Installation and Configuration

Installation Path

The default installation path for PlantPulse AI is as follows:

/home/kopens/plantpulse-ai/

Directory Structure

plantpulse-ai/
├── bin/                    # 오케스트레이션 스크립트
│   ├── setup.sh           # 초기 설치 (의존성, 모델 등)
│   ├── start.sh           # 전체 시스템 시작
│   ├── stop.sh            # 전체 시스템 종료
│   ├── status.sh          # 서비스 상태 확인
│   ├── start-daemon.sh    # 백그라운드 데몬 시작
│   ├── clean.sh           # 로그/임시파일 정리
│   └── log-viewer.sh      # 통합 로그 뷰어
├── template/
│   ├── env.sh             # 환경 변수 설정 (핵심 설정 파일)
│   └── proxy.yaml         # LiteLLM 프록시 설정
├── db/                     # 데이터베이스
│   ├── postgres/          # PostgreSQL
│   ├── cassandra/         # Apache Cassandra
│   ├── neo4j/             # Neo4j
│   ├── qdrant/            # Qdrant 벡터 DB
│   ├── valkey/            # Valkey (Redis)
│   └── minio/             # MinIO (S3)
├── lib/                    # AI 서비스
│   └── litellm/           # LLM 프록시 서버 (채팅 LLM 게이트웨이)
├── web/                    # AI Chat Web (내장 에이전트)
│   ├── app/               # WAR 애플리케이션
│   ├── config/            # application.properties
│   ├── bin/               # 시작/종료 스크립트
│   └── logs/              # 웹 서비스 로그
├── timeseries/             # TimeSeries-Insight
│   ├── app/               # FastAPI 애플리케이션
│   ├── bin/               # 시작/종료 스크립트
│   ├── frontend/          # Next.js 대시보드
│   ├── tests/             # 테스트 코드
│   └── logs/              # 시계열 로그
├── logs/                   # 중앙 집중식 로그
└── README.md               # 프로젝트 설명

info

With the 2026-07 restructuring, the MCP and ontology tools were moved to the platform's integrated MCP (server-web /api/v5/mcp), and RAG was replaced by the LightRAG Docker compose service. The legacy mcp/, ontology/, rag/, and component/ (Langflow components) directories, along with Langflow, Phoenix, and Jupyter, are no longer used.

---

Initial Installation

Step 1. Configure Environment Variables

Before installation, edit the template/env.sh file to match your environment:

vi /home/kopens/plantpulse-ai/template/env.sh

Key configuration items:

Variable	Description	Default
PP_SERVER_HOST	PlantPulse IIoT server IP	100.68.69.41
PP_SERVER_PORT	PlantPulse IIoT server port	7443
PP_HOME	Home directory	/home/kopens
PP_DATA_DIR	Data storage path	/data1/pp-data
PP_TEMP_DIR	Temporary file path	/data1/pp-temp

Step 2. Install Dependencies

cd /home/kopens/plantpulse-ai/bin
./setup.sh

setup.sh installs the following in order:

1. Python libraries
2. LiteLLM proxy (chat LLM gateway)
3. LightRAG service (Docker compose)
4. TimeSeries analysis models (IBM Granite TSPulse, TTM)

info

Installation logs are written to logs/setup.log.

Step 3. Start the System

# 포그라운드 실행
./start.sh

# 또는 백그라운드 데몬으로 실행
./start-daemon.sh

Step 4. Check Status

./status.sh

When running normally, all services are shown in the [OK] state.

---

Startup Sequence

start.sh starts services in the following order:

Stage 1: Databases

Order	Service	Port	Wait Time
1	Valkey (Redis)	6379	3 s
2	PostgreSQL	5432	3 s
3	Cassandra	9042	20 s
4	MinIO	9000/9001	1 s
5	Qdrant	6333	1 s
6	Neo4j	7474/7687	1 s

The process waits until all database ports respond before proceeding to the next stage.

Stage 2: Cassandra Initialization

Creates the Cassandra keyspaces (on first run).

Stage 3: AI Services

Order	Service	Description
1	LiteLLM	LLM proxy (chat LLM gateway)
2	LightRAG	Document search server
3	TimeSeries	Anomaly detection/forecasting service

info

The MCP and ontology tools are provided by the platform's integrated MCP (server-web /api/v5/mcp), so they are not started separately in the AI stack.

Stage 4: Web Server

Starts the AI Chat Web service. Chat and agent orchestration are built into this service.

warning

On shutdown, services are stopped in the reverse order of startup: services → databases.