Run Django asynchronously

Django runs async def views under any server, but under a WSGI server, such as the Gunicorn that Ingot starts by default, each async view runs in a one-off event loop, without the concurrency it’s written for. Served over ASGI, async views run on an event loop. On Sevalla, you install Uvicorn and name the project’s ASGI application in the settings, and Ingot starts Uvicorn on asgi.py instead of Gunicorn on wsgi.py.

At a glance

This guide changes one thing in the application from Deploy a Django application:


Start command	`uvicorn storefront.asgi:application --host 0.0.0.0 --port "$PORT"`, which Ingot detects from `ASGI_APPLICATION` and Uvicorn

Environment variables

This guide adds no environment variables. WEB_CONCURRENCY sets the number of Uvicorn worker processes, as it does for Gunicorn; the deploy guide lists the other variables.

Before you begin

A Django project deployed on Sevalla, as in Deploy Django.

Install Uvicorn

Install Uvicorn, the ASGI server, and update requirements.txt:

pip install uvicorn
pip freeze > requirements.txt

Add an async view

Add a view that awaits, and that reports which kind of request object Django built:

storefront/views.py

import asyncio

from django.http import JsonResponse


async def status(request):
    await asyncio.sleep(0.1)
    return JsonResponse({"interface": type(request).__name__})

storefront/urls.py

from django.contrib import admin
from django.urls import path

from storefront.views import status

urlpatterns = [
    path("admin/", admin.site.urls),
    path("status/", status),
]

In async views, use the ORM’s async methods, such as await Product.objects.aget(pk=1) and async for product in Product.objects.all(), or wrap synchronous code in asgiref.sync.sync_to_async.

Serve the project over ASGI

In storefront/settings.py, add ASGI_APPLICATION under WSGI_APPLICATION:

storefront/settings.py

WSGI_APPLICATION = "storefront.wsgi.application"
ASGI_APPLICATION = "storefront.asgi.application"

While Gunicorn is installed, Ingot serves a Django project over WSGI with it, even when Uvicorn is installed too. When the settings set ASGI_APPLICATION and Uvicorn is installed, Ingot starts Uvicorn on that application instead:

uvicorn storefront.asgi:application --host 0.0.0.0 --port "$PORT"

The web process still listens on the port in the PORT environment variable, 8080 by default. Uvicorn runs 1 worker process. On a larger size, set the WEB_CONCURRENCY environment variable to the number of worker processes, for example 2: Uvicorn reads it at start. If you set a Custom start command on the web process, it overrides what Ingot detects: clear it on the Processes tab (click ⋯ on the Web process, then click Update process). See Set the start command. If the application builds from your own Dockerfile, the Dockerfile’s CMD starts the server, not Ingot: replace the Gunicorn CMD with Uvicorn. exec replaces the shell with Uvicorn, which then handles SIGTERM itself:

Dockerfile

CMD ["sh", "-c", "exec uvicorn storefront.asgi:application --host 0.0.0.0 --port \"$PORT\""]

Commit and push the changes, requirements.txt included. If Auto-deploy is off, click Deploy now on the application’s Overview tab.

Verify

When the deployment’s status is Deployed, open it on the Deployments tab. The build plan’s Start row reads uvicorn storefront.asgi:application --host 0.0.0.0 --port "$PORT". With a Dockerfile, the build log has no build plan: steps 2 and 3 show that Uvicorn serves the project.
Request the view, with your application’s domain:
curl https://storefront-x7k2m.sevalla.app/status/
The response is {"interface": "ASGIRequest"}. Under Gunicorn, the same view returns {"interface": "WSGIRequest"}.
The web process’s runtime logs, on the application’s Logs tab, include Uvicorn running on http://0.0.0.0:8080. Uvicorn also logs ASGI 'lifespan' protocol appears unsupported.: Django doesn’t implement ASGI lifespan events, and the line is harmless. See Logs.

With the API, Get application deployment logs (getApplicationDeploymentLogsV3) finds the build plan’s Start row, and Get application runtime logs (getApplicationRuntimeLogsV3) finds Uvicorn’s line. APP_ID comes from Deploy with the API and DEPLOYMENT_ID from Deploy with the API:

curl -s "$API/applications/$APP_ID/deployments/$DEPLOYMENT_ID/logs?search=uvicorn%20storefront" \
  -H "Authorization: Bearer $SEVALLA_API_KEY" | jq -r '.[].message'
curl -s "$API/applications/$APP_ID/runtime-logs?search=Uvicorn%20running" \
  -H "Authorization: Bearer $SEVALLA_API_KEY" | jq -r '.[].message'

│ Start            uvicorn storefront.asgi:application --host 0.0.0.0 --port "$PORT"  Django                 │
INFO:     Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)

CLI: sevalla apps deployments logs "$DEPLOYMENT_ID" --app-id "$APP_ID" --search "uvicorn storefront" and sevalla apps logs runtime "$APP_ID" --search "Uvicorn running". MCP: get_deployment_logs and search_runtime_logs.

Deploy with the API

The application from Deploy with the API needs no new settings: Ingot picks Uvicorn from the code. Push the changes, then deploy in the same shell, with API and APP_ID set. Each step names the CLI command and the Sevalla MCP server tool that do the same.

Deploy

Trigger deployment (triggerDeploymentV3) deploys the latest commit. With Auto-deploy on, the push already started a deployment of the same commit; this one queues behind it, and you can wait for either:

DEPLOYMENT_ID=$(curl -s -X POST "$API/applications/$APP_ID/deployments" \
  -H "Authorization: Bearer $SEVALLA_API_KEY" -H "Content-Type: application/json" -d '{}' | jq -r .id)

CLI: sevalla apps deployments trigger "$APP_ID". MCP: trigger_deployment.

Wait for the deployment

Get deployment (getDeploymentV3) reports the status, which ends as success, failed, cancelled or skipped. The loop keeps polling through a failed request, for up to 30 minutes:

for i in $(seq 1 120); do
  STATUS=$(curl -s "$API/applications/$APP_ID/deployments/$DEPLOYMENT_ID" \
    -H "Authorization: Bearer $SEVALLA_API_KEY" | jq -r '.status // empty')
  case $STATUS in success|failed|cancelled|skipped) break ;; esac
  sleep 15
done
echo "$STATUS"

CLI: sevalla apps deployments get "$DEPLOYMENT_ID" --app-id "$APP_ID". MCP: get_deployment.

Then run the checks in Verify.

Get started

Migrate

AI

JavaScript

Python

PHP

Ruby

Go

Rust

Java, Kotlin and Scala

.NET

Elixir

Gleam

Swift

Dart

Static site generators

Run Django asynchronously

At a glance

Environment variables

Before you begin

Install Uvicorn

Add an async view

Serve the project over ASGI

Verify

Deploy with the API

​At a glance

​Environment variables

​Before you begin

​Install Uvicorn

​Add an async view

​Serve the project over ASGI

​Verify

​Deploy with the API

At a glance

Environment variables

Before you begin

Install Uvicorn

Add an async view

Serve the project over ASGI

Verify

Deploy with the API