We need to test our web application with 100+ users with JMeter tool.
Should we use JMeter webdriver plugin to launch the real browsers or can we use the blazemeter plugin to generate the script and run in JMeter? Will the script created with blazemeter simulate the real browser load?
Theoretically you can use real browsers, but be aware that browsers are very resource intensive so my expectation is that you will need at least one CPU core and at 1 GB of RAM per browser instance in order to have enough resources.
This means you will have to go for distributed testing and on average you will have one slave per 5 browsers. If this is something you can afford - go for it. If don't - be aware that you can configure JMeter to behave like a real browser, to wit:
Tell JMeter to Retrieve All Embedded Resources from HTML Files and Use concurrent pool of 6 threads. This can be done using HTTP Request Defaults, the settings live under "Advanced" tab
Add HTTP Cache Manager to simulate browser cache
Add HTTP Cookie Manager to mimic browser cookies
Add HTTP Header Manager to send browser-like headers (i.e. User-Agent, Accept-Encoding, etc.)
If your application uses AJAX requests - put the relevant HTTP Request samplers under the Parallel Controller
Related
I want to perform load test on website built on react js with 500 users.
Is it possible with jmeter without selenium plugin?
You can create JMeter test plans and simulate the 500 concurrent users accessing the system.
Record the test script
Correlate the session variables
Add timers
Configure the thread group
Run the test
There are a number of ways to build a JMeter test plan. You can use HTTPS Test Script Recorder or Blazemeter Chrome Extension to record the steps. Then you will have to add timers, correlate the sessions with the JMeter post-processors, etc to simulate the real users.
Ten ways to build your JMeter test plans
In your Thread Group set the Number of Threads (users) to 500 to simulate 500 concurrent users. Add a ramp-up time (e.g 60 seconds).
Note: Timers will simulate the think time of the users or the delays between the requests. You may use Random Timers to simulate different think times.
It is, JMeter doesn't know/care anything about the system under test technology so as long as JMeter supports network protocols used by the application it should be able to replicate the network footprint of the real browser.
Browsers in their turn don't do any magic, they send HTTP requests, wait for responses and render the response in the browser window.
JMeter is capable of sending HTTP requests via HTTP Request sampler so it can load test web applications, including but not limited to react js ones.
Just don't forget to configure JMeter to behave like a real browser:
Configure HTTP Request samplers to retrieve embedded resources (images, scripts, styles, fonts, etc.) from web pages
Use HTTP Cache Manager to simulate browser cache
Use HTTP Cookie Manager to simulate browser cookies
Use HTTP Header Manager to simulate browser headers
Use Timers to simulate real user "think times" (real users don't hammer the website non-stop, they need some time to "think" between operations, read something, type something, etc.)
I have been using Jmeter for performance testing my web application. I have recorded the jmeter script by excluding js,css and other static content files.
While running the script, Jmeter doesnt execute javascript files so ajax XHR request are not sent. To overcome this i have recorded the script with js, css and other static content and it recorded all the Ajax xhr request too. But the performance results seems to be different from the browser loading time. Also i need to use cache during my performance testing.
Below is how my test plan will look like,
Included Retrieve all embedded resources in HTTP Request manager.
Concurrent pool size is 6
I have added HTTP cookie manager and Cache manager.
I have added a loop controller (This is for caching, jmeter will cache the files on first iteration and it will use the cached files
after that)
The problem i am facing is that the time taken for rest call are double the time shown in the browser console for single user. I have tried all other combinations but always i am getting higher time than the browser console.
I have tried to use the Selenium webdriver plugin to simulate the browser behavior but it doesnt seems to be using the cache. (https://www.blazemeter.com/blog/how-load-test-ajaxxhr-enabled-sites-jmeter)
Is there any other way to solve this problem? I want to take the metrics with cache so kindly suggest me any solution that must include cache. Or is there any other tool similar to jmeter that could solve this issue. My goal is to take web page load time with cache for 'n' number of users.
PS : I am even interested to write any scripts in jmeter but the scripts should not overload the performance of jmeter.
Thanks in advance.
You should not be recording calls to embedded resources (images, scripts, styles, fonts, etc.) as if you record them - they will be executed sequentially while real browsers do this in parallel. So remove recorded requests for the embedded resources and "tell" JMeter to download them (and do it in parallel) using HTTP Request Defaults
You should be recording AJAX requests, however real browsers execute them in parallel while JMeter runs them sequentially. In order to make JMeter's behavior closer to real browser you need to put these AJAX calls under the Parallel Controller
You can install the Parallel Controller extension using JMeter Plugins Manager
Jmeter and caching are unrelated - everything that happens after data went over the wire is out of scope by design. You should only ever simulate requests that you expect NOT to be cached. So this is the feasible part: drop all requests that you expect the browser to cache from the Jmeter script (or move them outside the loop). On the load time of XHR: the browser will most certainly use HTTP keepAlive. Result is that all requests except the very first skip the setup and teardown phase of TCP sockets and are much faster - esp. when the request itself is small and quick. You can simulate this in JMeter also by checking the KeepAlive option AND selecting http commons as implementation. You can read up on this in the docs here: http://jmeter.apache.org/usermanual/component_reference.html#HTTP_Request
I want to do browser based load testing with JMeter.This I can achieve by using web diver sampler and a thread group with multiple concurrent users. This limits me to test only with about 10 users as opening more browser instances eats away my system resources.My objective is to see how the GUI behaves when the server is under heavy load (say 1000) users.
As per the Web Driver Tutorial
Note: It is NOT the intention of this project to replace the HTTP Samplers included in JMeter. Rather it is meant to compliment them by measuring the end user load time.
So you should not be creating the main load using the WebDriver Sampler, you can use it in another Thread Group in parallel with the main load conducted by the HTTP Request samplers to measure page rendering speed, catch any markup or JavaScript errors, etc.
Well-behaved JMeter virtual user looks just like a real user using the real browser from backend perspective so make sure you properly mimic real user behavior, to wit:
"embedded resources" handling (images, scripts, styles, fonts, etc.)
cache
headers
cookies
"think times"
Check out How to make JMeter behave more like a real browser for more details.
This is the case to work with Remove/Distributed Testing
Note: The same test plan is run by all the servers. JMeter does not distribute the load between servers, each runs the full test plan. So if you set 1000 Threads and have 6 JMeter server, you end up injecting 6000 Threads.
So in your case if you have 100 JMeter servers each will execute 10 users opening browser.
How can I execute those JMeter recorded script in parallel as we do while creating HTTP SAMPLER (Embedded HTTP Resource) ?
While recording I also checked Retrieve All Embedded Resource with pool of 6. Because of this I getting incorrect response time(Varies from browser timing of page).
Is there any way that we could execute our recorded HTTP Samplers in parallel?
As of JMeter version 3.2 it is not possible to kick off extra threads to run a specific sampler(s) group in parallel, each thread (virtual user) executes samplers upside down.
So you should be very careful with what you are recording. For example, you must not be recording any embedded resources calls, there is `URLs Patterns to Exclude" input on the "Requests Filtering" tab of the HTTP(S) Test Script Recorder where you can define which resources need to be excluded from the recording.
Remember that well-behaved JMeter test should be producing response time similar to real browser, but you need to configure JMeter to mimic real browser, to wit:
Add HTTP Header Manager to represent browser headers
Add HTTP Cookie Manager to represent browser cookies and deal with cookie-based authentication
Add HTTP Cache Manager to act like browser's memory and disk caches
See How to make JMeter behave more like a real browser guide for more information on JMeter fine tuning.
I am trying to setup a JMeter load test, using the Recording Controller.
This results in duplication of all the calls the web browser would make - including css files, js files and images.
Given that a web browser would get these concurrently (performing approximately 10 web requests concurrently). The jmeter documentation tends to equate 1 user to 1 thread - but this does not line up with the way in which modern web browsers work.
Do I need to use 10 threads for each user when setting up my load test?
Strictly Speaking, JMeter can't simulate a browser. JMeter has its own limitations, like not parsing .js files (as per 3.0 version), sending ajax requests from events etc.
But, for your question, there is support in JMeter.
So, you can specify the concurrent pool size as per your requirements, say concurrent pool size value set to 10, which would solve your issue.
Today browser normally sends requests concurrently and each browser has its own max value for maximum connections that can be made concurrently.
Refer following links:
Max parallel http connections in a browser?
and How to solve Chrome's 6 connection limit when using xhr polling.
When you use "Retrieve Embedded Resources" and set Concurrent Pool Size, you ONLY need to add parent sampler, which triggers resource requests like .css, .js files by Jmeter itself. (Eg: add only stackoverflow.com and it will load all the resources like .css, .js automatically). During recording , all these requests are seperately recorded, so you need to remove all of them or create a new plan/thread group containing only parent sampler.
Note: As I mentioned, requests triggered from .js, .csv won't be sent as Jmeter won't parse them like Browser. You need to add ONLY these requests as samplers explicitly. Add View Results Tree and compare the requests triggered by Jmeter and Browser (F12 -> Network tab) to know which request are missing by Jmeter.
First of all, don't record requests to embedded resources. If you click Add Suggested Excludes button the HTTP(S) Test Script Recorder will automatically populate a regular expression to filter images, scripts and styles.
Second. To simulate browser behaviour: 1 main request to the page and several parallel requests to retrieve content you can use "Advanced" tab of the HTTP Request Sampler (or even better HTTP Request Defaults). You can also limit embedded resources to your application under test domain there
Real browsers download images, styles and scripts. However well-behaved browsers do it only once, on subsequent requests these entities are being returned from cache. So make sure you add HTTP Cache Manager to your Test Plan to represent browser cache so you could avoid overwhelming your server with extra requests which don't happen in the reality.
See How To Make JMeter Behave More Like A Real Browser for more detailed explanation and instructions on using aforementioned Test Elements