How-To: Run Dapr in self-hosted mode with Docker

How to deploy and run Dapr in self-hosted mode using Docker

This article provides guidance on running Dapr with Docker outside of Kubernetes. There are a number of different configurations in which you may wish to run Dapr with Docker that are documented below.


Select a Docker image

Dapr provides a number of prebuilt Docker images for different components, you should select the relevant image for your desired binary, architecture, and tag/version.


There are published Docker images for each of the Dapr components available on Docker Hub.



  • latest: The latest release version, ONLY use for development purposes.
  • edge: The latest edge build (master).
  • major.minor.patch: A release version.
  • major.minor.patch-rc.iteration: A release candidate.


  • latest-arm: The latest release version for ARM, ONLY use for development purposes.
  • edge-arm: The latest edge build for ARM (master).
  • major.minor.patch-arm: A release version for ARM.
  • major.minor.patch-rc.iteration-arm: A release candidate for ARM.

Run app as a process

For development purposes ONLY

If you are running Dapr in a Docker container and your app as a process on the host machine, then you need to configure Docker to use the host network so that Dapr and the app can share a localhost network interface. Unfortunately, the host networking driver for Docker is only supported on Linux hosts. If you are running your Docker daemon on a Linux host, you should be able to run the following to launch Dapr.

docker run --net="host" --mount type=bind,source="$(pwd)"/components,target=/components daprio/daprd:edge ./daprd -app-id <my-app-id> -app-port <my-app-port>

Then you can run your app on the host and they should connect over the localhost network interface.

However, if you are not running your Docker daemon on a Linux host, it is recommended you follow the steps below to run both your app and the Dapr runtime in Docker containers using Docker Compose.

Run app and Dapr in a single Docker container

For development purposes ONLY

It is not recommended to run both the Dapr runtime and an application inside the same container. However, it is possible to do so for local development scenarios. In order to do this, you’ll need to write a Dockerfile that installs the Dapr runtime, Dapr CLI and your app code. You can then invoke both the Dapr runtime and your app code using the Dapr CLI.

Below is an example of a Dockerfile which achieves this:

FROM python:3.7.1
# Install dapr CLI
RUN wget -q -O - | /bin/bash

# Install daprd
ENV PATH="/opt/dapr/:${PATH}"

# Install your app
COPY python .
RUN pip install requests
CMD ["run", "--app-id", "nodeapp", "--app-port", "3000", "node", "app.js"]

Remember that if Dapr needs to communicate with other components i.e. Redis, these also need to be made accessible to it.

Run on a Docker network

If you have multiple instances of Dapr running in Docker containers and want them to be able to communicate with each other i.e. for service invocation, then you’ll need to create a shared Docker network and make sure those Dapr containers are attached to it.

You can create a simple Docker network using

docker network create my-dapr-network

When running your Docker containers, you can attach them to the network using

docker run --net=my-dapr-network ...

Each container will receive a unique IP on that network and be able to communicate with other containers on that network.

Run using Docker-Compose

Docker Compose can be used to define multi-container application configurations. If you wish to run multiple apps with Dapr sidecars locally without Kubernetes then it is recommended to use a Docker Compose definition (docker-compose.yml).

The syntax and tooling of Docker Compose is outside the scope of this article, however, it is recommended you refer to the offical Docker documentation for further details.

In order to run your applications using Dapr and Docker Compose you’ll need to define the sidecar pattern in your docker-compose.yml. For example:

version: '3'
    build: ./node
      - "50001:50001" # Dapr instances communicate over gRPC so we need to expose the gRPC port
      - redis
      - placement
      - hello-dapr
    image: "daprio/daprd:edge"
    command: [
     "-app-id", "nodeapp",
     "-app-port", "3000",
     "-placement-host-address", "placement:50006" # Dapr's placement service can be reach via the docker DNS entry
        - "./components/:/components" # Mount our components folder for the runtime to use
      - nodeapp
    network_mode: "service:nodeapp" # Attach the nodeapp-dapr service to the nodeapp network namespace

  ... # Deploy other daprized services and components (i.e. Redis)

    image: "daprio/dapr"
    command: ["./placement", "-port", "50006"]
      - "50006:50006"
      - hello-dapr

For those running the Docker daemon on a Linux host, you can also use network_mode: host to leverage host networking if needed.

To further learn how to run Dapr with Docker Compose, see the Docker-Compose Sample.

Run on Kubernetes

If your deployment target is Kubernetes then you’re probably better of running your applicaiton and Dapr sidecars directly on a Kubernetes platform. Running Dapr on Kubernetes is a first class experience and is documented separately. Please refer to the Dapr on Kubernetes docs