Using Granian 1.7.5, tested it with both Django and Quart under asginl and asgi modes respectively. The Python process under Granian seems to increase in RAM usage over time, most noticeably if the Python application is returning large responses like files.

Here's a small example using Quart

from quart import Quart

app = Quart(__name__)

@app.route("/")
async def hello():
    return "A" * 4 * 1000 * 1024

and a small example with Django (generate a new project and then throw this into the root urls.py)

from django.urls import path
from django.http import HttpResponse

def image(request):
    return HttpResponse("A" * 4 * 1000 * 1024, content_type="image/png")


urlpatterns = [
    path("", image, name="image")
]

Here's a graph showing the problem, this before when I was running with 1.7.5 and then after I downgraded back to 1.6.4. The large downspikes in memory usage on the left side is Kubernetes restarting the workers.

Git bisected the issue to this commit here 0741845

Thank you for this project!