indigodatacloud/deep-training-dashboard

DEEP-Hybrid-DataCloud训练仪表板

⚠️ 注意：这是从https://github.com/indigo-dc/orchestrator-dashboard%E5%88%86%E5%8F%89%E7%9A%84%E9%A1%B9%E7%9B%AE%EF%BC%8C%E6%AD%A3%E5%9C%A8%E5%AE%9A%E5%88%B6%E4%BB%A5%E9%80%82%E5%BA%94DEEP%E6%9C%8D%E5%8A%A1%E4%B8%8A%E7%9A%84ML/DL%E5%B7%A5%E4%BD%9C%E8%B4%9F%E8%BD%BD%E3%80%82%E7%9B%AE%E5%89%8D%E4%BB%8D%E5%9C%A8%E5%BC%80%E5%8F%91%E4%B8%AD%EF%BC%8C%E5%88%9D%E6%AD%A5%E7%89%88%E6%9C%AC%E5%B7%B2%E9%83%A8%E7%BD%B2%E5%9C%A8%E6%AD%A4%E5%A4%84%E3%80%82

核心功能

• IAM认证
• 显示用户的部署列表
• 查看部署详情及模板
• 删除部署
• 创建新部署

deep-training-dashboard是基于Flask微框架构建的Python应用；使用Flask-Dance实现Openid-Connect/OAuth2集成。Docker镜像采用Gunicorn作为WSGI HTTP服务器来运行Flask应用。

部署指南

步骤1：注册IAM客户端

在DEEP-IAM注册客户端，需满足以下条件：

• 重定向URI：https://<DASHBOARD_HOST>:<PORT>/login/iam/authorized
• 权限范围：openid、email、profile、offline_access
• 启用自省端点（introspection endpoint）

步骤2：克隆TOSCA模板仓库

获取仪表板所需的TOSCA模板集合：

bash
git clone https://github.com/indigo-dc/tosca-templates

步骤3：创建配置文件

在/app目录下创建config.json文件（参考示例），替换以下参数：

• IAM_CLIENT_ID：你的IAM客户端ID
• IAM_CLIENT_SECRET：你的IAM客户端密钥
• TOSCA_TEMPLATES_DIR：TOSCA模板目录路径
• （可选）GITHUB_SECRET：与GitHub Webhook密钥一致，用于验证更新请求

配置文件示例：

json
{
    "IAM_CLIENT_ID": "my_client_id",
    "IAM_CLIENT_SECRET": "my_client_secret",
    "IAM_BASE_URL": "https://iam.deep-hybrid-datacloud.eu",
    "ORCHESTRATOR_URL": "https://paas.cloud.cnaf.infn.it/orchestrator",
    "SLAM_URL": "https://paas.cloud.cnaf.infn.it:8443",
    "CMDB_URL": "http://paas.cloud.cnaf.infn.it/cmdb",
    "IM_URL": "https://paas.cloud.cnaf.infn.it/im",
    "MONITORING_URL": "https://deep-paas.cloud.ba.infn.it/monitoring-wrapper",
    "TOSCA_TEMPLATES_DIR": "../tosca-templates/deep-oc",
    "COMMON_TOSCAS": {
        "default": "deep-oc-marathon-webdav.yml",
        "minimal": "deep-oc-marathon-minimal.yml"
    },
    "MODULES_YML": "https://raw.githubusercontent.com/deephdc/deep-oc/master/MODULES.yml",
    "GITHUB_SECRET": "",
    "SUPPORT_EMAIL": "deep-support@listas.csic.es",
    "EXTERNAL_LINKS": [
        {
            "url": "https://marketplace.deep-hybrid-datacloud.eu",
            "menu_item_name": "DEEP Marketplace"
        },
        {
            "url": "https://docs.deep-hybrid-datacloud.eu",
            "menu_item_name": "Documentation"
        },
        {
            "url": "https://deep-hybrid-datacloud.eu",
            "menu_item_name": "DEEP-Hybrid-DataCloud project page"
        }
    ],
    "LOG_LEVEL": "info",
    "ENABLE_ADVANCED_MENU": "yes"
}

步骤4：启用HTTPS

仪表板需运行在HTTPS环境下，可选择以下两种方式：

直接启用HTTPS

需提供证书和密钥，并挂载到容器的/certs目录：

bash
docker run -d -p 443:5001 --name='deep-training-dashboard' \
           -e ENABLE_HTTPS=True \
           -v $PWD/cert.pem:/certs/cert.pem \
           -v $PWD/key.pem:/certs/key.pem \
           -v $PWD/config.json:/app/app/config.json \
           -v $PWD/tosca-templates:/opt/tosca-templates \
           docker.xuanyuan.run/indigodatacloud/deep-training-dashboard:latest

访问地址：https://<DASHBOARD_HOST>/

使用HTTPS代理

以Nginx为例，配置示例：

nginx
server {
      listen         80;
      server_name    YOUR_SERVER_NAME;
      return         301 https://$server_name$request_uri;
}

server {
    listen        443 ssl;
    server_name   YOUR_SERVER_NAME;
    access_log    /var/log/nginx/proxy-paas.access.log  combined;

    ssl on;
    ssl_protocols TLSv1 TLSv1.1 TLSv1.2;
    ssl_certificate           /etc/nginx/cert.pem;
    ssl_certificate_key       /etc/nginx/key.pem;
    ssl_trusted_certificate   /etc/nginx/trusted_ca_cert.pem;

    location / {
        proxy_pass http://127.0.0.1:5001/;
        proxy_set_header        X-Real-IP $remote_addr;
        proxy_set_header        X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header        X-Forwarded-Proto https;
        proxy_set_header        Host $http_host;
        proxy_redirect          http:// https://;
        proxy_buffering         off;
    }
}

运行容器：

bash
docker run -d -p 5001:5001 --name='deep-training-dashboard' \
           -v $PWD/config.json:/app/app/config.json \
           -v $PWD/tosca-templates:/opt/tosca-templates \
           docker.xuanyuan.run/indigodatacloud/deep-training-dashboard:latest

⚠️ 需更新IAM客户端的重定向URI为https://<PROXY_HOST>/login/iam/authorized，访问地址：https://<PROXY_HOST>/

自动更新仪表板

若需仪表板自动同步TOSCA模板或模块列表的更新，需配置GitHub Webhook：

• Payload URL：<dashboard_url>/reload
• Content type：application/json
• Secret：与config.json中的GITHUB_SECRET一致
• 事件类型：仅push事件
• 启用SSL（若使用HTTPS且证书有效）

性能调优

通过WORKERS环境变量调整Gunicorn worker数量，例如：

bash
docker run -d -p 5001:5001 -e WORKERS=2 ... indigodatacloud/deep-training-dashboard:latest

构建与运行镜像

bash
git clone https://github.com/indigo-dc/deep-training-dashboard.git
cd deep-training-dashboard
docker build -f docker/Dockerfile -t deep-training-dashboard .

运行镜像：

bash
docker run -d -p 5001:5001 -v $PWD/config.json:/app/app/config.json docker.xuanyuan.run/deep-training-dashboard

或直接使用DockerHub镜像：

bash
docker run -d -p 5001:5001 -v $PWD/config.json:/app/app/config.json docker.xuanyuan.run/indigodatacloud/deep-training-dashboard

开发环境搭建

bash
git clone https://github.com/indigo-dc/deep-training-dashboard.git
cd deep-training-dashboard
python3 -m venv venv
source venv/bin/activate
pip3 install -r requirements.txt

使用Flask启动：

bash
FLASK_app=orchdashboard flask run --host=0.0.0.0 --cert cert.pem --key privkey.pem --port 443

或使用Gunicorn：

bash
gunicorn --certfile=cert.pem --keyfile=key.pem --bind 0.0.0.0:443 orchdashboard:app --daemon

故障排除

SSL证书验证

若SLAM交互出现问题，需在config.json中指定SSL证书路径：

json
{
  ...
  "SLAM_URL": "https://indigo-slam.cloud.ba.infn.it:8443",
  "SLAM_CERT": "/path/to/certfile"
}

运行容器时需挂载证书文件：-v $PWD/certfile:/path/to/certfile

参考资料

• [***]