Configure readiness probes

For containerized applications that serve traffic, you might want to verify that your container is ready to handle incoming requests. Azure Container Instances supports readiness probes to include configurations so that your container can't be accessed under certain conditions. The readiness probe behaves like a Kubernetes readiness probe. For example, a container app might need to load a large data set during startup, and you don't want it to receive requests during this time.

This article explains how to deploy a container group that includes a readiness probe, so that a container only receives traffic when the probe succeeds.

Azure Container Instances also supports liveness probes, which you can configure to cause an unhealthy container to automatically restart.


Currently you cannot use a readiness probe in a container group deployed to a virtual network.

YAML configuration

As an example, create a readiness-probe.yaml file with the following snippet that includes a readiness probe. This file defines a container group that consists of a container running a small web app. The app is deployed from the public image. This containerized app is also demonstrated in Deploy a container instance in Azure using the Azure CLI and other quickstarts.

apiVersion: 2018-10-01
location: eastus
name: readinesstest
  - name: mycontainer
        - "/bin/sh"
        - "-c"
        - "node /usr/src/app/index.js & (sleep 240; touch /tmp/ready); wait"
      - port: 80
          cpu: 1.0
          memoryInGB: 1.5
          - "cat"
          - "/tmp/ready"
        periodSeconds: 5
  osType: Linux
  restartPolicy: Always
    type: Public
    - protocol: tcp
      port: '80'
tags: null
type: Microsoft.ContainerInstance/containerGroups

Start command

The deployment includes a command property defining a starting command that runs when the container first starts running. This property accepts an array of strings. This command simulates a time when the web app runs but the container isn't ready.

First, it starts a shell session and runs a node command to start the web app. It also starts a command to sleep for 240 seconds, after which it creates a file called ready within the /tmp directory:

node /usr/src/app/index.js & (sleep 240; touch /tmp/ready); wait

Readiness command

This YAML file defines a readinessProbe which supports an exec readiness command that acts as the readiness check. This example readiness command tests for the existence of the ready file in the /tmp directory.

When the ready file doesn't exist, the readiness command exits with a non-zero value; the container continues running but can't be accessed. When the command exits successfully with exit code 0, the container is ready to be accessed.

The periodSeconds property designates the readiness command should execute every 5 seconds. The readiness probe runs for the lifetime of the container group.

Example deployment

Run the following command to deploy a container group with the preceding YAML configuration:

az container create --resource-group myResourceGroup --file readiness-probe.yaml

View readiness checks

In this example, during the first 240 seconds, the readiness command fails when it checks for the ready file's existence. The status code returned signals that the container isn't ready.

These events can be viewed from the Azure portal or Azure CLI. For example, the portal shows events of type Unhealthy are triggered upon the readiness command failing.

Portal unhealthy event

Verify container readiness

After starting the container, you can verify that it's not accessible initially. After provisioning, get the IP address of the container group:

az container show --resource-group myResourceGroup --name readinesstest --query "ipAddress.ip" --out tsv

Try to access the site while the readiness probe fails:

wget <ipAddress>

Output shows the site isn't accessible initially:

$ wget
--2019-10-15 16:46:02--
Connecting to connected.
HTTP request sent, awaiting response... 

After 240 seconds, the readiness command succeeds, signaling the container is ready. Now, when you run the wget command, it succeeds:

$ wget
--2019-10-15 16:46:02--
Connecting to connected.
HTTP request sent, awaiting response...200 OK
Length: 1663 (1.6K) [text/html]
Saving to: ‘index.html.1’

index.html.1                       100%[===============================================================>]   1.62K  --.-KB/s    in 0s      

2019-10-15 16:49:38 (113 MB/s) - ‘index.html.1’ saved [1663/1663] 

When the container is ready, you can also access the web app by browsing to the IP address using a web browser.


The readiness probe continues to run for the lifetime of the container group. If the readiness command fails at a later time, the container again becomes inaccessible.

Next steps

A readiness probe could be useful in scenarios involving multi-container groups that consist of dependent containers. For more information about multi-container scenarios, see Container groups in Azure Container Instances.