I am working on web protocol on LR12 and created a script on application for which most of the activities of Application Launch and Login is handled at client side by implementing Java script. I have commented out the unwanted Java script requests in script and it working fine E2E.
After performing the load test, it has been found that the Avg. Response time values for Home Page Launch and Login is pretty high (> 25 sec).
Development team has suggested to increase the Browsers temporary internet files (Disk space ) from default 250 MB to 520 MB.
How can i simulate this setting in the script. Is there any available function which can help out?
I have not heard of any default setting where a browser would limit to 250MB for temp files.
What protocol(s) are used by the script? Is the slow login behavior only occurring in the script, or is it actually slow when accessing the application manually? If only in the script, is is slow in single-user execution, or does it degrade as load increases?
Additional details will be needed to be able to provide better input.
I also haven't heard of simulating the such browser settings, that's why wanted to take this with experts like you guys.
We are using web protocol in the script. Following is the few additional behavior that you asked.
1) When we recorded the script for the first time and enhanced it, despite 3-4 requests for content type text/html and few important JS, almost all request were JS, PNG and CSS content type. Replaying the script as it is were taking around 35 sec for login in Vugen for single user, while the manual reading on the browser for login was taking around 20 sec (even after cache clear) on httpwatch/fiddler.
I am puzzling why this is so? A lot of questions were being asked for this behavior. I checked with F12 developer tool while login manually, almost the same JS and CSS were there which was captured in script.
What would be the appropriate explanation for such behavior?
2) Then, i commented few JS (unwanted), all CSS and all PNG, and relapyed it successfully, this time it took around 19-21 sec, that is near abt. the manual reading.
3) When i put the commented script for load test with 1000 users, the response time of Login increased to 40 sec. The development team suggested to increase the temp. file size to 520 MB and on manual chekcing with single user we observed a significant improvement in Login response time. They are saying there is a minimal client-server traffic exchange for login and almost all the things are handled at cient side only.
The next challenge is since Loadrunner, works at middleware, Is it possible to simulate such setting and perform a load test? How should i check it?
Something like this:
I haven't went ahead and tally the Fiddler captured JS with that captured in script. In script, i commneted,
1) All CS
2) All PNG
3) All JS except few on which we captured the correlations.
We found the script to be running fine and recording and replay page is consistent for text/html content request.
Is there any way to find out that whether JS after executing is sending something to server or executing just at the client side doing some client side operation. If we would know the way to identify this, it would be easy to take decision, which JS to be commented out.
Of course, the easiest way is to have your developers provide you a list of JS that include logic to make server requests. If not possible, then if it were me I would first examine the comparison between JS downloaded by the script and what was captured with Fiddler. Then, I would review the JS content myself.