LOG8371 - TP2 - Team Alpha

• Francis Forget
• Michaël Salmon
• Vincent Labonté
• Samuel Chapleau

GitHub

https://github.com/vincentlabonte/jguwekarest

Video

https://youtu.be/HTc6p5JIkZw

Setup

1. Download, Install and Setup Docker
2. Download, Install and Setup JProfiler
3. Download, Install and Setup JMeter
4. Download or Fork our GitHub Repository
5. Follow instructions as specified below

Docker

Build the Docker Image

• Download the source code from our Github repository
  git clone [email protected]:vincentlabonte/jguwekarest.git
• Change into code directory
  cd jguwekarest
• Checkout branch master for OpenAPI 3.0.1
  git checkout master
• Compile the war (Web Application Archive) file with maven
  mvn clean package
• Build the docker image (replace dockerhubuser with your docker hub account user)
  docker build -t dockerhubuser/jguweka:OAS3 .
• Check images
  docker images

Run the Docker Container

• If you run the container locally don't forget to start also a mongodb container as a data base with:
  docker pull mongo; docker run --name mongodb -d mongo
• Run the image as a local container (replace dockerhubuser with your docker hub account user)
  docker run -p 8080:8080 --link mongodb:mongodb dockerhubuser/jguweka:OAS3
• Load the Swagger-UI representation in a web-browser at port 8080
  e.g.: firefox http://0.0.0.0:8080

JProfiler

Profile the execution using JProfiler

• Replace dockerhubuser with your docker hub account user in the docker-compose.yml file
• Launch the containers
  docker-compose up
• Connect to the first container running the WEKA RESTful API Webservice
  docker exec -it jguwekarest_jguweka_1 bash
• Attach JProfiler to the running process in the Weka container /usr/local/jprofiler10.1.5/bin/jpenable -g -p 8849
• JProfiler GUI -> Session -> New Session
• Session Settings -> Session Type -> Attach Type -> Attach to remote JVM
• Execute docker ps and take note of the port {rest_port} linked to port 8849 of the jguwekarest_jguweka_1 container
• Profiled JVM Settings -> Direct network connection to 127.0.0.1 -> Profiling port {rest_port}
• Session Startup -> Initial Profiling Settings -> Sampling (Recommended) -> OK
• Send requests to the server
• JProfiler GUI -> Save Snapshot

JMeter

Create a Test Plan

• Open JMeter
  jmeter
• Add a Thread Group to the Test Plan
  • Set the number of threads and the ramp-up period
• Add a HTTP Request Defaults to the Thread Group
  • Set the protocol (http), the server ip (127.0.0.1) and the port number (80)
  • Set the client implementation to Java in the Advanced tab
• Add a HTTP Header Manager to the Thread Group
  • Add header Content-Type:multipart/form-data
  • Add header Accept:application/json
• Add a HTTP Request to the Thread Group
  • Set the method (POST), the path (/algorithm/BayesNet)
  • Check multipart/form-data
  • Add a File Upload with the file path, the parameter name (file) and the MIME Type (text/plain)
• Add a listener to the Thread Group
  e.g.: View Results Tree
• Save the Test Plan as a jmx file

Execute the Test Plan

• Launch the containers
  docker-compose up
• Execute the test
  jmeter -n -t [jmx file] -l [csv results file]

Profile the execution of scenarios with JProfiler

• Profile the execution using JProfiler
• Create a Test Plan using JMeter
• Execute the Test Plan
• Save a snapshot using JProfiler
• Repeat for each scenario (small, average, augmented, exceptional)