If you just want to run your containers in the cloud, Azure Container Instances is easy

Introduction

It was decided to use Azure for the project I'm doing now, and I was thinking of developing the screen with Java (Springboot) and deploying it in a container. You can deploy containers with AKS and WebApp on Linux, but ACI (Azure Container Instances) can be deployed more easily.

I would like to introduce the flow of deploying to Azure Container Instances using Jib for deploying to the registry.

Solution introduction

Azure Container Registry

• Azure container registry. If you want to deploy manually, please see the article below.
• Push private images to Azure Container Registry

jib-gradle-plugin

• Up to now used spotify's dockerfile-maven-plugin to build a Dockerfile prepared by myself and even deploy it to ECR. ..
• jib can create a container without a Dockerfile and deploy it to Azure Container Registry, Elastic Container Registry, Dockerhub, Google Container Registry with a little setting by inserting a plugin. (The base image can be changed)

Azure Container Instances

• It is a service like Fargate in AWS, in which you are charged only for the operating time according to the size of the allocated CPU and memory without preparing an instance in which the container operates.
• It seems that you can't do as many things as Fargate, but you can expect more functions in the future.
• Prices were about the same as they are today when compared to Fargate.
• How much does Fargate cost per month
• When I calculate it, it's 1vCPU, 2GB = 4,965 yen, 2vCPU, 4GB = 9,929 yen, which is within the margin of error.

Push to ACR using Jib

Add plugin

build.gradle

plugins {
    id 'com.google.cloud.tools.jib' version '1.4.0'
}

Create an ACR repository and change the image name part. The user name and password should be described in gradle.properties.

• [Reference] Let's push a private image to Azure Container Registry

build.gradle

jib {
    to {
        image = 'xxxxxx.azurecr.io/web'
        auth {
            username = "${USERNAME}"
            password = "${PASSWORD}"
        }
    }
}

gradle.properties

USERNAME=xxxx
PASSWORD=xxxx

After that, if you type a command, it will create a container and push it to ACR.

$ gradlew jib

To create only the container, type the following command. (Once the container is created, it is good to check the operation with docker run)

$ gradlew jibDockerBuild

Deploy to Container Instances

After logging in to Azure, just start Cloud Shell and type the following command.

az container create -g TEST-RG \
   --name poc-web \
   --image xxxxx.azurecr.io/web \
   --ip-address public \
   --registry-username xxxxx \
   --registry-password yyyyyyyyy \
   --ports 80 \
   --dns-name-label zzzzzzz \
   --environment-variables SPRING_PROFILES_ACTIVE=development 

• For xxxx, yyyyyy, enter the ACR user name and password.
• If you add --dns-name, you can access by name.
  • http://zzzzzz.japaneast.azurecontainer.io

Other operations on Container Instances

Get the log

az container logs --resource-group TEST-RG --name poc-web

Attach to log (leads to standard output of container)

az container attach --resource-group TEST-RG --name poc-web

Delete Containr Instances

az container delete -g TEST-RG --name poc-web 

Surveillance

You can monitor using Azure Monitor, but it seems that you can not see the details with CPU, memory, network In / Out. For now, I think it would be like using SideCar to monitor for more metrics.

Summary

Anyway, it's Container Instances to easily deploy a container to Azure.