refactor(A1): remove Flamenco, simplify render pipeline to Celery-only

- Remove flamenco-manager and flamenco-worker from docker-compose.yml
- Delete flamenco_client.py, flamenco_tasks.py, docker_scaler.py
- Simplify render_dispatcher.py to Celery-only (removes ~300 lines)
- Remove Flamenco beat schedule from celery_app.py
- Clean admin.py: remove flamenco settings, endpoints, threejs validation
- Clean orders.py cancel-render: Celery revoke only
- Clean worker.py: remove flamenco_job_id from activity response
- Migration 032: cancel lingering flamenco jobs, remove flamenco settings
- PLAN.md: mark all decisions confirmed, status IN UMSETZUNG

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-03-06 15:38:37 +01:00
parent 552922eb8a
commit 1d6864fb64
13 changed files with 1524 additions and 1151 deletions
-177
View File
@@ -1,177 +0,0 @@
"""Scale Flamenco worker containers via the Docker socket.
Uses the Docker Python SDK (docker>=6.1.0) to list, start, and stop containers.
Requires /var/run/docker.sock to be mounted into the backend container.
"""
import os
import logging
log = logging.getLogger(__name__)
COMPOSE_PROJECT = os.getenv("COMPOSE_PROJECT_NAME", "schaefflerautomat")
SERVICE_NAME = "flamenco-worker"
def _get_client():
import docker
return docker.from_env()
def get_worker_containers(client=None):
"""Return all flamenco-worker containers (running + stopped) sorted by name."""
if client is None:
client = _get_client()
return sorted(
client.containers.list(
all=True,
filters={
"label": [
f"com.docker.compose.project={COMPOSE_PROJECT}",
f"com.docker.compose.service={SERVICE_NAME}",
]
},
),
key=lambda c: c.name,
)
def get_running_worker_count(client=None) -> int:
"""Return how many flamenco-worker containers are currently running."""
try:
if client is None:
client = _get_client()
containers = get_worker_containers(client)
return sum(1 for c in containers if c.status == "running")
except Exception as exc:
log.warning("docker_scaler: could not read worker count: %s", exc)
return -1
def scale_workers(target: int) -> dict:
"""Scale flamenco-worker containers to *target* count.
Returns a dict with keys:
previous containers running before
current containers running after
delta change (negative = stopped, positive = started)
message human-readable summary
"""
import docker
from docker.types import Mount
client = _get_client()
all_workers = get_worker_containers(client)
running = [c for c in all_workers if c.status == "running"]
previous = len(running)
if target == previous:
return {"previous": previous, "current": previous, "delta": 0,
"message": f"Already at {previous} worker(s) — no change"}
# ── Scale down ────────────────────────────────────────────────────────────
if target < previous:
# Stop highest-numbered containers first to minimise disruption
to_stop = sorted(running, key=lambda c: c.name, reverse=True)[: previous - target]
for c in to_stop:
log.info("docker_scaler: stopping %s", c.name)
c.stop(timeout=20)
c.remove()
return {
"previous": previous,
"current": target,
"delta": target - previous,
"message": f"Stopped {len(to_stop)} worker(s): {[c.name for c in to_stop]}",
}
# ── Scale up ──────────────────────────────────────────────────────────────
template = running[0] if running else (all_workers[0] if all_workers else None)
if template is None:
raise RuntimeError(
"No existing flamenco-worker container found to clone configuration from. "
"Ensure at least one worker container exists (even if stopped)."
)
attrs = template.attrs
image = attrs["Config"]["Image"]
env = attrs["Config"].get("Env") or []
# Reconstruct mounts from the template container
mounts = []
for m in (attrs.get("Mounts") or []):
mount_type = m.get("Type", "bind")
source = m.get("Name", "") if mount_type == "volume" else m.get("Source", "")
mounts.append(
Mount(
target=m["Destination"],
source=source,
type=mount_type,
read_only=not m.get("RW", True),
)
)
# Reconstruct GPU device requests (nvidia)
device_requests = None
raw_dr = (attrs.get("HostConfig") or {}).get("DeviceRequests") or []
if raw_dr:
device_requests = []
for dr in raw_dr:
device_requests.append(
docker.types.DeviceRequest(
driver=dr.get("Driver", ""),
count=dr.get("Count", -1),
device_ids=dr.get("DeviceIDs") or [],
capabilities=dr.get("Capabilities") or [],
options=dr.get("Options") or {},
)
)
# Network(s) the template is connected to
network_names = list(
(attrs.get("NetworkSettings") or {}).get("Networks", {}).keys()
)
restart_policy_name = (
(attrs.get("HostConfig") or {})
.get("RestartPolicy", {})
.get("Name", "unless-stopped")
) or "unless-stopped"
started = []
for i in range(previous + 1, target + 1):
new_name = f"{COMPOSE_PROJECT}-{SERVICE_NAME}-{i}"
labels = {
"com.docker.compose.project": COMPOSE_PROJECT,
"com.docker.compose.service": SERVICE_NAME,
"com.docker.compose.container-number": str(i),
}
log.info("docker_scaler: creating %s from image %s", new_name, image)
container = client.containers.create(
image=image,
name=new_name,
environment=env,
labels=labels,
mounts=mounts,
restart_policy={"Name": restart_policy_name},
device_requests=device_requests,
)
for net_name in network_names:
try:
net = client.networks.get(net_name)
net.connect(container)
log.info("docker_scaler: connected %s to network %s", new_name, net_name)
except Exception as exc:
log.warning("docker_scaler: could not connect to network %s: %s", net_name, exc)
container.start()
started.append(new_name)
log.info("docker_scaler: started %s", new_name)
return {
"previous": previous,
"current": target,
"delta": target - previous,
"message": f"Started {len(started)} new worker(s): {started}",
}
-121
View File
@@ -1,121 +0,0 @@
"""Flamenco Manager REST API client.
Uses httpx (sync) for compatibility with Celery tasks and FastAPI endpoints.
"""
import logging
from typing import Any
import httpx
logger = logging.getLogger(__name__)
DEFAULT_TIMEOUT = 10.0
class FlamencoClient:
"""Thin wrapper around the Flamenco Manager v3 REST API."""
def __init__(self, manager_url: str):
self.base_url = manager_url.rstrip("/")
def _url(self, path: str) -> str:
return f"{self.base_url}{path}"
# ── Job management ──────────────────────────────────────────────────────
def submit_job(
self,
name: str,
job_type: str,
settings: dict[str, Any],
metadata: dict[str, str] | None = None,
priority: int = 50,
) -> dict:
"""Submit a new render job to Flamenco Manager.
Returns the created job dict (includes 'id').
"""
payload = {
"name": name,
"type": job_type,
"submitter_platform": "linux",
"settings": settings,
"metadata": metadata or {},
"priority": priority,
}
resp = httpx.post(
self._url("/api/v3/jobs"),
json=payload,
timeout=DEFAULT_TIMEOUT,
)
resp.raise_for_status()
return resp.json()
def get_job(self, job_id: str) -> dict:
"""Get job details by ID."""
resp = httpx.get(
self._url(f"/api/v3/jobs/{job_id}"),
timeout=DEFAULT_TIMEOUT,
)
resp.raise_for_status()
return resp.json()
def cancel_job(self, job_id: str) -> None:
"""Request cancellation of a job."""
resp = httpx.post(
self._url(f"/api/v3/jobs/{job_id}/setstatus"),
json={"status": "cancel-requested"},
timeout=DEFAULT_TIMEOUT,
)
resp.raise_for_status()
# ── Workers ─────────────────────────────────────────────────────────────
def list_workers(self) -> list[dict]:
"""List all registered workers."""
resp = httpx.get(
self._url("/api/v3/worker-mgt/workers"),
timeout=DEFAULT_TIMEOUT,
)
resp.raise_for_status()
data = resp.json()
return data.get("workers", data) if isinstance(data, dict) else data
# ── Farm status ─────────────────────────────────────────────────────────
def get_farm_status(self) -> dict:
"""Get overall farm status from the Manager."""
resp = httpx.get(
self._url("/api/v3/configuration"),
timeout=DEFAULT_TIMEOUT,
)
resp.raise_for_status()
return resp.json()
def health_check(self) -> dict:
"""Check if the Flamenco Manager is reachable and return version info."""
try:
resp = httpx.get(
self._url("/api/v3/version"),
timeout=5.0,
)
resp.raise_for_status()
data = resp.json()
return {
"available": True,
"version": data.get("version", "unknown"),
"name": data.get("name", "Flamenco"),
}
except Exception as exc:
logger.warning(f"Flamenco health check failed: {exc}")
return {
"available": False,
"version": None,
"name": None,
"error": str(exc)[:200],
}
def get_flamenco_client(manager_url: str) -> FlamencoClient:
"""Factory that creates a FlamencoClient from a manager URL."""
return FlamencoClient(manager_url)
+8 -286
View File
@@ -1,12 +1,7 @@
"""Render dispatcher — routes render jobs to Celery or Flamenco.
"""Render dispatcher — routes render jobs to Celery.
Backend selection priority:
1. OutputType.render_backend per-type override ("celery" / "flamenco")
2. OutputType.is_animation — animations default to Flamenco
3. System setting render_backend — global default ("celery" / "flamenco" / "auto")
4. "auto" mode: stills → Celery, animations → Flamenco
All renders run via Celery workers (Flamenco removed in v2 refactor).
"""
import json
import logging
from datetime import datetime
@@ -14,7 +9,6 @@ from sqlalchemy import select, update as sql_update
from sqlalchemy.orm import Session, joinedload
from app.models.order_line import OrderLine
from app.models.output_type import OutputType
from app.models.product import Product
from app.models.system_setting import SystemSetting
@@ -29,113 +23,11 @@ def _load_setting(session: Session, key: str, default: str = "") -> str:
return row.value if row else default
def resolve_backend(output_type: OutputType | None, system_backend: str) -> str:
"""Determine which backend to use for a given output type.
Returns "celery" or "flamenco".
"""
if output_type is None:
return "celery"
# Priority 1: explicit per-type override
ot_backend = output_type.render_backend
if ot_backend in ("celery", "flamenco"):
return ot_backend
# Priority 2+3: is_animation + system setting
if system_backend in ("celery", "flamenco"):
return system_backend
# Priority 4: auto mode — animations → Flamenco, stills → Celery
if output_type.is_animation:
return "flamenco"
return "celery"
def build_flamenco_job_settings(
output_type: OutputType,
product: Product,
step_path: str,
output_dir: str,
system_settings: dict[str, str],
lighting_only: bool = False,
shadow_catcher: bool = False,
camera_orbit: bool = True,
cycles_device: str = "auto",
rotation_x: float = 0.0,
rotation_y: float = 0.0,
rotation_z: float = 0.0,
) -> dict:
"""Build Flamenco job settings from output type and product metadata."""
render_settings = output_type.render_settings or {}
engine = render_settings.get("engine", system_settings.get("blender_engine", "cycles"))
samples_key = f"blender_{engine}_samples"
samples = render_settings.get("samples", int(system_settings.get(samples_key, "256")))
stl_quality = render_settings.get("stl_quality", system_settings.get("stl_quality", "low"))
width = render_settings.get("width", 1920 if output_type.is_animation else 1024)
height = render_settings.get("height", 1080 if output_type.is_animation else 1024)
part_colors = {}
part_names_ordered = []
if product.cad_file and product.cad_file.parsed_objects:
part_names_ordered = product.cad_file.parsed_objects.get("objects", [])
materials_source = product.cad_part_materials
if materials_source:
from app.services.step_processor import build_part_colors
part_colors = build_part_colors(part_names_ordered, materials_source)
transparent_bg = bool(output_type.transparent_bg) if hasattr(output_type, 'transparent_bg') else False
settings = {
"step_path": step_path,
"engine": engine,
"samples": samples,
"stl_quality": stl_quality,
"width": width,
"height": height,
"part_colors_json": json.dumps(part_colors),
"transparent_bg": transparent_bg,
"template_path": "",
"target_collection": "Product",
"material_library_path": "",
"material_map_json": "{}",
"part_names_ordered_json": json.dumps(part_names_ordered),
"lighting_only": lighting_only,
"shadow_catcher": shadow_catcher,
"cycles_device": cycles_device,
"rotation_x": rotation_x,
"rotation_y": rotation_y,
"rotation_z": rotation_z,
}
for dk in ('noise_threshold', 'denoiser', 'denoising_input_passes',
'denoising_prefilter', 'denoising_quality', 'denoising_use_gpu'):
settings[dk] = str(render_settings.get(dk, ""))
if output_type.is_animation:
# Turntable-specific settings
output_name = render_settings.get("output_name", "turntable")
settings["output_dir"] = output_dir
settings["output_name"] = output_name
settings["frame_count"] = render_settings.get("frame_count", 120)
settings["fps"] = render_settings.get("fps", 30)
settings["turntable_degrees"] = render_settings.get("turntable_degrees", 360)
settings["turntable_axis"] = render_settings.get("turntable_axis", "world_z")
settings["bg_color"] = render_settings.get("bg_color", "")
settings["camera_orbit"] = camera_orbit
else:
# Still-specific settings
ext = output_type.output_format or "png"
settings["output_path"] = f"{output_dir}/render.{ext}"
return settings
def dispatch_render(order_line_id: str) -> dict:
"""Route a render job to Celery or Flamenco based on configuration.
"""Dispatch a render job to Celery.
Must be called from a sync context (Celery task or sync wrapper).
Returns {"backend": "celery"|"flamenco", "job_ref": str}.
Returns {"backend": "celery", "job_ref": str}.
"""
from app.config import settings as app_settings
from app.services.render_log import emit, clear
@@ -179,196 +71,26 @@ def dispatch_render(order_line_id: str) -> dict:
cad_name = line.product.cad_file.original_name if line.product.cad_file else "?"
emit(order_line_id, f"CAD file: {cad_name}")
emit(order_line_id, "Dispatching to Celery render worker")
# Load system settings
system_backend = _load_setting(session, "render_backend", "celery")
flamenco_url = _load_setting(session, "flamenco_manager_url", "http://flamenco-manager:8080")
backend = resolve_backend(line.output_type, system_backend)
emit(order_line_id, f"Resolved backend: {backend}")
# Mark as processing
now = datetime.utcnow()
session.execute(
sql_update(OrderLine)
.where(OrderLine.id == line.id)
.values(
render_status="processing",
render_backend_used=backend,
render_backend_used="celery",
render_started_at=now,
)
)
session.commit()
if backend == "flamenco":
emit(order_line_id, f"Submitting job to Flamenco Manager ({flamenco_url})")
result = _dispatch_flamenco(session, line, flamenco_url)
if result.get("error"):
emit(order_line_id, f"Flamenco submit failed: {result['error']}", "error")
else:
emit(order_line_id, f"Flamenco job submitted: {result.get('job_ref', '?')}")
return result
else:
emit(order_line_id, "Dispatching to Celery render worker")
return _dispatch_celery(order_line_id)
engine_db.dispose()
return _dispatch_celery(order_line_id)
def _dispatch_celery(order_line_id: str) -> dict:
"""Dispatch to the existing Celery render task."""
"""Dispatch to the Celery render task."""
from app.tasks.step_tasks import render_order_line_task
result = render_order_line_task.delay(order_line_id)
return {"backend": "celery", "job_ref": result.id}
def _dispatch_flamenco(session: Session, line: OrderLine, flamenco_url: str) -> dict:
"""Submit a job to Flamenco Manager."""
import re
from app.services.flamenco_client import get_flamenco_client
# Load all needed system settings
all_keys = ["blender_engine", "blender_cycles_samples", "blender_eevee_samples", "stl_quality", "cycles_device"]
sys_settings = {}
for key in all_keys:
sys_settings[key] = _load_setting(session, key, "")
output_type = line.output_type
product = line.product
cad_file = product.cad_file
# Load render_position for rotation values
rotation_x = rotation_y = rotation_z = 0.0
if line.render_position_id:
from app.models.render_position import ProductRenderPosition
rp = session.get(ProductRenderPosition, line.render_position_id)
if rp:
rotation_x, rotation_y, rotation_z = rp.rotation_x, rp.rotation_y, rp.rotation_z
# Flamenco mounts the uploads volume at /shared, backend uses /app/uploads
raw_path = cad_file.stored_path if cad_file else ""
step_path = raw_path.replace("/app/uploads/", "/shared/") if raw_path else ""
output_dir = f"/shared/renders/{line.id}"
job_type = "schaeffler-turntable" if (output_type and output_type.is_animation) else "schaeffler-still"
# Resolve render template + material library BEFORE building job settings
# (template.lighting_only is needed by build_flamenco_job_settings)
from app.services.template_service import resolve_template, get_material_library_path
category_key = product.category_key if product else None
ot_id = str(line.output_type_id) if line.output_type_id else None
template = resolve_template(category_key=category_key, output_type_id=ot_id)
material_library = get_material_library_path()
# Resolve cycles_device: per-output-type override wins, fall back to system setting
ot_cycles_device = output_type.cycles_device if output_type else None
effective_cycles_device = ot_cycles_device or sys_settings.get("cycles_device", "gpu") or "gpu"
settings = build_flamenco_job_settings(
output_type=output_type,
product=product,
step_path=step_path,
output_dir=output_dir,
system_settings=sys_settings,
lighting_only=bool(template.lighting_only) if template else False,
shadow_catcher=bool(template.shadow_catcher_enabled) if template else False,
camera_orbit=bool(template.camera_orbit) if template else True,
cycles_device=effective_cycles_device,
rotation_x=rotation_x,
rotation_y=rotation_y,
rotation_z=rotation_z,
)
if template:
# Remap path for Flamenco shared volume
tmpl_path = template.blend_file_path.replace("/app/uploads/", "/shared/")
settings["template_path"] = tmpl_path
settings["target_collection"] = template.target_collection
logger.info(
f"Flamenco job: using render template '{template.name}' "
f"(id={template.id}, path={tmpl_path}, collection={template.target_collection})"
)
else:
logger.info(
f"Flamenco job: no render template found for "
f"category_key={category_key!r}, output_type_id={ot_id!r} — using factory settings"
)
# Material library + material map: send whenever library exists and product
# has material assignments — works with or without a render template.
# When a template is present, only apply if material_replace_enabled is set.
materials_source = product.cad_part_materials
use_materials = bool(material_library and materials_source)
if template and not template.material_replace_enabled:
use_materials = False
if use_materials:
mat_lib_path = material_library.replace("/app/uploads/", "/shared/")
settings["material_library_path"] = mat_lib_path
mat_map = {
m["part_name"]: m["material"]
for m in materials_source
if m.get("part_name") and m.get("material")
}
# Resolve raw material names to SCHAEFFLER library names via aliases
from app.services.material_service import resolve_material_map
mat_map = resolve_material_map(mat_map)
settings["material_map_json"] = json.dumps(mat_map)
# Output naming: meaningful filename instead of generic render.ext
def _sanitize(s: str) -> str:
return re.sub(r'[^\w\-.]', '_', s.strip())[:100]
product_name = product.name or product.pim_id or "product"
ot_name = output_type.name if output_type else "render"
if not (output_type and output_type.is_animation):
ext = output_type.output_format or "png" if output_type else "png"
filename = f"{_sanitize(product_name)}_{_sanitize(ot_name)}.{ext}"
settings["output_path"] = f"{output_dir}/{filename}"
metadata = {
"order_line_id": str(line.id),
"order_id": str(line.order_id),
"product_name": product.name or "",
"output_type": output_type.name if output_type else "",
"category": product.category_key or "",
}
job_name = f"{product.name or product.pim_id} - {output_type.name if output_type else 'render'}"
try:
client = get_flamenco_client(flamenco_url)
job = client.submit_job(
name=job_name[:200],
job_type=job_type,
settings=settings,
metadata=metadata,
)
job_id = job.get("id", "")
# Save flamenco_job_id
session.execute(
sql_update(OrderLine)
.where(OrderLine.id == line.id)
.values(flamenco_job_id=job_id)
)
session.commit()
logger.info(f"Flamenco job submitted: {job_id} for OrderLine {line.id}")
return {"backend": "flamenco", "job_ref": job_id}
except Exception as exc:
logger.error(f"Flamenco submit failed for OrderLine {line.id}: {exc}")
session.execute(
sql_update(OrderLine)
.where(OrderLine.id == line.id)
.values(
render_status="failed",
render_completed_at=datetime.utcnow(),
render_log={"error": f"Flamenco submit failed: {str(exc)[:500]}"},
)
)
session.commit()
return {"backend": "flamenco", "job_ref": "", "error": str(exc)}