Source Code Deploy Guide

This document describes how to deploy each module through source code compilation. Assume that it is deployed according to the following architecture:

There are three main modules that need to be compiled and deployed:

info

From v1.11.0 and later versions, the API service will integrate the UI service. After starting the API service, the root path to access the API service is the FeatureProbe UI page. According to the requirements of the deployment architecture, you can choose whether to deploy UI services separately.

Sample machine	Deployment module	Port
10.100.1.1	FeatureProbe API	4008
10.100.1.1	FeatureProbe UI	4009(Nginx)(Optional)
10.100.1.2	FeatureProbe Server	4007

Create database

1. Environment preparation

   • MySQL 5.7+

2. Create the feature_probe database:

   CREATE DATABASE feature_probe;

note

There is no need to manually create table structures. All tables and initialization data are automatically created when the API service is started for the first time. If you want to import DML/DDL manually, please refer to the Database Setup document.

Compile and deploy API service

Compilation steps

1. Compilation environment

   • JDK 1.8+
   • Maven 2.0+ : how to install

2. Obtain the source code and compile the deployment package:

git clone https://gitee.com/FeatureProbe/FeatureProbe.git
cd FeatureProbe/api
mvn clean package

After the compilation is completed, a jar deployment file named after the version will be generated in the current directory, such as target/api-1.1.0.jar.

Deployment steps

1. Deployment environment

   • JDK 1.8+

2. Place api-1.1.0.jar in the deployment server, fill in the database link configuration, and start it with 4008 port:

   caution

   The database-related information in the following script needs to be replaced with the actual database information you use.

   java -jar api-1.1.0.jar --server.port=4008 \
        --spring.datasource.jdbc-url=jdbc:mysql://{MYSQL_DATABASE_IP}:{MYSQL_PORT}/feature_probe \  # 数据库 IP/端口和库名
        --spring.datasource.username={MYSQL_USERNAME} \
        --spring.datasource.password={MYSQL_PASSWORD} 
   info

   For a more detailed description of the startup parameters of the API service, see FeatureProbe API parameters

3. Check whether the service is running successfully

   Run on the deployment machine:

   curl "http://localhost:4008/actuator/health"

   If the following information is displayed, the startup is successful:

   {
       status: "UP"
   }

Compile and deploy Server service

Compilation steps

1. Environment preparation

   • Rust

     • Official website installation

       $ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

     • Or, install from domestic mirror

       1. Modify ~/.zshrc or ~/.bashrc:
       ~~~bash
       export RUSTUP_DIST_SERVER="https://rsproxy.cn"
       export RUSTUP_UPDATE_ROOT="https://rsproxy.cn/rustup"
       ~~~
       2. Install
       ~~~bash
       $ curl --proto '=https' --tlsv1.2 -sSf https://rsproxy.cn/rustup-init.sh | sh
       ~~~
       caution

       Pay attention to the prompt on the last line after installation, follow the prompt command to reload the environment variable, and the installation will take effect.

   • Configure dependant download mirror, create file: ~/.cargo/config

     [source.crates-io]
     # To use sparse index, change 'rsproxy' to 'rsproxy-sparse'
     replace-with = 'rsproxy'
     
     [source.rsproxy]
     registry = "https://rsproxy.cn/crates.io-index"
     [source.rsproxy-sparse]
     registry = "sparse+https://rsproxy.cn/index/"
     
     [registries.rsproxy]
     index = "https://rsproxy.cn/crates.io-index"
     
     [net]
     git-fetch-with-cli = true

2. Get the source code and compile the deployment package

   git clone https://github.com/FeatureProbe/FeatureProbe.git

   Compile in the source code directory:

   cd server
   cargo build --release --verbose

   After compiling, an executable binary file will be generated in the target/release/ directory:

   $ ls target/release/ 
   feature_probe_server

Deployment steps

1. Environment preparation

   • none

2. Put the generated feature_probe_server file on the server and create the startup script start-feature-probe-server.sh:

   caution

   In actual use, you need to modify the ip and port in the script below to the ip and port of the actual deployment environment of your API service.

   #/bin/bash
   
   export FP_SERVER_PORT=4007 # FeatureProbe Server port
   export FP_TOGGLES_URL=http://10.100.1.1:4008/internal/server/toggles # FeatureProbe API IP and port
   export FP_EVENTS_URL=http://10.100.1.1:4008/internal/server/events
   export FP_KEYS_URL=http://10.100.1.1:4008/internal/server/sdk_keys
   export FP_REFRESH_SECONDS=1
   
   ./feature_probe_server
   info

   For a more detailed description of the startup parameters of the Server service, see FeatureProbe Server parameters

3. Execute the startup script to run the service: sh ./start-feature-probe-server.sh

4. Check whether the service is running successfully, and run it on the deployment machine:

   curl "http://localhost:4007/"

   If the following information is displayed, the startup is successful:

   <h1>Feature Probe Server</h1>

Compile and deploy UI service

Compilation steps

1. Environment preparation

   • Node.js 16.13+ : Download
   info

   It is recommended to switch to the npm China mirror station in China: npm config set registry https://registry.npmmirror.com/

2. Obtain the source code and compile a deployable static file:

   git clone https://gitee.com/FeatureProbe/FeatureProbe.git
   cd FeatureProbe/ui
   npm install
   npm run build

   After compiling, deployable static files will be generated in the build directory. As follows:

   $ ls build
   asset-manifest.json favicon.ico index.html static/

3. [Optional] Configure output path:

   For production environment deployment, there are two options for domain name configuration:

   • Separately deploy the FeatureProbe service as an independent domain name, for example: directly access the FeatureProbe entry page through https://your.domain.com
   • Share a domain name with several services, and distinguish services in the form of path (/path), for example: access the FeatureProbe entry page through https://your.domain.com/featureprobe

   At present, we support the first method by default. If you deploy in the second method, you need to make some changes to the source code:

   (1) Open craco.config.js file

   cd ui
   vi craco.config.js

   (2) On the basis of the original configuration, add the publicPath field in the webpack - configure - output object, for example, set the value to: /featureprobe/

   module.exports = {
      webpack: {
         configure: {
            output: {
               publicPath: '/featureprobe/'
            },
         }
      }
   };

   (3) After executing the yarn build command, view the index.html file through vi build/index.html, you can find that the static resource js and css files in the file have been added with the prefix configured in publicPath

   

   (4) In order to speed up access, you can choose to upload static resource (js, css, etc.) files to the CDN server, and you can also configure publicPath as the address of the CDN during the compilation phase:

   module.exports = {
      webpack: {
         configure: {
            output: {
               publicPath: 'https://cdn.domain.com/'
            },
         }
      }
   };

   The static resources (js, css, etc.) in the compiled html are all paths carrying the CDN address:

   

Deployment steps

1. Environment preparation

   • Nginx

2. Copy all the files and folders under build generated by compilation to the /usr/share/nginx/html/ directory.

Configure Nginx

1. Create Nginx configuration: /etc/nginx/conf.d/feature_probe.conf

   caution

   In actual use, you need to modify the ip and port in the script below to the ip and port of the actual deployment environment of the service.

   upstream featureProbeAPI {
       server 10.100.1.1:4008; # FeatureProbe API IP and port
   }
   
   
   upstream featureProbeServer {
       server 10.100.1.2:4007; # FeatureProbe Server IP and port
   }
   
   server {
       listen 4009; # UI port
   
       location / {
           index index.html index.htm;
           root /usr/share/nginx/html; # UI static file directory
           try_files $uri /index.html;
       }
   
       location /api { # Unified forwarding to featureProbeAPI service when accessing /api
           proxy_set_header X-Real-IP $remote_addr;
           proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
           proxy_set_header X-NginX-Proxy true;
           proxy_pass http://featureProbeAPI/api;
           proxy_ssl_session_reuse off;
           proxy_set_header Host $http_host;
           proxy_cache_bypass $http_upgrade;
           proxy_redirect off;
       }
   
       location /server/api { # Unified forwarding to featureProbeServer service when accessing /server/api
           proxy_set_header X-Real-IP $remote_addr;
           proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
           proxy_set_header X-NginX-Proxy true;
           proxy_pass http://featureProbeServer/api;
           proxy_ssl_session_reuse off;
           proxy_set_header Host $http_host;
           proxy_cache_bypass $http_upgrade;
           proxy_redirect off;
       }
   }
   [optional]

   In actual use, if you configure a custom output path, for example: configure publicPath to /featureprobe/, you need to change location / in the above nginx configuration to location /featureprobe/ to be correct matches static files such as html, js and css.

   location /featureprobe/ {
      index index.html index.htm;
      root /usr/share/nginx/html; # UI static file directory
      try_files $uri /index.html;
   }

2. Execute reload nginx configuration to make the above configuration take effect:

   nginx -s reload

Service verification

Platform

caution

In actual use, you need to modify the following ip and port to the ip and port of the actual deployment environment of your ngnix service.

Visit http://10.100.1.1:4009 in the browser and use the following account and password to log in to verify whether the deployment is successful:

• username: admin
• password: Pass1234

Server Side SDK Access

The server remote_uri address used by the SDK connection is the ngnix machine address and service port configured above, plus the path /server, for example: http://10.100.1.1:4009/server in the above example.

For specific steps, please refer to Rollout Tutorial