11 Answers, 1 is accepted
In short, yes you must install some version of Test Studio on the machine you want to be able to gather performance data from, i.e Windows Performance Counters. Test Studio connects to our profiling service running on the remote machine in order to gather the counters you select during setup. Our profiling service comes with both the full edition and the run-time edition. It is not packaged separately or available any other way.
Yes this means you need a Run-Time license for each machine you want/need to gather performance information from. Fortunately the retail price for this is only $199 (compared to the $2499 for our full edition).
Does that adequately answer your questions?
the Telerik team
Test Studio Trainings
You are correct in that the Profiler service must be installed on the remote machines of which you want to gather performance data on. We do not have a separate license/product just for that service. The Profiler service is a component of the Run-time edition. You must install Run-time edition on each remote machine, after you have purchased the number of Run-time licenses needed, one for each machine Run-time edition will be installed on.
We have installed telerik run time edition on the server(Remote Machine) and added the server name( remote machine- to monitor the performance) in telerik which was installed on my local machine(telerik stand alone version) and executed the load test scripts. but when l look at the load test results(over all), it' s showing my local machine name, but not showing the server name(Remote machine name) which I have configured to monitor the performance.
Due to this issue , I'm not able to figure out exactly the following points
1. When I ran the load test scripts, whether the load generator is creating the virtual load on the configured server (Remote Machine) or not ?
2. On over all test results, I couldn't find the Server name (remote machine), which I have added in Telerik to monitor the performance.
3. Do I need to configure any thing on my local machine.. Kindly advice.. please.. Thank You .
Item 1) That depends on your setup and which method you used to run your load test. If you used "Run This Test" as shown on this page (which is the most common method) then your local machine generated and sent the traffic to your web server.
If you put the load test into a Test List and ran the test list on a remote machine, the machine you selected generated the load traffic and sent it to your web server. The load will not come from the machine you selected to monitor the performance of (unless it happens to be the same machine you setup to be an execution server and told the test list to run on that execution server).
Item 2) I've contacted development on this problem. I'm waiting for their assistance to further diagnose why this is happening.
Item 3) Short answer is none that I am aware of.
I'll get back to you as soon as I get some questions answered by my development team.
Thanks for the update. Could you please help me on item#2.
Item #2 On over all test results, I couldn't find the Server name (remote machine), which I have added in Telerik to monitor the performance.
Yes I'm finally ready to move on this problem. Here's the normal sequence of events:
1) During test initialization the Load Controller process asks the remote execution server where the Profiler process is. In other words the machine that will act as your load controller, generating the load, will reach out to the machine you are trying to profile (on port 55555) and ask it for connection info to the Profiler service that should be running on it. The usual response is something like this: "http://dad-pc:8031/".
2) The same Load Controller machine will make a connection to the Profiler service (usually running on port 8031) and gives it instructions what counters to profile and how often to report.
3) The Profiler service running on the remote machine starts making connections back to the Load Controller process reporting data samples taken. It repeats this at the specified Sampling Rate. The Load Controller stores the received data sample.
I'm ready to look at this problem via an online meeting. When are you available? Keep in mind I'm located in Austin, Texas which is Central US timezone.
I have designed a load test script for a web application and tried running with a user load of 2 users for 10 mins, but it ran for 2 mins 8 seconds and test got failed. I could see only the home page url on the page level metrics. I do not see any error codes on the page metrics Section. could you please help me to debug the issue.
Avg response time- 3000 value reached 1381
Current virtual users for over all test - 2 value reached 2
HTTP Errors/ Second for over all test -10 value reached 0 ; Responses Received/ sec for over all test -2 -value reached 0 ( zero)
That description sounds like you're getting faulted users. A Faulted User happens when something goes seriously wrong during the load test such that the VU is unable to run the remaining test steps. Instead it aborts the test in the middle of the steps, stops and a new VU is started which starts from step 1.
To diagnose what is causing faulted users I need two things from you:
1) A copy of the load test. This is the .tstest file on disk. You'll need to place it into a .zip file before attaching it.
2) A Fiddler trace of the load test trying to run, but starting and ending with only 1 user. I can't follow a Fiddler trace with multiple VU's in it.
I have installed run-time edition on the server and tried to configure the server on my stand lone edition to run the scripts on remote machine. but on entering the server name and clicking on test connection, it's showing an error message as "Cannot connect Scheduling Server").
I have verified the server machine ( where Run time edition installed) both scheduling server and execution servers are running.
Please advice me on this issue. Thank You.