Hello Reza,

"I wonder how much error logging for purpose of maintenance is used routinely for complex systems by experienced users?"

Now that Michael and Kai have already replied, nothing much to add. Anyway a few more aspects which came to my mind :

1. Make the error logging event based and flag the logged event so that it is not repeatedly logged till cleared. Actually I have seen some error logs reading " Zone 1 over temperature " for every 1 second or so till the user resets. Makes the log quite useless and as boring as a cargo ship's loading manifest.

2. Once you decide to have the error logging event based, then you can decide as to which events will trigger the logging. And more important, in case of no error, make sure you have a " All systems Normal" marker at regular intervals - someone should know that you logger is alive and not got into an error itself !

Logging of error is a very involved process by itself and make sure you spend most of the time analysing and designing the process - coding for it is trivial. And if you can tell a little bit more on your project, maybe more useful guidelines can follow.

Raghu