Skip to main content

Deployment Checklist

Used for checking and verifying the preparation before deployment, configuration during deployment, and functionality after deployment in the CoStrict backend service deployment process.

I. Pre-Deployment​

1.1 Models​

  • Chat Model: 4 * H20 or 4 * RTX4090
  • Code Review Model: 2 * H20 or 2 * RTX4090
  • Completion Model: 1 * H20 or 1 * RTX4090
  • Embedding Model: 0.5 * H20 or 0.5 * RTX4090
  • Rerank Model: 0.5 * H20 or 0.5 * RTX4090
  • Chat Model: GLM-4.5-FP8, GLM-4.5-106B-A12B-FP8
  • Code Review Model: Qwen2.5-Coder-32B-Instruct
  • Completion Model: DeepSeek-Coder-V2-Lite-Base
  • Embedding Model: gte-modernbert-baseRAG/Embedding
  • Rerank Model: gte-reranker-modernbert-baseRAG/Rerank

Note: Confirm and record the model NAMES, APIKEYs, and CONTEXT LENGTHS are accurate.

1.1.3 Checks​

  • GPU Resource Check

    # Provided by the model deployment party (AICP or User)

  • Model Interfaces

    • Chat Model: {CHAT_BASEURL}/v1/chat/completions interface
    • Code Review Model: {CODEREVIEW_BASEURL}/v1/chat/completions interface
    • Completion Model: {COMPLETION_BASEURL} interface
    • Embedding Model: {EMBEDDER_BASEURL} interface
    • Rerank Model: {RERANKER_BASEURL} interface

1.2 Backend Server​

1.2.1 Hardware Requirements​

  • CPU: Intel x64 architecture, minimum 16 cores
  • Memory: Minimum 32GB RAM
  • Storage: Minimum 512GB available storage space

1.2.2 Software Requirements​

  • Operating System: CentOS 7+ or Ubuntu 18.04+
  • Container Runtime: Docker 20.10+
  • Orchestration Tool: Docker Compose 2.0+
  • Git: Not required if obtaining Deployment Project offline

1.2.3 Checks​

  • Check CPU

    lscpu

  • Check Memory

    free -h

  • Check Storage

    df -h

  • Check Operating System

    cat /etc/os-release

  • Check Docker, Docker Compose

    docker --version
    systemctl status docker
    docker-compose --version

1.3 Backend Services​

1.3.1 Deployment Project​

Project URL: https://github.com/zgsm-ai/zgsm-backend-deploy

Project storage path: /opt/zgsm-backend-deploy (Assuming the directory is /opt/zgsm-backend-deploy, modify according to the actual project storage path, the same below)

1.3.2 Service Port Requirements​

Local port list: /opt/zgsm-backend-deploy/configure.sh

Online port list: https://github.com/zgsm-ai/zgsm-backend-deploy/blob/main/configure.sh

  • Port list: Lines 6~34

1.3.3 Checks​

  • Project Check

    ls /opt/zgsm-backend-deploy

  • Port Check

    sudo ss -tlnp | grep -E ':(9180|9080|9091|9092|9093|2382|6379|5432|5003|9081|5000|8765|5001|9090|3000|9200|8080|8001|9001|5173|9003|9004|9005|9006|9007|9008|9009|9010|7890|9011|8888|8889) '

II. During Deployment​

2.1 Startup Configuration​

2.1.1 Model Configuration​

Local model configuration list: /opt/zgsm-backend-deploy/configure.sh

  • Large model configuration: Lines 71~100

2.1.2 Backend Address Configuration​

Local backend address configuration list: /opt/zgsm-backend-deploy/configure.sh

  • Backend address configuration: Line 143

2.1.3 Checks​

  • Check if file exists

    ls -l /opt/zgsm-backend-deploy | grep configure.sh

  • View configuration file

    cat /opt/zgsm-backend-deploy/configure.sh

2.2 Backend Images​

Local image list: /opt/zgsm-backend-deploy/scripts/newest-images.list

Online image list: https://github.com/zgsm-ai/zgsm-backend-deploy/blob/main/scripts/newest-images.list

2.2.1 Checks​

  • Image Check

    cd /opt/zgsm-backend-deploy
    bash docker-download-images.sh

2.3 Execute Script​

Execution script: /opt/zgsm-backend-deploy/deploy.sh

2.3.1 Checks​

  • Check if file exists and has executable permissions

    ls -l /opt/zgsm-backend-deploy | grep deploy.sh

2.4 Service Configuration​

2.4.1 AI Gateway Configuration​

Local AI gateway configuration: /opt/zgsm-backend-deploy/docs/higress.zh-CN.md

2.4.2 Checks​

  • AI Gateway Configuration

    1. Access Higress page via browser
    http://{COSTRICT_BACKEND}:{PORT_HIGRESS_CONTROL}

    2. Check "AI Traffic Entry Management -> AI Service Provider Management" page

    3. Check "AI Traffic Entry Management -> AI Route Management" page

    4. Check "Plugin Configuration -> AI Quota Management" configuration

III. Post-Deployment​

3.1 Service Checks​

3.1.1 Backend Service Operation​

Local service image list: /opt/zgsm-backend-deploy/scripts/newest-images.list

Online service image list: https://github.com/zgsm-ai/zgsm-backend-deploy/blob/main/scripts/newest-images.list

3.1.2 Service Connectivity​

apisix gateway service probe interface: http://{COSTRICT_BACKEND}:{PORT_APISIX_ENTRY}/health

3.1.3 Checks​

  • Backend Service Operation Check

    docker ps

  • Backend Service Gateway Connectivity Check

    curl -v http://{COSTRICT_BACKEND}:{PORT_APISIX_ENTRY}/health

3.2 Functionality Testing​

3.2.1 Login​

Login Baseurl: {COSTRICT_BACKEND_BASEURL}

3.2.2 Agent Mode​

  • Ask
  • Code
  • Architect
  • Debug
  • Orchestrator

3.2.3 CodeReview​

  • Code block
  • File

3.2.4 Completion​

  • Code completion
  • Text completion

3.2.5 Checks​

  • Login Function Verification

  • Agent Model Function Verification

    # Code
    ```
    In the current project, add a test.go file and implement a quick sort algorithm function
    ```

    # Ask
    ```
    Read the current project and provide a description of the main functions
    ```

    # Architect
    ```
    Based on the content of the @test.go file, design a complete sorting algorithm
    ```

    # Debug
    ```
    Check the content in the @test.go file to determine if the sorting logic is in descending order
    ```

    # Orchestrator
    ```
    Check the content in the @test.go file:
    1. Determine if all sorting algorithms exist, if not, design and implement complete sorting algorithms;
    2. Additionally, determine if all sorting algorithms are implemented in descending order, if not, adjust the code logic to implement descending order and modify the corresponding design documents
    ```

  • CodeReview Function Verification

  • Completion Function Verification

3.3 Log Checks​

3.3.1 Service List​

Local service image list: /root/zgsm-backend-deploy/scripts/newest-images.list

Online service image list: https://github.com/zgsm-ai/zgsm-backend-deploy/blob/main/scripts/newest-images.list

3.3.2 Checks​

  • Service Log Check

    # Services: chat-rag,issue-manager,review-manager,review-checker,code-completion,codebase-embedder
    docker-compose logs [service_name]