The default for starting RTserver is to start the optimized version, in which validations and checking are turned off. Your first step in diagnosing RTserver problems is to ensure you are running the check version of RTserver so you can collect more information about whatever problem you are having. Start RTserver using:
This starts the check (also called debug) version of RTserver.
Because RTserver does not have any API functions, no C/C++ code can be added to RTserver to help diagnose problems. There are several files, command-line arguments, options, and commands available to debug RTserver without requiring any programming. Also, RTmon can be used to poll or watch information in RTserver. These and other debugging features are described below.
When RTserver starts, it creates a debug file as specified by the -trace_file
argument specified on the rtserver
command that you used when you started the RTserver. If you did not specify the -trace_file
argument, the default is to use standard output (stdout
), which is printed to the console.
If RTserver crashes or is having problems, the debug file (if you specified one) is the first place to look.
Each time RTserver prints something to its debug file, it immediately flushes the data to disk so that it can be easily typed out. When RTserver exits cleanly, it checks the size of the debug file and automatically removes the debug file if the file is empty. This automatic cleanup prevents hundreds or even thousands of useless debug files from clogging the filesystem that contains the directory where the debug files are written. The size of the debug file can be regulated with the Trace_File_Size option. If RTserver crashes, this debug file is a good place to look for error messages. The operating system modification date of the debug file can be used to correlate debug files with previous RTserver sessions.
On OpenVMS, RTserver also creates these files when RTserver starts a detached background process:
stdout
is stored in SYS$SCRATCH:RTSERVER_OUT_
Node
_
User
_
Counter
.TMP
.stderr
is stored in SYS$SCRATCH:RTSERVER_ERR_
Node
_
User
_
Counter
.TMP
. This file is not created unless RTserver writes something to stderr
.SYS$SCRATCH:RUN_RTSERVER_
Node
_
User
_
Counter
.TMP
.SYS$SCRATCH:RTSERVER_
Node
_
Pid
_OK.TMP
. RTserver waits for the detached background process to create this file once it has finished initialization.For OpenVMS:
If RTserver fails to start successfully on OpenVMS, you can look at these files for information such as stack traces and error messages.
The most useful RTserver command-line arguments for debugging are:
There are many different levels of trace information you can have written to the debug file. The most detailed setting is debug
. For complete information on the settings for trace level, see Starting RTserver.
These RTserver options are useful for debugging:
The Log_* options allow you to easily log messages into message files entering or leaving RTserver. The Trace_Level and Verbose options provide a way to watch the detailed operations of RTserver. When the Verbose option is set to TRUE
, RTserver prints out much information to the debug file as new processes connect, existing processes disconnect, and subjects are operated on. Setting Verbose to TRUE
is equivalent in level of detail to setting Trace_Level to verbose
.
TIBCO SmartSockets™ User’s Guide Software Release 6.8, July 2006 Copyright © TIBCO Software Inc. All rights reserved www.tibco.com |