ContextQMD
Back to Remotion

Deploying to Azure Container Apps

packages/docs/docs/azure-container-apps.mdx

4.0.4716.4 KB
Original Source

:::note This guide has been contributed by the community and has not yet been tested by the Remotion team. :::

This guide provides a walkthrough for deploying Remotion rendering service with basic render queuing on Azure Container Apps.

Prerequisites

  • A working Remotion project (e.g., from npm create video) with provision to use Docker.
  • An Azure account.
  • A basic understanding of Docker and having it installed on your system is required.
  • The Azure CLI installed, authenticated (azure login) and a subscription is selected after authentication.
  • The Azure account is provisioned to use GPU on Azure Container Apps, if not, request here.
  • A docker hub account to host the docker image and authenticate (docker login).

Step 1: Pick a project

For simplicity, we'll use remotion-gpu which is configured to leverage GPU acceleration. This project is based with Remotion render server template and uses the scene composition from gpu-scene project.

Step 2: Build and tag the docker image

From the project directory, build the Docker image and push it to Docker Hub with your account, specifying an image name and version tag. The image-name is remotion-docker-gpu and version is 1.0.0.

bash
docker build -t your-dockerhub-username/remotion-docker-gpu:1.0.0 .

Step 3: Push the image to docker hub

bash
docker push your-dockerhub-username/remotion-docker-gpu:1.0.0

Step 4: Set environment variables

Set the environment variables for the azure cli

bash
CONTAINER_IMAGE="docker.io/your-dockerhub-username/remotion-docker-gpu:1.0.0"
RESOURCE_GROUP="<RESOURCE_GROUP>"
ENVIRONMENT_NAME="<ENVIRONMENT_NAME>"
LOCATION="swedencentral"
CONTAINER_APP_NAME="<CONTAINER_APP_NAME>"
WORKLOAD_PROFILE_NAME="NC8as-T4"
WORKLOAD_PROFILE_TYPE="Consumption-GPU-NC8as-T4"

Variable representation

  • CONTAINER_IMAGE represents the docker.io domain and docker image.
  • LOCATION represents the region where the GPU is allocated.
  • WORKLOAD_PROFILE_NAME and WORKLOAD_PROFILE_TYPE are constant variables from Azure documentation representing GPU family to use.

Step 5: Create the resource group

bash
az group create \
  --name $RESOURCE_GROUP \
  --location $LOCATION \
  --query "properties.provisioningState"

Step 6: Register insights for monitoring

bash
az provider register -n Microsoft.OperationalInsights --wait

Step 7: Create a Container Apps environment

bash
az containerapp env create \
  --name $ENVIRONMENT_NAME \
  --resource-group $RESOURCE_GROUP \
  --location "$LOCATION" \
  --query "properties.provisioningState"

Step 8: Add a workload profile to your environment

bash
az containerapp env workload-profile add \
  --name $ENVIRONMENT_NAME \
  --resource-group $RESOURCE_GROUP \
  --workload-profile-name $WORKLOAD_PROFILE_NAME \
  --workload-profile-type $WORKLOAD_PROFILE_TYPE

Step 9: Deploy container app

bash
az containerapp create \
  --name $CONTAINER_APP_NAME \
  --resource-group $RESOURCE_GROUP \
  --environment $ENVIRONMENT_NAME \
  --image $CONTAINER_IMAGE \
  --target-port 3000 \
  --ingress external \
  --cpu 8.0 \
  --memory 56.0Gi \
  --workload-profile-name $WORKLOAD_PROFILE_NAME \
  --query properties.configuration.ingress.fqdn

Azure CLI will deploy the image to Azure Container Apps and when completed will provide a URL which accepts HTTP requests.

bash
Container app created. Access your app at https://remotion-gpu.grayocean-24741fc9.australiaeast.azurecontainerapps.io/
"remotion-gpu.grayocean-24741fc9.australiaeast.azurecontainerapps.io"

:::note The target-port is the port that the Node.js server listens on for incoming requests. Steps 4 to 9 are based on the Azure Container App documentation. :::

Step 10: Trigger a Render

Send a POST request to the service's /renders endpoint. Replace YOUR_SERVICE_URL with the URL from the deployment output, https://remotion-gpu.grayocean-24741fc9.australiaeast.azurecontainerapps.io/.

bash
curl -X POST {YOUR_SERVICE_URL}/renders \
-H "Content-Type: application/json" \
-d '{}'

Response

If successful, the API will respond with JSON containing the jobId which can be used to get the status of render.

json
{
  "jobId": "955338f6-2607-48bd-bedb-6d4c98f7b4dc"
}

Step 11: Get render status

Send a GET request to get the render status, replace YOUR_SERVICE_URL with the URL from the deployment output and the jobId from the render request output.

bash
curl  {YOUR_SERVICE_URL}/renders/{jobId}

Response

json
{
  "status": "completed",
  "videoUrl": "/renders/955338f6-2607-48bd-bedb-6d4c98f7b4dc.mp4",
  "data": {"titleText": "Hello, world!"}
}

If status is completed, combine YOUR_SERVICE_URL and the videoUrl response to generate the download link (https://remotion-gpu.grayocean-24741fc9.australiaeast.azurecontainerapps.io/renders/955338f6-2607-48bd-bedb-6d4c98f7b4dc.mp4).

Step 12: (Optional) Delete the resource group

bash
az group delete --name $RESOURCE_GROUP

This will delete all the resources inside the resource group including the application container.

:::warning Cost: This implementation does not include cost management. Refer to Azure pricing documentation for usage estimates. Use at your own risk. Performance: Generated videos may appear blurry. GPU acceleration is configured in the project config, but actual GPU utilization has not been confirmed. :::

See also