neops-remote-lab 0.0.1b2

No project description provided

Remote Lab Manager

Run Netlab topologies on a remote host while keeping your pytest suite local. The service exposes a small REST API that schedules exclusive sessions in a FIFO queue so your CI jobs or multiple developers can share the same infrastructure safely.

---

✨ Key Points

• One-lab rule – only one Netlab topology may run per host; the manager enforces this with a queue and automatic reference counting.
• Zero-config client – set a single environment variable and your existing fixtures will transparently switch to remote mode.
• Stateless HTTP API – every request is authenticated via an X-Session-ID header issued when the session is created.
• Python client available – import RemoteLabClient for programmatic use.

---

You to author tests that consume these labs? See the Testing Framework guide: /development/testing-framework/

For secure reachability to the Remote Lab subnet(s) using Tailscale clients managed by a self‑hosted control plane, see: Headscale + Headplane with Docker Compose.

🚀 Quick-Start

1. Start the Server

Native Python

# Install deps (inside a venv)
poetry install --extras remote-lab  # includes FastAPI, Uvicorn, etc.

# Run the service
poetry run remote_lab --host 0.0.0.0 --port 8000 --log-level info

2. Configure Your Tests

export REMOTE_LAB_URL=http://<host>:8000

# Hetzner neops-labs VM:
export REMOTE_LAB_URL=http://91.99.184.46:8000 
# Optional: put this into a .env file – test suite auto-loads it via python-dotenv

3. Run pytest as usual

pytest tests/function_blocks/

If REMOTE_LAB_URL is set the fixtures automatically use the remote server; otherwise they fall back to local netlab execution.

---

🔌 REST API

Method & Path	Purpose	Notes
POST /session	Create a new queue entry	Returns 201 with session_id & current position
GET /session/{id}	Poll session state	status: waiting/active, queue position
GET /active-session	Get active session details	Returns 200 with session_id, status and position
DELETE /session/{id}	End a session prematurely	Frees lab if active, returns 204
POST /session/heartbeat	Keep-alive	Must include X-Session-ID header, returns 204
POST /lab	Upload topology & acquire lab	multipart/form-data; `reuse=true
GET /lab	Lab status & device list	Only valid for active sessions
GET /lab/devices	Shortcut to device list	–
POST /lab/release	Decrement ref-count	If it drops to zero the lab becomes idle
DELETE /lab?force=true	Destroy lab	202 accepted; force=false fails if busy
GET /healthz	Liveness check	204 No Content

⚠️ All /lab* endpoints require the X-Session-ID header of an active session. Non-active sessions receive 423 Locked.

ℹ️ A debug-only endpoint GET /debug/health returns rich server stats (uptime, queue length, etc.) and is useful during development.

---

🛠️ Example cURL Session

# 1) Create session
SESSION=$(curl -s -X POST http://localhost:8000/session | jq -r .session_id)

# 2) Wait until it becomes ACTIVE (simplified polling)
while true; do
  STATUS=$(curl -s http://localhost:8000/session/$SESSION | jq -r .status)
  [[ $STATUS == "active" ]] && break
  sleep 2
done

# 3) Upload topology & acquire lab
curl -X POST http://localhost:8000/lab \
     -H "X-Session-ID: $SESSION" \
     -F "topology=@tests/topologies/simple_frr.yml" \
     -F "reuse=true"
# Optionally attach supporting files (repeatable)
#    -F "extra_files=@path/to/vars.yml" -F "extra_files=@path/to/your_special_config.yml"

# 4) Release when finished
curl -X POST http://localhost:8000/lab/release -H "X-Session-ID: $SESSION"

# 5) End session
curl -X DELETE http://localhost:8000/session/$SESSION

---

⚙️ Environment Variables

Variable	Description	Default
REMOTE_LAB_URL	Base URL used by client and fixtures	–

---

🧹 House-Keeping & Timeouts

• Waiting sessions – dropped after 600 s without a heartbeat.
• Active sessions – deemed stale after 300 s of silence; the lab is cleaned up and the next session in queue is promoted.
• Cleanup cadence – adaptive background task: ~5 s when busy, ~15 s with a single active session, ~30 s when idle.

Constants are defined in neops_worker_sdk/testing/remote_lab/server.py.

---

🪵 Logging

The server emits structured logs:

2024-05-27 12:34:56 | INFO     | remote-lab-server | sid=24f... topo=simple_frr.yml | Created session

Use --log-level debug or the --debug flag when starting the service to see queue promotions and Netlab command output. The --debug flag also enables streaming of Netlab output via NEOPS_NETLAB_STREAM_OUTPUT=1. You can override logging with --log-config <yaml>; see neops_worker_sdk/testing/remote_lab/logging_config.yaml for the default.

---

🧪 Tests

• Remote lab API: tests/testing/remote_lab/test_server.py
  • Covers queueing/promotion, heartbeats, active-session, acquire/release/destroy, status codes (400/409/423/202/204), device listing, and extra_files directory preservation. Uses a stubbed LabManager; no Netlab required.
• Fixture selection: tests/testing/netlab/test_netlab_fixture_logic.py
  • Verifies create_netlab_fixture local vs remote behavior, REMOTE_LAB_URL auto-selection, and conversion to NetlabDevice.
• Harness: tests/conftest.py
  • Loads .env, defines example fixtures, and adds handy pytest markers.

pytest tests/testing/remote_lab/test_server.py
pytest tests/testing/netlab/test_netlab_fixture_logic.py
pytest -m testing # Run all tests with "testing" marker

---

❓ Troubleshooting

Symptom	Checklist
Server won’t start	netlab --version, correct module path
Tests hang in queue	Port 8000 reachable? Heartbeats sent? Check server logs
Containers unreachable	Using network_mode: host? Firewall rules?
Lab stuck busy	Someone forgot to release? Use DELETE /lab?force=true

---

📚 Interactive Docs

Browse http://<host>:8000/docs for an auto-generated, interactive OpenAPI UI and experiment with the endpoints directly.