Sterling doesn't crash the system, it crashes itself.
I guess you could develop some kind of system to automate restarting sterling. How would that work..
Constantly search for the error message box (using FindWindow() Win32 API), then when it appears, force terminate the sterling process. Restart it, wait a while, find the login box and emulate some user input to type in the login info. Maybe this is possible with AutoHotKey? Even if it is though, my god it would be such an ugly kludge.
I am primarily into Linu; some of the standard techniques that we normally use in such situation might help you…
1. Write a python/ant cron job (win equivalent) that kills the sterlingpro process at a certain time of the day and restarts in at a certain type. In linux we normally issue a “kill -9 <pid>”… I am sure, windows has something similar in the command line of in win32/mfc...
2. Run a daemon that reboots your machine at a certain time of the day (whenever sterling database starts up)… put sterling pro init in the system startup routines… so whenever the machine reboots, the startup procedure starts up sterling pro…
This is how we can normally handle such situation in Linux… you can use “expect” to enter log in username and password in linux command line…. Not sure if expect can help you in windows.
Also, I know there are a few GUI automation QA tools in windows. You can automate the process of entering the username and password using these tools… see if any of these can help you http://blog.dreamcss.com/tools/gui-testing-tools/
You can try all these techniques if you decide not to leave sterling Pro… but if you do, I am sure you have a lot of options.
Hope this helps