In my previous article, I launched the open supply take a look at instrument JMeter and used a easy HTTP take a look at for example to show its capabilities. This article exhibits you the way to construct take a look at scripts for advanced take a look at situations.
The person interface shows a JMeter take a look at script within the “tree” format. The saved take a look at script (within the .jmx format) is XML. The JMeter script tree treats a take a look at plan as the basis node, and the take a look at plan consists of all take a look at elements. In the take a look at plan, you may configure user-defined variables known as by elements all through the whole take a look at plan. Variables may also thread group conduct, library information used within the take a look at, and so forth. You can construct wealthy take a look at situations utilizing varied take a look at elements within the take a look at plan.
Test elements in JMeter usually have the next classes:
- Thread group
- Sampler
- Logic controller
- Listener
- Configuration factor
- Assertion
- Timer
- Pre-processor
- Post-processor
Thread teams
A thread group is the start level for all take a look at plans (so all samplers and controllers have to be positioned beneath a thread group). A thread group could be considered a digital person pool during which every thread is actually a digital person, and a number of digital customers carry out the identical batch of duties concurrently. Each thread is impartial and would not have an effect on the others. During the execution of 1 thread, the variable of the present thread would not have an effect on the variable worth of different threads.
(Chongyuan Yin, CC BY-SA 4.0)
In this interface, the thread group could be configured in varied methods.
1. Action to be taken after a sampler error
The following configuration objects management whether or not a take a look at continues when an error is encountered:
- Continue: Ignore errors and proceed execution.
- Start Next Thread Loop: Ignore the error, terminate the present loop of the thread, and execute the subsequent loop.
- Stop Thread: Stop executing the present thread with out affecting the traditional execution of different threads.
- Stop Test: Stop the whole thread after executing threads have completed the present sampling.
- Stop Test Now: The complete take a look at execution stops instantly, even when it interrupts presently executing samplers.
2. Number of threads
This is the variety of concurrent (digital) customers. Each thread runs the take a look at plan fully independently with out interfering with any others. The take a look at makes use of a number of threads to simulate concurrent entry to the server.
3. Ramp-up interval
The Ramp-up time units the time required to start out all threads. For instance, if the variety of threads is about to 10 and the ramp-up time is about to 100 seconds, then JMeter makes use of 100 seconds to start out and runs 10 threads (every thread begins 10 seconds after the earlier thread was began).
If the ramp-up worth is about small and the variety of threads is about giant, there’s a whole lot of stress on the server in the beginning of the take a look at.
4. Loop rely
Sets the variety of loops per thread within the thread group earlier than ending.
5. Delay thread creation till wanted
By default, all threads are created when the take a look at begins. If this selection is checked, threads are created when they’re wanted.
6. Specify thread lifetime
Control the execution time of thread teams. You can set the period and startup delay (in seconds).
Samplers
A sampler simulates person operations. It’s a working unit that sends requests to the server and receives response knowledge from the server. A sampler is a element inside a thread group, so it have to be added to the thread group. JMeter natively helps quite a lot of samplers, together with a TCP Sampler, HTTP Request, FTP Request, JDBC Request, Java Request, and so forth. Each sort of sampler sends completely different requests to the server based on the set parameters.
TCP Sampler
The TCP Sampler connects to the required server over TCP/IP, sends a message to the server after the connection is profitable, after which waits for the server to answer.
(Chongyuan Yin, CC BY-SA 4.0)
The properties that may be set within the TCP Sampler are as follows:
TCPClient classname
This represents the implementation class that handles the request. By default, org.apache.jmeter.protocol.tcp.sampler.TCPClientImpl is used, and plain textual content is used for transmission. In addition, JMeter additionally has built-in assist for BinaryTCPClientImpl and SizePrefixedBinaryTCPClientImpl. The former makes use of hexadecimal packets, and the latter provides a 2-byte size prefix to BinaryTCPClientImpl.
You may also present customized implementation lessons by extending org.apache.jmeter.protocol.tcp.sampler.TCPClient.
- Target server settings: Server Name or IP and Port Number specify the hostname or IP deal with and port variety of the server utility.
- Connection Options: Determines the way you hook up with the server.
-
- Re-use connection: If enabled, this connection is all the time open; in any other case, it is closed after studying knowledge.
- Close Connection: If enabled, this connection is closed after the TCP sampler has completed working.
- Set No-Delay: If enabled, the Nagle algorithm is disabled, and the sending of small packets is allowed.
- SO_LINGER: Controls whether or not to attend for knowledge within the buffer to finish transmission earlier than closing the connection.
- End of line (EOL) byte worth: Determines the byte worth on the finish of the road. The EOL test is skipped if the required worth is bigger than 127 or lower than -128. For instance, if a string returned by the server ends with a carriage return, you may set this selection to 10.
- Timeouts: Set the join timeout and response timeout.
- Text to ship: Contains the payload you wish to ship.
- Login configuration: Sets the username and password used for the connection.
HTTP Request Sampler
The HTTP Sampler sends HTTP and HTTPS requests to the online server.
(Chongyuan Yin, CC BY-SA 4.0)
Here are the settings out there:
- Name and feedback
- Protocol: Set the protocol to ship the request to the goal server, which could be HTTP, HTTPS, or FILE. The default is HTTP.
- Server identify or IP deal with: The hostname or IP deal with of the goal server to which the request is distributed.
- Port quantity: The port quantity that the online service listens on. The default port is 80 for HTTP and 443 for HTTPS.
- Request technique: The technique for sending the request, generally together with GET, POST, DELETE, PUT, TRACE, HEAD, OPTIONS, and so forth.
- Path: The goal URL (excluding server deal with and port) to request.
- Content encoding: How to encode the request (relevant to POST, PUT, PATCH, and FILE).
- Advanced request choices: A number of further choices, together with:
-
- Redirect Automatically: Redirection isn’t handled as a separate request and isn’t recorded by JMeter.
- Follow Redirects: Each redirection is handled as a separate request and is recorded by JMeter.
- Use KeepAlive: If enabled,
Connection: keep-aliveis added to the request header when JMeter communicates with the goal server. - Use multipart/form-data for POST: If enabled, requests are despatched utilizing
multipart/form-data or utility/x-www-form-urlencoded.
- Parameters: JMeter makes use of parameter key-value pairs to generate request parameters and ship these request parameters in several methods relying on the request technique. For instance, for GET, DELETE requests, parameters are appended to the request URL.
- Message physique knowledge: If you wish to go parameters in JSON format, you will need to configure the Content-Type as
utility/jsonwithin the request header. - File add: Send a file within the request. The HTTP file add conduct could be simulated on this method (normally).
Logic Controllers
The JMeter Logic Controller controls the execution logic of elements. The JMeter web site explains it like this: “Logic Controllers determine the order in which Samplers are processed.”
The Logic Controller can management the execution order of the samplers. Therefore, the controller must be used along with the sampler. Except for the as soon as–solely controller, different logic controllers could be nested inside one another.
Logic controllers in JMeter are primarily divided into two classes. They can management the logical execution order of nodes through the execution of the take a look at plan (a loop or conditional controller), or they will act in response to particular throughput or transaction rely.
Transaction Controller
Sometimes, you wish to rely the general response time of a bunch of associated requests. In this case, you want to use a Transaction Controller.
The Transaction Controller counts the sampler execution time of all youngster nodes beneath the controller. If a number of samplers are outlined beneath the Transaction Controller, then the transaction is taken into account profitable solely when all samplers run efficiently.
Add a transaction controller utilizing the contextual menu:
(Chongyuan Yin, CC BY-SA 4.0)
Generate guardian pattern: If enabled, the Transaction Controller is used as a guardian pattern for different samplers. Otherwise, the Transaction Controller is simply used as an impartial pattern.
(Chongyuan Yin, CC BY-SA 4.0)
For instance, the unchecked Summary Report is as follows:
(Chongyuan Yin, CC BY-SA 4.0)
If checked, the Summary Report is as follows:
(Chongyuan Yin, CC BY-SA 4.0)
Include period of timer: If enabled, embody a timer (a delay is added earlier than and after the sampler runs).
Once Only Controller
The Once Only Controller, as its identify implies, is a controller that executes solely as soon as. The request beneath the controller is executed solely as soon as through the loop execution course of beneath the thread group. For checks that require a login, you may think about placing the login request in a Once Only Controller as a result of the login request solely must be executed as soon as to determine a session.
(Chongyuan Yin, CC BY-SA 4.0)
If you set the loop rely to 2 and test the consequence tree after working, you may see that the HTTP request 3 beneath the Once Only Controller is simply executed as soon as, and different requests are executed twice.
(Chongyuan Yin, CC BY-SA 4.0)
Listeners
A listener is a collection of elements that course of and visualize take a look at consequence knowledge. View Results Tree, Graph Results, and Aggregate Report are widespread listener elements.
View Results Tree
This element shows the consequence, request content material, response time, response code, and response content material of every sampler in a tree construction. Viewing the knowledge can help in analyzing whether or not there’s a drawback. It supplies varied viewing codecs and filtering strategies and may also write the outcomes to specified information for batch evaluation and processing.
(Chongyuan Yin, CC BY-SA 4.0)
Configuration factor
Configuration factor supplies assist for static knowledge configuration. It could be outlined on the take a look at plan stage, or on the thread group or sampler stage, with completely different scopes for various ranges. Configuration parts primarily embody User Defined Variables, CSV Data Set Config, TCP Sampler Config, HTTP Cookie Manager, and so forth.
User-Defined Variables
(Chongyuan Yin, CC BY-SA 4.0)
By setting a collection of variables, you trigger a random collection of values for use within the efficiency take a look at. Variable names could be referenced throughout the scope, and variables could be referenced as ${variable identify}.
In addition to the User Defined Variables element, variables can be outlined in different elements, resembling Test Plans and HTTP Requests:
(Chongyuan Yin, CC BY-SA 4.0)
For instance, an outlined variable is referenced in an HTTP Request:
(Chongyuan Yin, CC BY-SA 4.0)
Viewing the execution outcomes, you may see that the worth of the variable has been obtained:
(Chongyuan Yin, CC BY-SA 4.0)
CSV Data Set Config
During a efficiency take a look at, chances are you’ll want parameterized enter, such because the username and password, within the login operation. When the quantity of concurrency is comparatively giant, the information technology at runtime causes a heavy burden on the CPU and reminiscence. The CSV Data Set Config can be utilized because the supply of parameters required on this situation.
(Chongyuan Yin, CC BY-SA 4.0)
The descriptions of some parameters within the CSV Data Set Config:
- Variable identify: Defines the parameter identify within the CSV file, which the script can reference as
${variable identify}. - Recycle on EOF: If set to True, this permits looping once more from the start when reaching the top of the CSV file.
- Stop thread on EOF: If set to True, this stops working after studying the final file within the CSV file.
- Sharing mode: Sets the mode shared between threads and thread teams.
Assertions
The Assertion checks whether or not the request is returned as anticipated. Assertions are an essential a part of automated take a look at scripts, so it’s best to pay nice consideration to it.
JMeter generally used assertions embody Response Assertion, JSON Assertion, Size Assertion, Duration Assertion, Beanshell Assertion, and so forth. Below I introduce the frequently-used JSON Assertion.
JSON Assertion
This is used to claim the content material of the response in JSON format. A JSON Assertion is added on an HTTP Sampler on this instance, as proven within the following picture:
(Chongyuan Yin, CC BY-SA 4.0)
The root of the JSON path is all the time known as $, which could be represented by two completely different kinds: dot-notation (.) or bracket-notation ([]). For instance; $.message[0].identify or $['message'][0]['name'].
Here’s an instance of a request made to https://www.google.com/. The $[0].identify worth represents the ‘identify’ half within the first array factor within the response.
(Chongyuan Yin, CC BY-SA 4.0)
The Additionally assert worth specifies that the worth of 'identify' is to be verified, and the Expected worth is predicted to be ‘2022-world-cup-opening-day’.
Run the script and have a look at the outcomes. You can see that the assertion has handed.
(Chongyuan Yin, CC BY-SA 4.0)
Here are the potential situations and the way they’re handled:
- If a response consequence isn’t in JSON format, it is handled as a failure.
- If the JSON path can’t discover the factor, it fails.
- If the JSON path finds the factor, however no situations are set, it passes.
- If the JSON path finds a component that doesn’t meet the situations, it fails.
- If the JSON path finds the factor that meets the situations, it passes.
- If the JSON path returns an array, it iterates to find out whether or not any parts meet the situations. If sure, it passes. If not, it fails.
Go again to JSON Assertion and test the Invert assertion.
(Chongyuan Yin, CC BY-SA 4.0)
Run the script, test the outcomes, and you may see that the assertion failed:
(Chongyuan Yin, CC BY-SA 4.0)
Timers
The pause time between requests within the efficiency take a look at is named “thinking time.” In the true world, the pause time could be spent on content material search or studying, and the Timer simulates this pause.
All timers in the identical scope are executed earlier than the samplers.
If you need the timer to be utilized to solely one of many samplers, add the timer to the kid node of the sampler.
JMeter timers primarily embody Constant Timer, Uniform Random Timer, Precise Throughput Timer, Constant Throughput Timer, Gaussian Random Timer, JSR223 Timer, Poisson Random Timer, Synchronizing Timer, and BeanShell Timer.
Constant Timer
A Constant Timer implies that the interval between every request is a hard and fast worth.
(Chongyuan Yin, CC BY-SA 4.0)
After configuring the thread delay to 100 and 1000, respectively, run the script:
(Chongyuan Yin, CC BY-SA 4.0)
Check the information within the desk, the place #1 and #2 are the working outcomes when the configuration is 100 milliseconds, and #4 and #5 are the working outcomes when the configuration is 1000 milliseconds. You can see that the interval between #4 and #5 is considerably higher than that between #1 and #2:
(Chongyuan Yin, CC BY-SA 4.0)
Constant Throughput Timer
The Constant Throughput Timer controls the execution of requests based on the required throughput.
(Chongyuan Yin, CC BY-SA 4.0)
Configure the goal throughput as 120 (notice that the unit is minutes), after which choose All energetic threads in present thread group (shared) primarily based on the calculated throughput:
(Chongyuan Yin, CC BY-SA 4.0)
Run the script, test the outcomes, and observe that the throughput is roughly 2/second (120/60).
(Chongyuan Yin, CC BY-SA 4.0)
Pre-processors and post-processors
A pre-processor performs some operations earlier than the sampler request. It’s usually used to switch parameters, set atmosphere variables, or replace variables.
Similarly, a post-processor performs some operations after the sampler request. Sometimes, the response knowledge must be utilized in subsequent requests, and you want to course of the response knowledge. For instance, if the jwt token within the response is obtained and used for authentication in subsequent requests, the post-processor is used.
Using JMeter
The above is the introduction to the principle take a look at elements of JMeter, and now you may really feel assured in beginning your individual checks. In one other article, I’ll clarify utilizing the MQTT plugin in JMeter.
