Files
fuelish/docker-compose.yml
T
ml 7a91a37049
build / build (push) Successful in 3m4s
Add optional API key auth to the server
Deployments reachable outside a trusted LAN need some form of access
control. Set FUELISH_API_KEY and the server requires a matching
X-API-Key header on /jobs; unset, behavior is unchanged (no auth).
The Android app gets a matching Settings field. This only checks the
key - TLS termination is left to a reverse proxy in front of the
server, not handled here.
2026-10-04 11:41:55 +02:00

44 lines
1.6 KiB
YAML

# Runs the pipeline as a long-lived HTTP API for the Android app, on top
# of the same tooling-container image the numbered host scripts use.
# `docker compose up` alone is enough: the api service waits for Ollama
# and pulls the model itself (see tooling-container/src/server.py).
#
# The numbered host scripts (0_run_ollama.sh, 1/2/3_*.sh) and
# test_batch.sh are untouched and remain the way to inspect/test the
# pipeline stage by stage; this is the additive "serve it for real" path.
#
# Usage: docker compose up -d
# API reachable at http://<this-host>:8080 from any device on the LAN.
#
# For a deployment reachable outside a trusted LAN, copy .env.example to
# .env and set FUELISH_API_KEY - every /jobs request then needs a
# matching X-API-Key header (see tooling-container/src/server.py).
# Unset (the default), the server has no auth, same as before. This
# server does not terminate TLS itself - put a reverse proxy in front of
# it for anything public.
services:
ollama:
image: ollama/ollama
volumes:
- fuelish-ollama-data:/root/.ollama
restart: unless-stopped
api:
build:
context: ./tooling-container
restart: unless-stopped
depends_on:
- ollama
environment:
OLLAMA_URL: http://ollama:11434/api/chat
OLLAMA_BASE_URL: http://ollama:11434
OLLAMA_MODEL: qwen2.5:3b
FUELISH_API_KEY: ${FUELISH_API_KEY:-}
command: ["uvicorn", "server:app", "--app-dir", "/opt/app/src", "--host", "0.0.0.0", "--port", "8000"]
ports:
- "8080:8000"
volumes:
fuelish-ollama-data:
name: fuelish-ollama-data