local-llm-server/llm_server/routes/stats.py

from datetime import datetime

from llm_server.routes.cache import redis

# proompters_5_min = 0
# concurrent_semaphore = Semaphore(concurrent_gens)

server_start_time = datetime.now()


# TODO: do I need this?
# def elapsed_times_cleanup():
#     global wait_in_queue_elapsed
#     while True:
#         current_time = time.time()
#         with wait_in_queue_elapsed_lock:
#             global wait_in_queue_elapsed
#             wait_in_queue_elapsed = [(end_time, elapsed_time) for end_time, elapsed_time in wait_in_queue_elapsed if current_time - end_time <= 60]
#         time.sleep(1)


def calculate_avg_gen_time():
    # Get the average generation time from Redis
    average_generation_time = redis.get('average_generation_time')
    if average_generation_time is None:
        return 0
    else:
        return float(average_generation_time)


def get_total_proompts():
    count = redis.get('proompts')
    if count is None:
        count = 0
    else:
        count = int(count)
    return count


def get_active_gen_workers():
    active_gen_workers = redis.get('active_gen_workers')
    if active_gen_workers is None:
        count = 0
    else:
        count = int(active_gen_workers)
    return count