mirror of
https://github.com/Monadical-SAS/reflector.git
synced 2026-02-04 09:56:47 +00:00
1b33fba3ba
Critical bug fix: padding_workflow was registered on CPU worker (slots=1), causing all padding tasks to run serially instead of in parallel. Changes: - Moved padding_workflow from run_workers_cpu.py to run_workers_llm.py - LLM worker has slots=10, allowing up to 10 parallel padding operations - Padding is I/O-bound (S3 download/upload), not CPU-intensive - CPU worker now handles only mixdown_tracks (compute-heavy, serialized) Impact: - Before: 4 tracks × 5s padding = 20s serial execution - After: 4 tracks × 5s padding = ~5s parallel execution (4 concurrent) - Restores intended performance benefit of the refactoring
59 lines
1.6 KiB
Python
59 lines
1.6 KiB
Python
"""
|
|
LLM/I/O worker pool for all non-CPU tasks.
|
|
Handles: all tasks except mixdown_tracks (padding, transcription, LLM inference, orchestration)
|
|
"""
|
|
|
|
from reflector.hatchet.client import HatchetClientManager
|
|
from reflector.hatchet.workflows.daily_multitrack_pipeline import (
|
|
daily_multitrack_pipeline,
|
|
)
|
|
from reflector.hatchet.workflows.padding_workflow import padding_workflow
|
|
from reflector.hatchet.workflows.subject_processing import subject_workflow
|
|
from reflector.hatchet.workflows.topic_chunk_processing import topic_chunk_workflow
|
|
from reflector.hatchet.workflows.transcription_workflow import transcription_workflow
|
|
from reflector.logger import logger
|
|
from reflector.settings import settings
|
|
|
|
SLOTS = 10
|
|
WORKER_NAME = "llm-worker-pool"
|
|
POOL = "llm-io"
|
|
|
|
|
|
def main():
|
|
if not settings.HATCHET_ENABLED:
|
|
logger.error("HATCHET_ENABLED is False, not starting LLM workers")
|
|
return
|
|
|
|
hatchet = HatchetClientManager.get_client()
|
|
|
|
logger.info(
|
|
"Starting Hatchet LLM worker pool (all tasks except mixdown)",
|
|
worker_name=WORKER_NAME,
|
|
slots=SLOTS,
|
|
labels={"pool": POOL},
|
|
)
|
|
|
|
llm_worker = hatchet.worker(
|
|
WORKER_NAME,
|
|
slots=SLOTS,
|
|
labels={
|
|
"pool": POOL,
|
|
},
|
|
workflows=[
|
|
daily_multitrack_pipeline,
|
|
topic_chunk_workflow,
|
|
subject_workflow,
|
|
padding_workflow,
|
|
transcription_workflow,
|
|
],
|
|
)
|
|
|
|
try:
|
|
llm_worker.start()
|
|
except KeyboardInterrupt:
|
|
logger.info("Received shutdown signal, stopping LLM workers...")
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|