Skip to content

PROJECT_STATUS.md

Latest commit

 

History

History
97 lines (78 loc) · 3.01 KB

PROJECT_STATUS.md

File metadata and controls

97 lines (78 loc) · 3.01 KB

Vyse Project Status

Last verified: 2026-04-25 on Pop!_OS with Docker Compose and local Ollama.

Working

  • Docker Compose starts the full stack:
    • dashboard on http://localhost:3000
    • public gateway on http://localhost:8080
    • admin gateway on http://localhost:8081
    • engine gRPC on localhost:50051
    • PostgreSQL on host port 55432
    • Redis on host port 6379
  • Authenticated inference works through the full runtime path:
    • client curl
    • Go gateway
    • Rust engine
    • host Ollama via host.docker.internal:11434
    • tinyllama
  • Public and admin health endpoints return 200 OK.
  • Engine connects to Redis and PostgreSQL and runs migrations.
  • Gateway no longer panics on gRPC calls; Go protobuf stubs have been regenerated from proto/vyse.proto.
  • Engine and gateway read the Docker environment variables used by docker-compose.yml.
  • Dashboard production build succeeds.

Verified Commands

cd /home/sarvadubey/Desktop/Projects/Vyse

cargo test --manifest-path engine/Cargo.toml
cd gateway && go test ./...
cd ../dashboard && npm run build
cd ..

docker compose config
docker compose up --build -d
curl http://localhost:8080/health
curl -i \
  -H 'Content-Type: application/json' \
  -H 'X-Vyse-Key: abcdef0123456789abcdef0123456789' \
  -H 'X-Vyse-Session: smoke-test' \
  -d '{"prompt":"Say hello in one short sentence."}' \
  http://localhost:8080/api/infer

Local Host Requirements

Ollama must listen on all interfaces so Docker containers can reach it through host.docker.internal:

ollama pull tinyllama

sudo mkdir -p /etc/systemd/system/ollama.service.d
printf '[Service]\nEnvironment="OLLAMA_HOST=0.0.0.0:11434"\n' | \
  sudo tee /etc/systemd/system/ollama.service.d/override.conf
sudo systemctl daemon-reload
sudo systemctl restart ollama
ss -ltnp | grep 11434

Expected listener:

0.0.0.0:11434

Pending For Full Completion

  • Add engine/models/all-MiniLM-L6-v2.onnx.
  • Add engine/models/nli-MiniLM2-L6-H768.onnx.
  • Ensure the engine Docker runtime includes or mounts engine/models.
  • Replace development secrets in .env:
    • VYSE_AUTH_JWT_SECRET
    • VYSE_AUTH_ADMIN_PASSWORD
    • VYSE_AUTH_API_KEY
  • Decide Rekor mode:
    • keep disabled for local-only development, or
    • configure a real Rekor/Sigstore endpoint for Tier 3 audit entries.
  • Add production transport security:
    • TLS at the public edge
    • private network or TLS between gateway and engine
  • Add container healthchecks for the gateway, engine, and dashboard.
  • Optimize Dockerfiles for dependency-layer caching.
  • Add integration tests that start the Compose stack and verify /health plus /api/infer.

Current Degraded Health Reason

The current degraded status is caused by missing ONNX model files:

embedding model not found at ./models/all-MiniLM-L6-v2.onnx
intent model not found at ./models/nli-MiniLM2-L6-H768.onnx
starting with degraded ML fallbacks

The fallback path is functional for local runtime verification, but full ML-backed scoring requires those model files.