I am running a loadtest consisting of a single recorded web test, with a constant 50 User load for 15 minutes.
The test requests data driven aspx pages from my localhost, with the necessary parameters being stored in csv files and bound to the test.
After running the test (and other ones like it) I consistently see values such as:
Total Request 125,338
Req/Sec: Range 1000, Min 0.0, Max 199, Avg. 138
Avg Response: Min 0.01, Max 1.15, Avg 0.05
Total Test: 1,043
The problem is that a colleague running the same app, same database data and same VS test project (all copied from my machine & ran on his) will continually get these results.
Total Request 17,884
Req/Sec: Range 100, Min 9.8, Max 30, Avg. 19.9
Avg Response: Min 0.009, Max 0.071, Avg 0.03
Total Test: 1,251
I can understand the difference in the response time from his development environment set up to my one.
But why would I be seeing over 100,000 more requests made during the 15 minutes than he would
Request/Sec Range value
Thomas Kurek
Can you look to the requests table to see if there are any clues there Do you see the same set of requests listed here Are all the requests passing in both cases Is it possible that an error page is being returned in one case and not the other
It wasn't clear to me if you are both hitting the exact same server for the pages, is that the case
Please check the requests table and reply back.
Thanks,
Rick
Waldo Wrangler
When I run the web test by itself it completes without error (All green ticks against each page).
After running the load test using that single scenario, the Request Table shows the 7 correct aspx pages listed, with the addition of the “DefaultErrorPage.aspx” listed 3 times.
There is a failed test count of 13 next to one of the pages, the same number as the loadtest failed test count.
The interesting thing is that the Content Length for that page is only 150, when it should be somewhere around 19,000
So it looks like the app has been failing when run on my set up, but maybe sending out the right header for the test to think that it was a success!
Unless the Content Length displayed is only the last received value, and the last request to that page was a failure.
I work remotely so I have a completely separate environment to my colleague.
He is testing solely on his single development box, while have the VS Test project on one, and the web application running on another.
Thanks for the suggestion Rick.
Cheers
William
themadmax
When running the web test by it's self, all the returned pages are correct within the results viewer.
When run within the load test, the test runs smoothly for about a minute and a half, and then the Request / Sec range jumps up by 10 fold.
It must be around here the app is actually failing, so by adding a validation rule in the right place may show this to be the case.
Thanks for your input.
Much appreciated
William
Christian Damborg
If the error page is returning a 200 Http status code then the request will be considered passing.
It would be a good idea to view the pages returned in the web test result viewer to visually verify that you're getting what you expect, or add a validation rule to help detect error situations.
For instance, you could add a validation rule to check for a unique string returned on the error page (or maybe a validation rule that looks for something in the response headers that would indicate the error page was returned). The request would be marked as failed if the validation rule fails.
One thing to note regarding validation rules... By default a validation rule is created with a level of "high". By default, a load test only runs validation rule at level "low" (thus, those marked "high" would not be run by the load test by default -- this is a property of the load test's run setting and can be changed). The idea is that you use the validation rules to verify the web test run before turning it lose on a load test. When you're confident your web test scripts are running correctly you're able to avoid the overhead of running validation rule during the load test. Of course, you can always change the load test run setting to always run validation rules, just be aware that it doesn't happen by default.
Thanks,
Rick