Multiple Tasks Arbitrary Generated after first tasks success

[LEECHOONGHO]

Describe the bug
I operate pytorch training server with 2 celery worker, redis broker, and fastapi server.
And when I give several tasks(3~4) to workers, tasks are well reserved and distributed.
But after first task is successfully done. Multiple tasks generated and reserved on my workers.

Training Process takes 1~3hour for each tasks.

To Reproduce

# FastAPI Server launch command
$ gunicorn train_master:app --workers 8 --worker-class uvicorn.workers.UvicornWorker --bind 0.0.0.0:$80000 --env MANAGERPORT=5555

# Celery Worker Server launch command 
$ celery -A train_worker worker -P threads -c 1 -n train-worker-$N -E --loglevel=debug

# Flower Monitoring Server launch command
$ celery -A train_worker.celery_app flower -A --broker=redis:// --port=5555 --address=0.0.0.0 --loglevel=debug

# Redis Server launch docker compose file contents
version: '3.9'
services:
    redis-server:
        environment:
            - REDISPORT=${REDISPORT}
        image: redis
        network_mode: host
        command: redis-server --port $REDISPORT

# train_worker.py

celery_app = Celery(
    __name__, 
    broker=BROKER_URL, 
    backend=CELERY_RESULT_BACKEND
)

@celery_app.task(
    bind=True,
    base=TrainManagerWrapper,
    name="train",
    retry=False,
)
def train(self, inputs:dict):
    try:
        # train process
    except Exception as e:
        return e

# train_master.py
from train_worker import train, celery_app, AsyncResult

app = FastAPI()

@app.post("/train", status_code=200)
async def train_master(inputs:TrainInput):
    task_id = train.delay(inputs.dict())
    return {"task_id" : str(task_id), "status" : "Processing"}

Expected behavior
I want workers not repeating same tasks.

Screenshots

System information
No Error Messages.

Name: flower
Version: 2.0.1

Name: celery
Version: 5.3.6

Name: kombu
Version: 5.3.4