ARTICLE AD BOX
I’m building a long-running background service in Python using asyncio, and I’m facing a memory usage issue that keeps growing over time. The service schedules multiple async tasks every few seconds, processes some I/O-bound work, and then should release memory once tasks are completed. However, after running for several hours, RAM usage increases steadily and never drops back down, even though tasks appear to finish correctly.
I suspect the issue might be related to how I’m storing references to tasks or callbacks, but I can’t pinpoint what I’m doing wrong. I’ve already tried forcing garbage collection and monitoring objects with tracemalloc, but the results are confusing. Below is a simplified version of my code that demonstrates the structure I’m using:
import asyncio tasks = [] async def worker(n): await asyncio.sleep(1) return n * 2 async def scheduler(): while True: task = asyncio.create_task(worker(10)) tasks.append(task) await asyncio.sleep(0.5) async def main(): await scheduler() asyncio.run(main())Even when tasks are completed, memory keeps increasing. What is the correct pattern to manage asyncio tasks in a long-running application to avoid memory leaks? Should I be awaiting or cleaning up tasks differently, or is there a better design approach for this kind of workload?
