Creating a file with timestamp 4781

Peter T. Breuer

You mean two processes can access the same file in read mode and not in write mode?

OK

Yes, a core can happen but the goal is as follows. I have created a simplified example just to get the point across:

sqlreturncode = executequery( "select * from test" )

at this point, if a dump core does not take place (due to some crazy database issue itself) an error code is returned, then the error encountered would come back (hopefully a DB error - something like ORA-1010). At this point, I want all processing to stop and an application log created. In the app log, there would be the sqlreturncode, the stack, etc. leading up to this point.

Which was the whole goal of my post: to figure out a way to generate this type of log with useful information but with unique file names.

Again, this code can be processed from within many processes in my own pesudo-system.

I thought that I mentioned that the application dump was similar (in idea) to the core file - that you run gdb (or the debugger) against.

The systems I have worked on in the past would be defined (as an *example*) as follows: 4 boxes - on each box, there is 8 work processes. Each work process connects to (as in the case of an Oracle database) "shadow process". All "shadow processes" reside on one box which is defined as the database server (hosting Oracle). I am biased with Oracle since that is the one I am most familiar with at a systems level.

This is my definition of ONE system - whose purpose is to run applications. The system itself has it's own programming language (similar to REXX but called ABAP) that access the database.

In such a system, whenever an application (i.e. something written with the programming language of the system) abends, there is a separate ASCII file created that has the contents of what I have described above (stack of calling programs, the variables used when the calls took place, buttembly file information, etc.). But again, this is when the application fails.

I think the problem is communication here - when I think application, I think the different applications written in the system as defined above using the home-grown programming language. When you mention application, you are thinking of something like a web server - i.e. an application running in the OS.

As mentioned earlier, for each individual work process, there is a log file. There is also ONE log file for all work processes to log to as well - kinda like a global log file. Of course, the Oracle shadow processes will use the Oracle alert log or background trace files. The OS system log is used though if the DB abends totally - kinda in a foobar situation.

OK - as an example with log files for the http servers, etc (running within the OS).

I had started to write something else but I will leave well enough alone :-P

Thanks to all of those who actually tried to help me or give me a hint - it is appreciated ;-)

Kindest Regards.