtsn32.log file, a file that CA Customer Support may ask you to send to help analyze the problem.
On UNIX, a crashed server creates a core file in $PSHOME/bin unless you have configured your server not to generate core files. If a core file is generated, please do not send it to CA unless instructed to do so. Instead, run the command pstack core > pstack.txt to capture the stack traces of all threads running within the crashed application. This output is valuable in diagnosing the failure.
On Windows, a hung server (one where one or more requests did not run to completion) can only be debugged using the provisioning server trace log (PSHOME\logs\etatranyyyymmdd-hhmm.log) in conjunction with the analyzelog utility. You will generally be asked to capture the provisioning server trace log (at logging level 7 if at all possible) and CA will use “analyzelog” to locate operations that have not yet completed. The C++ Connector Server trace log (satransyyyymmdd-hhmm.log) and sometimes other logs are also useful to collect.
On UNIX, capturing the provisioning server trace log and running analyzelog is still useful. But another option that often provides additional information is once again the pstack command. Locate the process ID (pid) of the hung service by reading the contents of the file $PSHOME/data/pid/servicename.pid, and then issue the command pstack pid > pstack.txt to capture the stack traces of all active threads within the running process. Please include this output file along with the provisioning server trace logs.
|
Copyright © 2013 CA.
All rights reserved.
|
|