Files
cosma-qc/app/main.py
Silver Surfer b7d957c806 scaffold — FastAPI + SQLite + HTMX dashboard, ingest + dispatcher
- app/main.py : dashboard /, partials /partials/{jobs,monitor} (htmx polling)
- app/templates/ : index, jobs table, monitor card par worker
- app/static/style.css : thème sombre cohérent
- scripts/ingest.py : scan SSD d'acquisition, EXIF CreateDate → segments
  continus par (AUV, GoPro serial) avec seuil configurable
- scripts/dispatcher.py : polling queue, pick worker selon VRAM free,
  extraction ffmpeg + lingbot-map windowed --offload_to_cpu, progression DB
- DB : SQLite (acquisitions + jobs), lifecycle queued→extracting→running→done
- Workers par défaut : .87 (3060 12GB) + .84 (3090 24GB)

Contexte : QC terrain le jour-même (avant photogrammétrie à 30 jours),
plusieurs heures × 2 GoPros × 2-3 AUVs d'enregistrement à traiter en parallèle.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-21 09:52:41 +00:00

169 lines
6.0 KiB
Python

from __future__ import annotations
import asyncio
import json
import os
import sqlite3
from contextlib import asynccontextmanager, closing
from datetime import datetime
from pathlib import Path
from typing import Any
from fastapi import FastAPI, Form, HTTPException, Request
from fastapi.responses import HTMLResponse, JSONResponse
from fastapi.staticfiles import StaticFiles
from fastapi.templating import Jinja2Templates
DB_PATH = Path(os.environ.get("COSMA_QC_DB", "/var/lib/cosma-qc/jobs.db"))
WORKERS = json.loads(os.environ.get("COSMA_QC_WORKERS", json.dumps([
{"host": "192.168.0.87", "ssh_alias": "gpu", "gpu": "RTX 3060 12GB"},
{"host": "192.168.0.84", "ssh_alias": "cosma-vm","gpu": "RTX 3090 24GB"},
])))
STATUSES = ("queued", "extracting", "running", "done", "error")
def db() -> sqlite3.Connection:
DB_PATH.parent.mkdir(parents=True, exist_ok=True)
conn = sqlite3.connect(DB_PATH, isolation_level=None)
conn.execute("PRAGMA journal_mode=WAL")
conn.execute("PRAGMA foreign_keys=ON")
conn.row_factory = sqlite3.Row
return conn
def init_schema() -> None:
with closing(db()) as conn:
conn.executescript("""
CREATE TABLE IF NOT EXISTS acquisitions (
id INTEGER PRIMARY KEY,
name TEXT NOT NULL,
source_path TEXT NOT NULL,
created_at TEXT NOT NULL DEFAULT (datetime('now'))
);
CREATE TABLE IF NOT EXISTS jobs (
id INTEGER PRIMARY KEY,
acquisition_id INTEGER NOT NULL REFERENCES acquisitions(id) ON DELETE CASCADE,
auv TEXT NOT NULL,
gopro_serial TEXT NOT NULL,
segment_label TEXT NOT NULL,
video_paths TEXT NOT NULL,
frame_count INTEGER,
frames_dir TEXT,
status TEXT NOT NULL DEFAULT 'queued',
worker_host TEXT,
viser_url TEXT,
ply_path TEXT,
progress INTEGER NOT NULL DEFAULT 0,
log_tail TEXT,
error TEXT,
started_at TEXT,
finished_at TEXT,
created_at TEXT NOT NULL DEFAULT (datetime('now'))
);
CREATE INDEX IF NOT EXISTS jobs_status_idx ON jobs(status);
CREATE INDEX IF NOT EXISTS jobs_acq_idx ON jobs(acquisition_id);
""")
@asynccontextmanager
async def lifespan(_: FastAPI):
init_schema()
yield
app = FastAPI(title="cosma-qc", lifespan=lifespan)
templates = Jinja2Templates(directory=Path(__file__).parent / "templates")
app.mount("/static", StaticFiles(directory=Path(__file__).parent / "static"), name="static")
@app.get("/", response_class=HTMLResponse)
async def index(request: Request):
with closing(db()) as conn:
jobs = conn.execute("""
SELECT j.*, a.name AS acquisition_name
FROM jobs j
LEFT JOIN acquisitions a ON a.id = j.acquisition_id
ORDER BY j.created_at DESC
LIMIT 200
""").fetchall()
return templates.TemplateResponse("index.html", {
"request": request,
"jobs": jobs,
"workers": WORKERS,
})
@app.get("/api/jobs")
async def list_jobs():
with closing(db()) as conn:
rows = conn.execute("SELECT * FROM jobs ORDER BY created_at DESC LIMIT 500").fetchall()
return [dict(r) for r in rows]
@app.get("/partials/jobs", response_class=HTMLResponse)
async def partial_jobs(request: Request):
with closing(db()) as conn:
jobs = conn.execute("""
SELECT j.*, a.name AS acquisition_name
FROM jobs j
LEFT JOIN acquisitions a ON a.id = j.acquisition_id
ORDER BY j.created_at DESC
LIMIT 200
""").fetchall()
return templates.TemplateResponse("_jobs_table.html", {"request": request, "jobs": jobs})
@app.get("/partials/monitor", response_class=HTMLResponse)
async def partial_monitor(request: Request):
stats = await asyncio.gather(*[_worker_stats(w) for w in WORKERS])
return templates.TemplateResponse("_monitor.html", {"request": request, "workers": stats})
async def _worker_stats(worker: dict) -> dict:
alias = worker["ssh_alias"]
try:
proc = await asyncio.create_subprocess_exec(
"ssh", "-o", "ConnectTimeout=3", "-o", "BatchMode=yes", alias,
"nvidia-smi --query-gpu=memory.used,memory.total,utilization.gpu --format=csv,noheader,nounits && df -h / | tail -1",
stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE,
)
out, _ = await asyncio.wait_for(proc.communicate(), timeout=4)
text = out.decode().strip().splitlines()
gpu_line = text[0].split(",") if text else ["?", "?", "?"]
disk = text[1].split() if len(text) > 1 else ["?"] * 6
return {
**worker,
"online": True,
"vram_used_mib": int(gpu_line[0].strip()) if gpu_line[0].strip().isdigit() else None,
"vram_total_mib": int(gpu_line[1].strip()) if gpu_line[1].strip().isdigit() else None,
"gpu_util_pct": int(gpu_line[2].strip()) if gpu_line[2].strip().isdigit() else None,
"disk_used_pct": disk[4] if len(disk) > 4 else "?",
}
except Exception as e:
return {**worker, "online": False, "error": str(e)[:80]}
@app.post("/jobs/{job_id}/cancel")
async def cancel_job(job_id: int):
with closing(db()) as conn:
conn.execute(
"UPDATE jobs SET status='error', error='cancelled by user', finished_at=datetime('now') "
"WHERE id=? AND status IN ('queued','extracting','running')",
(job_id,),
)
return {"ok": True}
@app.post("/jobs/{job_id}/retry")
async def retry_job(job_id: int):
with closing(db()) as conn:
conn.execute(
"UPDATE jobs SET status='queued', error=NULL, progress=0, started_at=NULL, "
"finished_at=NULL, worker_host=NULL WHERE id=? AND status='error'",
(job_id,),
)
return {"ok": True}