We are using WinRunner 9.2 with scripts stored in Quality Center 9.2. We have been having a problem with scripts failing after two or more hours with an Explorer crash. In trouble-shooting, we found several instances in our framework where Explorer is stopped and started. We eliminated all (I think) of them.
Now my script failed after 13 hours, 47 minutes with an Internet Explorer crash in ie4_trap.dll on one machine (with the debugger running), after 5 hours, 49 minutes on another (with IE crash notice), and 2 hours, 21 minutes (with IE crash notice) on the third. My script is a fairly tight loop that takes about 1 minute each iteration to launch my application, logon to a data server, then logoff and close the application. The crash occurs after the launch and before the logon. It occurs even when the logon and logoff are eliminated from the script.
As far as I know, there is no call to Internet Explorer. (The next several hours will be devoted to reading through the framework again looking for IE calls.) However, since we are attached to Quality Center, do you know whether and how WinRunner is using the ie4_trap.dll?
Any ideas for further investigation?
It is imperative that we automate overnight testing and this problem is preventing unattended testing.
Thanks for any assistance.
Try running without the WinRunner WebTest addin loaded - that will tell you if it's a WinRunner or QC issue. Almost guranteed it's a WR issue - there are known memory leaks, and long tests usually fail with similar issues. It's a good practice to keep tests fairly short (< 30 min), and run them in a suite (especiall if you have QC). If you're running a longevity test, you may be better off using LoadRunner, or a similar tool which implements HTTP requests directly (instead of through the Browser).