M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

# NornicDB AMD64 CUDA + Heimdall (Cognitive Guardian) # Self-contained build with local embedding AND Heimdall SLM support # # This image includes: # - BGE-M3 embedding model for vector search # - Qwen2.5-0.5B-Instruct (Q4_K_M) for Heimdall (cognitive database guardian) # # Build: # docker build -f docker/Dockerfile.amd64-cuda-heimdall -t timothyswt/nornicdb-amd64-cuda-bge-heimdall . # # Run: # docker run --gpus all -p 7474:7474 -p 7687:7687 -v nornicdb-data:/data timothyswt/nornicdb-amd64-cuda-bge-heimdall # # This is a "batteries included" deployment - no additional model downloads required. # Heimdall provides: anomaly detection, runtime diagnosis, and memory curation via Bifrost chat. ARG LLAMA_CUDA_IMAGE=timothyswt/llama-cuda-libs:7285 # ============================================================================= # Stage 1: UI # ============================================================================= FROM node:20-alpine AS ui WORKDIR /ui COPY ui/package*.json ./ RUN npm ci 2>/dev/null || npm install --legacy-peer-deps COPY ui/ . RUN npm run build # ============================================================================= # Stage 2: Pre-built CUDA libs # ============================================================================= FROM ${LLAMA_CUDA_IMAGE} AS llama # ============================================================================= # Stage 3: Go build # ============================================================================= FROM nvidia/cuda:12.6.3-devel-ubuntu22.04 AS builder ENV GO_VERSION=1.25.5 RUN apt-get update && apt-get install -y wget git gcc g++ libgomp1 && \ wget -q https://go.dev/dl/go${GO_VERSION}.linux-amd64.tar.gz && \ tar -C /usr/local -xzf go${GO_VERSION}.linux-amd64.tar.gz && rm go*.tar.gz ENV PATH="/usr/local/go/bin:${PATH}" CUDA_HOME=/usr/local/cuda WORKDIR /build # Copy llama artifacts COPY --from=llama /output/lib/*.a /build/lib/llama/ COPY --from=llama /output/include/*.h /build/lib/llama/ # Go dependencies COPY go.mod go.sum ./ RUN go mod download # Source + UI COPY . . COPY --from=ui /ui/dist ./ui/dist # Build with CUDA + localllm + heimdall RUN echo "Building with CUDA + localllm + heimdall..." && \ CGO_ENABLED=1 go build -tags "cuda localllm heimdall" \ -ldflags="-s -w -X main.buildTime=$(date -u +%Y%m%d-%H%M%S)" \ -o nornicdb ./cmd/nornicdb # Build APOC plugin RUN echo "Building APOC plugin..." && \ mkdir -p apoc/built-plugins && \ cd apoc/plugin-src/apoc && go build -buildmode=plugin -o ../../../apoc/built-plugins/apoc.so apoc_plugin.go && \ echo "✓ Built plugin:" && ls -lh /build/apoc/built-plugins/*.so # ============================================================================= # Stage 4: Runtime # ============================================================================= FROM nvidia/cuda:12.6.3-runtime-ubuntu22.04 WORKDIR /app RUN apt-get update && apt-get install -y --no-install-recommends \ ca-certificates tzdata wget libgomp1 && rm -rf /var/lib/apt/lists/* && \ mkdir -p /data /app/models COPY --from=builder /build/nornicdb /app/ COPY --from=builder /build/apoc/built-plugins /app/plugins/ COPY docker/entrypoint.sh /app/ RUN chmod +x /app/entrypoint.sh # Embed both models: BGE-M3 for embeddings + Qwen for Heimdall RUN --mount=type=bind,source=models,target=/models,ro \ echo "Embedding BGE-M3 model..." && \ if [ -f /models/bge-m3.gguf ]; then \ cp /models/bge-m3.gguf /app/models/ && \ echo "✓ Embedded bge-m3.gguf ($(du -h /app/models/bge-m3.gguf | cut -f1))"; \ else \ echo "ERROR: models/bge-m3.gguf not found" && exit 1; \ fi && \ echo "Embedding Heimdall SLM model..." && \ if [ -f /models/qwen2.5-0.5b-instruct-q4_k_m.gguf ]; then \ cp /models/qwen2.5-0.5b-instruct-q4_k_m.gguf /app/models/ && \ echo "✓ Embedded qwen2.5-0.5b-instruct-q4_k_m.gguf ($(du -h /app/models/qwen2.5-0.5b-instruct-q4_k_m.gguf | cut -f1))"; \ else \ echo "ERROR: models/qwen2.5-0.5b-instruct-q4_k_m.gguf not found" && exit 1; \ fi && \ echo "✓ Heimdall cognitive features enabled" EXPOSE 7474 7687 HEALTHCHECK --interval=30s --timeout=10s --start-period=10s --retries=3 \ CMD wget --spider -q http://localhost:7474/health || exit 1 ENV NORNICDB_DATA_DIR=/data \ NORNICDB_HTTP_PORT=7474 \ NORNICDB_BOLT_PORT=7687 \ NORNICDB_EMBEDDING_PROVIDER=local \ NORNICDB_EMBEDDING_MODEL=bge-m3 \ NORNICDB_EMBEDDING_DIMENSIONS=1024 \ NORNICDB_MODELS_DIR=/app/models \ NORNICDB_EMBEDDING_GPU_LAYERS=-1 \ NORNICDB_NO_AUTH=true \ NORNICDB_GPU_ENABLED=true \ NORNICDB_PLUGINS_DIR=/app/plugins \ NORNICDB_HEIMDALL_ENABLED=true \ NORNICDB_HEIMDALL_MODEL=qwen2.5-0.5b-instruct-q4_k_m \ NVIDIA_VISIBLE_DEVICES=all \ NVIDIA_DRIVER_CAPABILITIES=compute,utility \ LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/nvidia/lib64:${LD_LIBRARY_PATH} ENTRYPOINT ["/app/entrypoint.sh"]