<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Mayank Thakur</title>
    <description>The latest articles on DEV Community by Mayank Thakur (@mayankthakur001).</description>
    <link>https://dev.to/mayankthakur001</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4089964%2Fc1654db4-8e98-44bb-9ec0-8a42bb19f29d.png</url>
      <title>DEV Community: Mayank Thakur</title>
      <link>https://dev.to/mayankthakur001</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/mayankthakur001"/>
    <language>en</language>
    <item>
      <title>Which Cloud Should You Choose? Help Me Build the Ultimate Cloud Selection Tool! ☁️</title>
      <dc:creator>Mayank Thakur</dc:creator>
      <pubDate>Tue, 17 Nov 2026 18:41:42 +0000</pubDate>
      <link>https://dev.to/mayankthakur001/which-cloud-should-you-choose-help-me-build-the-ultimate-cloud-selection-tool-39mf</link>
      <guid>https://dev.to/mayankthakur001/which-cloud-should-you-choose-help-me-build-the-ultimate-cloud-selection-tool-39mf</guid>
      <description>&lt;p&gt;Choosing the right cloud provider (AWS, Azure, GCP) can be an overwhelming task. With hundreds of services, complex pricing models, and constantly evolving features, developers and architects often spend hours comparing options before writing a single line of code.&lt;br&gt;
To solve this, I’ve been building Which Cloud — a lightweight, intuitive web app designed to simplify the cloud decision-making process!&lt;/p&gt;

&lt;p&gt;🛠️ What is "Which Cloud"?&lt;br&gt;
Which Cloud is built to help developers, students, and teams quickly compare cloud services and find the platform that best fits their project's technical and budgetary needs.&lt;br&gt;
Whether you're looking for the best serverless environment, managed databases, or identity access management (IAM) solutions, the goal is to make the choice clear and straightforward.&lt;/p&gt;

&lt;p&gt;🔮 What’s Next? (Future Implementations)&lt;br&gt;
The project is in its early stages, and there is so much potential to grow. Here are a few features and implementations I'm planning to work on next:&lt;/p&gt;

&lt;p&gt;Interactive Cloud Recommendation Wizard: A step-by-step questionnaire (e.g., "Do you need serverless?", "What is your budget limit?", "Are you looking for strong IAM integrations?") that dynamically recommends the best provider.&lt;/p&gt;

&lt;p&gt;Deep-Dive Service Mapping: A comprehensive directory that maps equivalent services across AWS, Azure, and GCP (e.g., comparing AWS Cognito with Azure AD / Google Identity Platform).&lt;/p&gt;

&lt;p&gt;Real-time Cost Estimator: A simple calculator to estimate and compare monthly costs for basic workloads (compute, storage, and database) across all three major clouds.&lt;/p&gt;

&lt;p&gt;Multi-Cloud Architecture Guides: Visual recommendations for hybrid or multi-cloud setups.&lt;/p&gt;

&lt;p&gt;🤝 We Want YOU to Help Us Build It!&lt;br&gt;
This is an open-source project, and I would love to build this alongside the developer community. Whether you are a:&lt;br&gt;
Frontend Developer (React/Next.js/Tailwind) looking to polish the UI/UX.&lt;/p&gt;

&lt;p&gt;Cloud Engineer or Architect who can help curate accurate service comparisons and architectural best practices.&lt;/p&gt;

&lt;p&gt;Technical Writer who wants to draft cloud comparison guides.&lt;/p&gt;

&lt;p&gt;How to get in touch:&lt;br&gt;
Drop a comment below with your thoughts, feedback, or what features you would find most useful!&lt;/p&gt;

&lt;p&gt;Ping me directly via email at: &lt;a href="mailto:Mayankthakur9181@gmail.com"&gt;Mayankthakur9181@gmail.com&lt;/a&gt;&lt;/p&gt;

&lt;h1&gt;
  
  
  cloud, #opensource, #webdev, #showdev
&lt;/h1&gt;

</description>
      <category>architecture</category>
      <category>cloud</category>
      <category>tools</category>
    </item>
    <item>
      <title>When I started working with AWS, Amazon S3 was one of the first services I came across. It looks simple at first, but there are several important concepts to understand when using it in real applications.
What is Amazon S3?
Amazon S3 (Simple Storage Servic</title>
      <dc:creator>Mayank Thakur</dc:creator>
      <pubDate>Tue, 17 Nov 2026 18:33:46 +0000</pubDate>
      <link>https://dev.to/mayankthakur001/when-i-started-working-with-aws-amazon-s3-was-one-of-the-first-services-i-came-across-it-looks-53pi</link>
      <guid>https://dev.to/mayankthakur001/when-i-started-working-with-aws-amazon-s3-was-one-of-the-first-services-i-came-across-it-looks-53pi</guid>
      <description></description>
    </item>
    <item>
      <title>AWS IAM: A Beginner-Friendly Guide</title>
      <dc:creator>Mayank Thakur</dc:creator>
      <pubDate>Mon, 21 Sep 2026 12:14:31 +0000</pubDate>
      <link>https://dev.to/mayankthakur001/aws-iam-a-beginner-friendly-guide-4878</link>
      <guid>https://dev.to/mayankthakur001/aws-iam-a-beginner-friendly-guide-4878</guid>
      <description>&lt;p&gt;`When you start working with AWS, one of the first services you should understand is AWS Identity and Access Management (IAM).&lt;/p&gt;

&lt;p&gt;IAM controls who can access your AWS resources, what they can access, and what actions they are allowed to perform.&lt;/p&gt;

&lt;p&gt;A simple way to remember it:&lt;/p&gt;

&lt;p&gt;IAM → Who can access → What they can access → What they can do&lt;/p&gt;

&lt;p&gt;Whether you are working with EC2, S3, Lambda, RDS, Terraform, or other AWS services, IAM is involved in controlling access securely.&lt;/p&gt;

&lt;p&gt;What is AWS IAM?&lt;/p&gt;

&lt;p&gt;AWS Identity and Access Management (IAM) is an AWS service that helps you securely manage access to AWS resources.&lt;/p&gt;

&lt;p&gt;With IAM, you can control:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Who can access AWS&lt;/li&gt;
&lt;li&gt;How they authenticate&lt;/li&gt;
&lt;li&gt;Which AWS resources they can access&lt;/li&gt;
&lt;li&gt;Which actions they can perform&lt;/li&gt;
&lt;li&gt;Which actions they are explicitly denied&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For example, imagine a developer only needs to upload files to a specific S3 bucket.&lt;/p&gt;

&lt;p&gt;Instead of giving the developer access to the entire AWS account, you can create a policy that allows access only to that particular bucket and only for the required operations.&lt;/p&gt;

&lt;p&gt;This is the foundation of Least Privilege.&lt;/p&gt;

&lt;p&gt;The Main IAM Components&lt;/p&gt;

&lt;p&gt;IAM becomes much easier to understand when you break it into four major components.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;IAM Users&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;An IAM User represents a person or application that needs AWS access.&lt;/p&gt;

&lt;p&gt;For example: &lt;br&gt;
Developer → IAM User&lt;br&gt;
Administrator → IAM User&lt;br&gt;
Data Analyst → IAM User&lt;/p&gt;

&lt;p&gt;A user can have permissions assigned through policies or groups.&lt;/p&gt;

&lt;p&gt;However, for workloads running on AWS, AWS generally recommends using IAM roles instead of long-term access keys whenever possible.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;IAM Groups&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;An IAM Group is a collection of IAM users.&lt;/p&gt;

&lt;p&gt;Instead of assigning the same permissions individually to 20 developers, you can create a group:&lt;br&gt;
Developers Group&lt;br&gt;
       |&lt;br&gt;
       |-- User A&lt;br&gt;
       |-- User B&lt;br&gt;
       |-- User C&lt;br&gt;
       |-- User D&lt;/p&gt;

&lt;p&gt;Then attach the required policy to the group.&lt;/p&gt;

&lt;p&gt;For example: &lt;br&gt;
Developers Group&lt;br&gt;
       |&lt;br&gt;
S3 Read Access Policy&lt;br&gt;
       |&lt;br&gt;
All developers receive S3 read permissions&lt;/p&gt;

&lt;p&gt;This makes permission management easier when working with teams.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;IAM Roles&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;IAM Roles are extremely important when working with AWS services.&lt;/p&gt;

&lt;p&gt;An IAM Role provides temporary permissions that can be assumed by trusted entities such as:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;EC2&lt;/li&gt;
&lt;li&gt;Lambda&lt;/li&gt;
&lt;li&gt;ECS&lt;/li&gt;
&lt;li&gt;AWS services&lt;/li&gt;
&lt;li&gt;IAM users&lt;/li&gt;
&lt;li&gt;Other AWS accounts&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For example, suppose an EC2 instance needs to read files from an S3 bucket.&lt;/p&gt;

&lt;p&gt;You should avoid putting AWS access keys directly inside the EC2 server.&lt;/p&gt;

&lt;p&gt;Instead: &lt;/p&gt;

&lt;p&gt;EC2 Instance&lt;br&gt;
     |&lt;br&gt;
 IAM Role&lt;br&gt;
     |&lt;br&gt;
 IAM Policy&lt;br&gt;
     |&lt;br&gt;
 Amazon S3&lt;/p&gt;

&lt;p&gt;The EC2 instance assumes the role and receives temporary credentials automatically.&lt;/p&gt;

&lt;p&gt;Why is this better?&lt;/p&gt;

&lt;p&gt;You do not need to hard-code credentials such as:&lt;/p&gt;

&lt;p&gt;AWS_ACCESS_KEY_ID&lt;br&gt;
AWS_SECRET_ACCESS_KEY&lt;/p&gt;

&lt;p&gt;inside your application.&lt;/p&gt;

&lt;p&gt;This reduces the risk of accidentally exposing long-term credentials.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;IAM Policies&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;IAM Policies define what actions are allowed or denied.&lt;/p&gt;

&lt;p&gt;Policies are written in JSON.&lt;/p&gt;

&lt;p&gt;A simplified example:&lt;/p&gt;

&lt;p&gt;{&lt;br&gt;
  "Version": "2012-10-17",&lt;br&gt;
  "Statement": [&lt;br&gt;
    {&lt;br&gt;
      "Effect": "Allow",&lt;br&gt;
      "Action": [&lt;br&gt;
        "s3:GetObject"&lt;br&gt;
      ],&lt;br&gt;
      "Resource": "arn:aws:s3:::my-bucket/*"&lt;br&gt;
    }&lt;br&gt;
  ]&lt;br&gt;
}&lt;/p&gt;

&lt;p&gt;Let’s break it down.&lt;/p&gt;

&lt;p&gt;Effect&lt;/p&gt;

&lt;p&gt;Defines whether the action is:&lt;/p&gt;

&lt;p&gt;Allow&lt;br&gt;
or &lt;br&gt;
Deny&lt;/p&gt;

&lt;p&gt;Action&lt;/p&gt;

&lt;p&gt;Specifies what the principal can do.&lt;/p&gt;

&lt;p&gt;For example:s3:GetObject&lt;br&gt;
s3:PutObject&lt;br&gt;
s3:DeleteObject&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;            IAM
             |
   +---------+---------+
   |         |         |
 Users     Roles     Groups
   |         |         |
   +---------+---------+
             |
         Policies
             |
      AWS Resources
             |
    +--------+--------+
    |        |        |
   S3       EC2     Lambda
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

&lt;p&gt;IAM Best Practices&lt;/p&gt;

&lt;p&gt;Here are some important IAM practices to remember.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;Enable MFA&lt;br&gt;
Protect privileged identities with MFA.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Avoid using the root user for daily operations&lt;br&gt;
Keep the root account protected.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Follow Least Privilege{% embed  %}&lt;br&gt;
Give only the permissions that are required.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Prefer IAM Roles for AWS Workloads&lt;br&gt;
For EC2, Lambda, ECS, and similar workloads, use roles rather than hard-coded long-term credentials.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Review Permissions Regularly&lt;br&gt;
Remove unnecessary permissions and credentials.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;`&lt;/p&gt;

</description>
      <category>aws</category>
      <category>beginners</category>
      <category>security</category>
    </item>
    <item>
      <title>GlobalMart — CI/CD Pipeline on AWS</title>
      <dc:creator>Mayank Thakur</dc:creator>
      <pubDate>Sun, 23 Aug 2026 17:01:00 +0000</pubDate>
      <link>https://dev.to/mayankthakur001/globalmart-cicd-pipeline-on-aws-4lh4</link>
      <guid>https://dev.to/mayankthakur001/globalmart-cicd-pipeline-on-aws-4lh4</guid>
      <description>&lt;p&gt;&lt;strong&gt;Zero-touch deployment of a React e-commerce app from GitHub to EC2 using AWS CodePipeline and CodeDeploy.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Push to &lt;code&gt;main&lt;/code&gt;, walk away. The pipeline builds the app, versions the artifact in S3, and rolls it&lt;br&gt;
out to EC2 with health checks and automatic rollback — no SSH, no manual &lt;code&gt;scp&lt;/code&gt;, no downtime window.&lt;/p&gt;




&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2j3sdu1bp9nrwcqzhrvp.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F2j3sdu1bp9nrwcqzhrvp.png" alt="Architecture" width="800" height="322"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Why this project exists
&lt;/h2&gt;

&lt;p&gt;Manually deploying a build to a server is fine once. It stops being fine the tenth time: someone&lt;br&gt;
copies the wrong folder, forgets to restart the web server, or deploys from a dirty local branch,&lt;br&gt;
and nobody can say which commit is live.&lt;/p&gt;

&lt;p&gt;This project replaces that with a pipeline where the commit SHA is the single source of truth for&lt;br&gt;
what is running in production. The goal was not "get a React app onto EC2" — it was to build the&lt;br&gt;
release path properly: versioned artifacts, an agent-driven deploy with lifecycle hooks, health&lt;br&gt;
verification before traffic is served, and rollback on failure.&lt;/p&gt;



&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fucdoc5uisqka5hh0hbun.jpeg" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fucdoc5uisqka5hh0hbun.jpeg" alt=" " width="800" height="263"&gt;&lt;/a&gt;&lt;br&gt;
&lt;/p&gt;

&lt;pre data-lang="mermaid"&gt;&lt;code&gt;flowchart LR
    DEV[Developer] --&amp;gt;|git push main| GH[GitHub repository]
    GH --&amp;gt;|CodeStar Connection webhook| CP[AWS CodePipeline]

    subgraph CP_STAGES [Pipeline]
      direction TB
      SRC[Source stage] --&amp;gt; BUILD[Build stage&amp;lt;br/&amp;gt;CodeBuild + Node.js]
      BUILD --&amp;gt; DEPLOY[Deploy stage&amp;lt;br/&amp;gt;CodeDeploy]
    end

    CP --&amp;gt; SRC
    BUILD --&amp;gt;|versioned artifact| S3[(S3 artifact bucket)]
    S3 --&amp;gt; DEPLOY
    DEPLOY --&amp;gt;|pull + lifecycle hooks| AGENT[CodeDeploy agent on EC2]
    AGENT --&amp;gt; NGINX[Nginx serving /var/www/globalmart]
    NGINX --&amp;gt; USER[End user]

    AGENT -.-&amp;gt;|deploy + app logs| CW[CloudWatch Logs]
    CP -.-&amp;gt;|state change events| SNS[SNS notification]

    IAM[IAM roles:&amp;lt;br/&amp;gt;pipeline, build, deploy, instance profile] -.-&amp;gt; CP_STAGES&lt;/code&gt;&lt;/pre&gt;



&lt;p&gt;Trust boundaries worth noting: the pipeline role can read the artifact bucket and start&lt;br&gt;
deployments; the &lt;strong&gt;EC2 instance profile&lt;/strong&gt; can only read that one bucket. The instance never&lt;br&gt;
holds credentials that can modify the pipeline. Compromising the web server does not give an&lt;br&gt;
attacker the ability to ship code.&lt;/p&gt;


&lt;h2&gt;
  
  
  Pipeline stages
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Stage&lt;/th&gt;
&lt;th&gt;Provider&lt;/th&gt;
&lt;th&gt;What happens&lt;/th&gt;
&lt;th&gt;Failure behaviour&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Source&lt;/td&gt;
&lt;td&gt;GitHub via CodeStar Connections&lt;/td&gt;
&lt;td&gt;Webhook fires on push to &lt;code&gt;main&lt;/code&gt;, source zipped to S3&lt;/td&gt;
&lt;td&gt;Pipeline stops, no artifact produced&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Build&lt;/td&gt;
&lt;td&gt;AWS CodeBuild (&lt;code&gt;buildspec.yml&lt;/code&gt;)&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;npm ci&lt;/code&gt; then &lt;code&gt;npm run build&lt;/code&gt;, output packaged with deploy scripts&lt;/td&gt;
&lt;td&gt;Pipeline stops before anything reaches the server&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Deploy&lt;/td&gt;
&lt;td&gt;AWS CodeDeploy (&lt;code&gt;appspec.yml&lt;/code&gt;)&lt;/td&gt;
&lt;td&gt;Agent pulls artifact, runs lifecycle hooks, swaps web root&lt;/td&gt;
&lt;td&gt;Automatic rollback to last successful revision&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;CodeDeploy runs &lt;strong&gt;OneAtATime&lt;/strong&gt; against the deployment group, with the deployment marked failed if&lt;br&gt;
&lt;code&gt;ValidateService&lt;/code&gt; cannot get a 200 from the app — so a broken build never stays live.&lt;/p&gt;
&lt;h3&gt;
  
  
  Deployment lifecycle hooks
&lt;/h3&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;ApplicationStop     → scripts/stop_server.sh      stop nginx gracefully
BeforeInstall       → scripts/before_install.sh   install deps, clear old web root
Install             → (CodeDeploy copies files)   artifact → /var/www/globalmart
AfterInstall        → scripts/after_install.sh    set ownership + permissions
ApplicationStart    → scripts/start_server.sh     start and enable nginx
ValidateService     → scripts/validate_service.sh curl health check, fail non-200
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Repository layout
&lt;/h2&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Globalmart-cicd-pipeline/
├── appspec.yml              # CodeDeploy instructions — MUST be at repo root
├── buildspec.yml            # CodeBuild build steps
├── scripts/
│   ├── before_install.sh
│   ├── after_install.sh
│   ├── start_server.sh
│   ├── stop_server.sh
│   └── validate_service.sh
├── src/                     # React application source
├── public/
├── screenshots/             # pipeline runs, deploy events, live app
└── README.md
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Configuration
&lt;/h2&gt;
&lt;h3&gt;
  
  
  &lt;code&gt;buildspec.yml&lt;/code&gt;
&lt;/h3&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;version&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;0.2&lt;/span&gt;

&lt;span class="na"&gt;phases&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;install&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;runtime-versions&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="na"&gt;nodejs&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;20&lt;/span&gt;
  &lt;span class="na"&gt;pre_build&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;commands&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;npm ci&lt;/span&gt;
  &lt;span class="na"&gt;build&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;commands&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;npm run build&lt;/span&gt;
  &lt;span class="na"&gt;post_build&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;commands&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;echo "Build completed for commit $CODEBUILD_RESOLVED_SOURCE_VERSION"&lt;/span&gt;

&lt;span class="na"&gt;artifacts&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;files&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;build/**/*&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;appspec.yml&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;scripts/**/*&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;&lt;code&gt;appspec.yml&lt;/code&gt; and &lt;code&gt;scripts/&lt;/code&gt; must be listed in artifacts — if they are not in the bundle,&lt;br&gt;
CodeDeploy has nothing to execute and fails immediately.&lt;/p&gt;
&lt;h3&gt;
  
  
  &lt;code&gt;appspec.yml&lt;/code&gt;
&lt;/h3&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;version&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;0.0&lt;/span&gt;
&lt;span class="na"&gt;os&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;linux&lt;/span&gt;

&lt;span class="na"&gt;files&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;source&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;/build&lt;/span&gt;
    &lt;span class="na"&gt;destination&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;/var/www/globalmart&lt;/span&gt;

&lt;span class="na"&gt;hooks&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;ApplicationStop&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;location&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;scripts/stop_server.sh&lt;/span&gt;
      &lt;span class="na"&gt;timeout&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;60&lt;/span&gt;
      &lt;span class="na"&gt;runas&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;root&lt;/span&gt;
  &lt;span class="na"&gt;BeforeInstall&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;location&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;scripts/before_install.sh&lt;/span&gt;
      &lt;span class="na"&gt;timeout&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;300&lt;/span&gt;
      &lt;span class="na"&gt;runas&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;root&lt;/span&gt;
  &lt;span class="na"&gt;AfterInstall&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;location&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;scripts/after_install.sh&lt;/span&gt;
      &lt;span class="na"&gt;timeout&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;120&lt;/span&gt;
      &lt;span class="na"&gt;runas&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;root&lt;/span&gt;
  &lt;span class="na"&gt;ApplicationStart&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;location&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;scripts/start_server.sh&lt;/span&gt;
      &lt;span class="na"&gt;timeout&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;120&lt;/span&gt;
      &lt;span class="na"&gt;runas&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;root&lt;/span&gt;
  &lt;span class="na"&gt;ValidateService&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;location&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;scripts/validate_service.sh&lt;/span&gt;
      &lt;span class="na"&gt;timeout&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="m"&gt;120&lt;/span&gt;
      &lt;span class="na"&gt;runas&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;root&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

&lt;h3&gt;
  
  
  &lt;code&gt;scripts/validate_service.sh&lt;/code&gt;
&lt;/h3&gt;

&lt;p&gt;The hook that makes this a real pipeline rather than a file copy:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;#!/bin/bash&lt;/span&gt;
&lt;span class="nb"&gt;set&lt;/span&gt; &lt;span class="nt"&gt;-e&lt;/span&gt;

&lt;span class="k"&gt;for &lt;/span&gt;i &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="o"&gt;{&lt;/span&gt;1..10&lt;span class="o"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="k"&gt;do
  &lt;/span&gt;&lt;span class="nv"&gt;code&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="si"&gt;$(&lt;/span&gt;curl &lt;span class="nt"&gt;-s&lt;/span&gt; &lt;span class="nt"&gt;-o&lt;/span&gt; /dev/null &lt;span class="nt"&gt;-w&lt;/span&gt; &lt;span class="s2"&gt;"%{http_code}"&lt;/span&gt; http://localhost/ &lt;span class="o"&gt;||&lt;/span&gt; &lt;span class="nb"&gt;true&lt;/span&gt;&lt;span class="si"&gt;)&lt;/span&gt;
  &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="o"&gt;[&lt;/span&gt; &lt;span class="s2"&gt;"&lt;/span&gt;&lt;span class="nv"&gt;$code&lt;/span&gt;&lt;span class="s2"&gt;"&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"200"&lt;/span&gt; &lt;span class="o"&gt;]&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="k"&gt;then
    &lt;/span&gt;&lt;span class="nb"&gt;echo&lt;/span&gt; &lt;span class="s2"&gt;"Health check passed on attempt &lt;/span&gt;&lt;span class="nv"&gt;$i&lt;/span&gt;&lt;span class="s2"&gt;"&lt;/span&gt;
    &lt;span class="nb"&gt;exit &lt;/span&gt;0
  &lt;span class="k"&gt;fi
  &lt;/span&gt;&lt;span class="nb"&gt;echo&lt;/span&gt; &lt;span class="s2"&gt;"Attempt &lt;/span&gt;&lt;span class="nv"&gt;$i&lt;/span&gt;&lt;span class="s2"&gt; returned &lt;/span&gt;&lt;span class="nv"&gt;$code&lt;/span&gt;&lt;span class="s2"&gt;, retrying..."&lt;/span&gt;
  &lt;span class="nb"&gt;sleep &lt;/span&gt;3
&lt;span class="k"&gt;done

&lt;/span&gt;&lt;span class="nb"&gt;echo&lt;/span&gt; &lt;span class="s2"&gt;"Health check failed — triggering rollback"&lt;/span&gt;
&lt;span class="nb"&gt;exit &lt;/span&gt;1
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;






&lt;h2&gt;
  
  
  IAM roles used
&lt;/h2&gt;

&lt;p&gt;Four distinct roles, each scoped to one job. This is the part most tutorials collapse into one&lt;br&gt;
over-permissioned role.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Role&lt;/th&gt;
&lt;th&gt;Attached to&lt;/th&gt;
&lt;th&gt;Permissions&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;CodePipelineServiceRole&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;CodePipeline&lt;/td&gt;
&lt;td&gt;Read/write artifact bucket, start CodeBuild, create CodeDeploy deployments, use the CodeStar connection&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;CodeBuildServiceRole&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;CodeBuild project&lt;/td&gt;
&lt;td&gt;Write artifacts to S3, write CloudWatch Logs&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;CodeDeployServiceRole&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;CodeDeploy application&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;AWSCodeDeployRole&lt;/code&gt; — describe and tag EC2 instances&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;EC2InstanceProfile&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;EC2 instance&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;s3:GetObject&lt;/code&gt; on the artifact bucket only, plus CloudWatch agent permissions&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The instance profile is intentionally read-only and single-bucket. No &lt;code&gt;s3:*&lt;/code&gt;, no wildcard resource.&lt;/p&gt;


&lt;h2&gt;
  
  
  Setup: reproduce this pipeline
&lt;/h2&gt;
&lt;h3&gt;
  
  
  1. Prepare the EC2 instance
&lt;/h3&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;# Amazon Linux 2023, t3.micro, security group: 22 (your IP only) + 80 (0.0.0.0/0)&lt;/span&gt;
&lt;span class="nb"&gt;sudo &lt;/span&gt;dnf update &lt;span class="nt"&gt;-y&lt;/span&gt;
&lt;span class="nb"&gt;sudo &lt;/span&gt;dnf &lt;span class="nb"&gt;install&lt;/span&gt; &lt;span class="nt"&gt;-y&lt;/span&gt; nginx ruby wget
&lt;span class="nb"&gt;sudo &lt;/span&gt;systemctl &lt;span class="nb"&gt;enable&lt;/span&gt; &lt;span class="nt"&gt;--now&lt;/span&gt; nginx

&lt;span class="c"&gt;# CodeDeploy agent (region-specific bucket)&lt;/span&gt;
&lt;span class="nb"&gt;cd&lt;/span&gt; /home/ec2-user
wget https://aws-codedeploy-ap-south-1.s3.ap-south-1.amazonaws.com/latest/install
&lt;span class="nb"&gt;chmod&lt;/span&gt; +x ./install
&lt;span class="nb"&gt;sudo&lt;/span&gt; ./install auto
&lt;span class="nb"&gt;sudo &lt;/span&gt;systemctl status codedeploy-agent    &lt;span class="c"&gt;# must be active&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;Tag the instance &lt;code&gt;Name=globalmart-prod&lt;/code&gt; — CodeDeploy targets the deployment group by tag.&lt;/p&gt;
&lt;h3&gt;
  
  
  2. Create the AWS resources
&lt;/h3&gt;

&lt;ul&gt;
&lt;li&gt;S3 bucket for artifacts, versioning enabled, all public access blocked&lt;/li&gt;
&lt;li&gt;CodeDeploy application (EC2/On-premises) + deployment group targeting the instance tag&lt;/li&gt;
&lt;li&gt;CodeBuild project pointing at &lt;code&gt;buildspec.yml&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;CodePipeline: Source (GitHub via CodeStar Connection) → Build → Deploy&lt;/li&gt;
&lt;li&gt;Attach the instance profile to the EC2 instance&lt;/li&gt;
&lt;/ul&gt;
&lt;h3&gt;
  
  
  3. Verify
&lt;/h3&gt;


&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;git commit &lt;span class="nt"&gt;--allow-empty&lt;/span&gt; &lt;span class="nt"&gt;-m&lt;/span&gt; &lt;span class="s2"&gt;"test: trigger pipeline"&lt;/span&gt;
git push origin main
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;Then confirm, in order:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;CodePipeline shows Source succeeded within ~30 seconds&lt;/li&gt;
&lt;li&gt;CodeBuild logs show &lt;code&gt;npm run build&lt;/code&gt; completing&lt;/li&gt;
&lt;li&gt;CodeDeploy shows all five lifecycle events green&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;curl http://&amp;lt;EC2-public-IP&amp;gt;&lt;/code&gt; returns the app&lt;/li&gt;
&lt;li&gt;On the instance: &lt;code&gt;tail -f /var/log/aws/codedeploy-agent/codedeploy-agent.log&lt;/code&gt;
&lt;/li&gt;
&lt;/ol&gt;


&lt;h2&gt;
  
  
  Troubleshooting: what actually broke
&lt;/h2&gt;

&lt;p&gt;Real failures hit while building this, and the fixes. Kept here because these are the things&lt;br&gt;
tutorials skip.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;ApplicationStop&lt;/code&gt; failed on the very first deployment&lt;/strong&gt;&lt;br&gt;
The hook script from the &lt;em&gt;previous&lt;/em&gt; revision runs during &lt;code&gt;ApplicationStop&lt;/code&gt;, and on deployment #1&lt;br&gt;
there is no previous revision. Fix: use "Ignore ApplicationStop lifecycle event failures" on the&lt;br&gt;
first run, or make the script idempotent with &lt;code&gt;systemctl stop nginx || true&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;InstallError: The deployment failed because a specified file already exists&lt;/code&gt;&lt;/strong&gt;&lt;br&gt;
CodeDeploy will not overwrite files it did not place there. Fix: &lt;code&gt;rm -rf /var/www/globalmart/*&lt;/code&gt;&lt;br&gt;
in &lt;code&gt;before_install.sh&lt;/code&gt;, or set the file overwrite behaviour on the deployment group.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;ScriptFailed&lt;/code&gt; with exit code 126&lt;/strong&gt;&lt;br&gt;
Shell scripts committed without the execute bit. Fix: &lt;code&gt;git update-index --chmod=+x scripts/*.sh&lt;/code&gt;,&lt;br&gt;
or &lt;code&gt;chmod +x&lt;/code&gt; inside &lt;code&gt;before_install.sh&lt;/code&gt; before the hooks run.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;npm run build&lt;/code&gt; killed on a t2.micro&lt;/strong&gt;&lt;br&gt;
1 GB of RAM is not enough for a Vite/CRA production build; the OOM killer takes the process and&lt;br&gt;
the log just says "Killed." Fix: build in CodeBuild rather than on the instance (which is what&lt;br&gt;
this pipeline does), or add swap on the instance if you must build there.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Deployment stuck at &lt;code&gt;Pending&lt;/code&gt;, then timed out&lt;/strong&gt;&lt;br&gt;
The CodeDeploy agent could not reach the service. Causes seen: agent not running, no instance&lt;br&gt;
profile attached, or a private subnet with no NAT and no VPC endpoint. Fix: &lt;code&gt;systemctl status&lt;br&gt;
codedeploy-agent&lt;/code&gt;, confirm the instance profile, check egress.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;AccessDenied&lt;/code&gt; pulling the artifact from S3&lt;/strong&gt;&lt;br&gt;
The instance profile lacked &lt;code&gt;s3:GetObject&lt;/code&gt; on the artifact bucket, and separately the bucket was&lt;br&gt;
in a different region than the deployment. Fix: scope the policy to that bucket ARN and keep&lt;br&gt;
bucket, pipeline, and instance in one region.&lt;/p&gt;


&lt;h2&gt;
  
  
  Monitoring and observability
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;CloudWatch Logs: CodeBuild build logs, plus the CodeDeploy agent log shipped from the instance&lt;/li&gt;
&lt;li&gt;CloudWatch Alarm on EC2 &lt;code&gt;StatusCheckFailed&lt;/code&gt; and high CPU&lt;/li&gt;
&lt;li&gt;SNS topic subscribed to CodePipeline state-change events, so a failed deploy emails immediately&lt;/li&gt;
&lt;li&gt;Nginx access and error logs streamed via the CloudWatch agent&lt;/li&gt;
&lt;/ul&gt;


&lt;h2&gt;
  
  
  Cost and teardown
&lt;/h2&gt;

&lt;p&gt;Running cost on free tier is effectively zero; outside it, roughly a few dollars a month —&lt;br&gt;
t3.micro on-demand, S3 storage in the low cents, CodePipeline at one dollar per active pipeline&lt;br&gt;
per month, CodeBuild billed per build minute.&lt;/p&gt;

&lt;p&gt;Teardown, in this order, to avoid orphaned charges:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;# 1. delete the pipeline   2. delete the CodeBuild project&lt;/span&gt;
&lt;span class="c"&gt;# 3. delete the CodeDeploy app + deployment group&lt;/span&gt;
&lt;span class="c"&gt;# 4. empty and delete the artifact bucket (versioned — delete all versions)&lt;/span&gt;
&lt;span class="c"&gt;# 5. terminate EC2   6. release any Elastic IP   7. delete the IAM roles&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;An unreleased Elastic IP on a terminated instance is the classic surprise bill.&lt;/p&gt;




&lt;h2&gt;
  
  
  Production hardening — what I would add next
&lt;/h2&gt;

&lt;p&gt;Honest list of what this pipeline does &lt;em&gt;not&lt;/em&gt; do yet:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt; &lt;strong&gt;Terraform&lt;/strong&gt; for the whole stack, so the pipeline itself is version-controlled and not click-configured&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;Blue/green deployment&lt;/strong&gt; behind an ALB and Auto Scaling group, replacing in-place deploys and eliminating the brief serve gap&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;Test and security gates&lt;/strong&gt; in the build stage: unit tests, &lt;code&gt;npm audit&lt;/code&gt;, and SAST as blocking steps&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;Manual approval action&lt;/strong&gt; before production, with a staging deployment group ahead of it&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;S3 + CloudFront&lt;/strong&gt; for the static React build instead of EC2 — cheaper, faster, and no server to patch&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;Secrets via Parameter Store / Secrets Manager&lt;/strong&gt; injected at build time rather than baked into the artifact&lt;/li&gt;
&lt;li&gt; &lt;strong&gt;Container path&lt;/strong&gt;: Docker image to ECR, deployed to ECS Fargate, for reproducible runtime environments&lt;/li&gt;
&lt;/ul&gt;




&lt;h2&gt;
  
  
  What this project demonstrates
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;End-to-end CI/CD design: source, build, artifact versioning, deploy, verify, roll back&lt;/li&gt;
&lt;li&gt;CodeDeploy lifecycle hooks and health-gated releases, not just automated file copying&lt;/li&gt;
&lt;li&gt;Least-privilege IAM across four separate service roles&lt;/li&gt;
&lt;li&gt;Linux service administration and web server configuration on EC2&lt;/li&gt;
&lt;li&gt;Debugging distributed deploy failures from agent logs&lt;/li&gt;
&lt;li&gt;Cost awareness and clean resource teardown&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;Pipeline executions, CodeDeploy lifecycle events, EC2 status, and the running application &lt;/p&gt;




&lt;h2&gt;
  
  
  Author
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Mayank Thakur&lt;/strong&gt; — Cloud, DevOps and SRE&lt;br&gt;
&lt;a href="https://github.com/itzmayank01" rel="noopener noreferrer"&gt;GitHub&lt;/a&gt; · &lt;a href="https://linkedin.com/in/mayankthakur1" rel="noopener noreferrer"&gt;LinkedIn&lt;/a&gt; · &lt;a href="mailto:mayankthakur9181@gmail.com"&gt;mayankthakur9181@gmail.com&lt;/a&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>3 Tier Application On EKS</title>
      <dc:creator>Mayank Thakur</dc:creator>
      <pubDate>Sat, 22 Aug 2026 16:45:45 +0000</pubDate>
      <link>https://dev.to/mayankthakur001/3-tier-application-on-eks-oc7</link>
      <guid>https://dev.to/mayankthakur001/3-tier-application-on-eks-oc7</guid>
      <description>&lt;p&gt;I built a 3-tier application running on Amazon EKS, deployed by a GitHub Actions pipeline that runs on every merge to &lt;code&gt;main&lt;/code&gt;. The first version of that pipeline had two secrets sitting in the repo settings: &lt;code&gt;AWS_ACCESS_KEY_ID&lt;/code&gt; and &lt;code&gt;AWS_SECRET_ACCESS_KEY&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Github- &lt;a href="https://github.com/itzmayank01/3-tier-user-platform-devops" rel="noopener noreferrer"&gt;https://github.com/itzmayank01/3-tier-user-platform-devops&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;They worked. They were also a bad idea, and getting rid of them taught me more about EKS authentication than anything else in the project.&lt;/p&gt;

&lt;p&gt;This post is the setup I ended up with. It includes the part that broke for me and that most tutorials skip: getting past IAM is only half the job, because EKS has its own authorization layer on top.&lt;/p&gt;

&lt;p&gt;Why long-lived keys are a problem&lt;/p&gt;

&lt;p&gt;An IAM access key does not expire. If it leaks, it stays valid until someone notices and revokes it. In a CI pipeline that risk is real:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The key lives in repository settings, and anyone with admin access to the repo can see who added it&lt;/li&gt;
&lt;li&gt;A malicious or careless workflow change can print it, exfiltrate it, or use it for something unrelated&lt;/li&gt;
&lt;li&gt;Rotating it means updating every repo that uses it, so in practice nobody rotates it&lt;/li&gt;
&lt;li&gt;A fork or a compromised third-party action widens the blast radius&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;OIDC replaces that with a token that lives for the length of a single job.&lt;/p&gt;

&lt;p&gt;How OIDC actually works here&lt;/p&gt;

&lt;p&gt;Four steps, and it helps to hold the whole picture in your head before you start:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;GitHub Actions mints a short-lived JSON Web Token describing the job: which repo, which branch, which workflow&lt;/li&gt;
&lt;li&gt;Your workflow sends that token to AWS STS&lt;/li&gt;
&lt;li&gt;AWS checks the token signature against a registered identity provider, then checks the claims inside it against your IAM role's trust policy&lt;/li&gt;
&lt;li&gt;If both pass, STS hands back temporary credentials that expire when the job ends&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;No secret is stored anywhere. The trust is in the claims, not in a shared string.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Register GitHub as an identity provider in IAM
&lt;/h2&gt;

&lt;p&gt;You only do this once per AWS account.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws iam create-open-id-connect-provider &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--url&lt;/span&gt; https://token.actions.githubusercontent.com &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--client-id-list&lt;/span&gt; sts.amazonaws.com
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If you have used GitHub OIDC before, you may remember passing a certificate thumbprint. AWS no longer requires you to manage that for this provider - it verifies and rotates the certificate itself. If your tooling still demands a value, older guides use &lt;code&gt;6938fd4d98bab03faadb97b34396831e3780aea1&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Check that it landed:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws iam list-open-id-connect-providers
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Step 2: Create the IAM role and its trust policy
&lt;/h2&gt;

&lt;p&gt;This is where most of the security lives. The trust policy decides &lt;em&gt;which&lt;/em&gt; GitHub jobs are allowed to assume this role.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;trust-policy.json&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"Version"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"2012-10-17"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"Statement"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Effect"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Allow"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Principal"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="nl"&gt;"Federated"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com"&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Action"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"sts:AssumeRoleWithWebIdentity"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Condition"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="nl"&gt;"StringEquals"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
          &lt;/span&gt;&lt;span class="nl"&gt;"token.actions.githubusercontent.com:aud"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"sts.amazonaws.com"&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="p"&gt;},&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="nl"&gt;"StringLike"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
          &lt;/span&gt;&lt;span class="nl"&gt;"token.actions.githubusercontent.com:sub"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"repo:my-org/my-repo:ref:refs/heads/main"&lt;/span&gt;&lt;span class="w"&gt;
        &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws iam create-role &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--role-name&lt;/span&gt; GitHubActionsEKSDeployRole &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--assume-role-policy-document&lt;/span&gt; file://trust-policy.json
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;Read the &lt;code&gt;sub&lt;/code&gt; condition twice.&lt;/strong&gt; It is the only thing standing between "my main branch can deploy" and "anyone who opens a pull request against my repo can deploy". Some patterns and what they mean:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Pattern&lt;/th&gt;
&lt;th&gt;Who can assume the role&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;repo:my-org/my-repo:ref:refs/heads/main&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Only jobs running on the main branch&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;repo:my-org/my-repo:environment:production&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Only jobs targeting the &lt;code&gt;production&lt;/code&gt; environment&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;repo:my-org/my-repo:pull_request&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Any pull request, including from forks&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;repo:my-org/*&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Every repo in the org&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;repo:*&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Anyone on GitHub. Never do this.&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;I use the environment form for production, because GitHub environments let me add a required reviewer on top. IAM enforces which jobs can assume the role; GitHub enforces who can trigger those jobs.&lt;/p&gt;

&lt;p&gt;Then attach permissions. The deploy role needs very little - enough to look up the cluster endpoint:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"Version"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"2012-10-17"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"Statement"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Effect"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Allow"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Action"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"eks:DescribeCluster"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
      &lt;/span&gt;&lt;span class="nl"&gt;"Resource"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"arn:aws:eks:ap-south-1:111122223333:cluster/my-cluster"&lt;/span&gt;&lt;span class="w"&gt;
    &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If the same job also pushes images, add the ECR actions it needs (&lt;code&gt;ecr:GetAuthorizationToken&lt;/code&gt;, &lt;code&gt;ecr:BatchCheckLayerAvailability&lt;/code&gt;, &lt;code&gt;ecr:PutImage&lt;/code&gt;, &lt;code&gt;ecr:InitiateLayerUpload&lt;/code&gt;, &lt;code&gt;ecr:UploadLayerPart&lt;/code&gt;, &lt;code&gt;ecr:CompleteLayerUpload&lt;/code&gt;). I prefer two separate roles - one for build and push, one for deploy - so a compromised build step cannot touch the cluster.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: The workflow
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;Deploy to EKS&lt;/span&gt;

&lt;span class="na"&gt;on&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;push&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;branches&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="pi"&gt;[&lt;/span&gt;&lt;span class="nv"&gt;main&lt;/span&gt;&lt;span class="pi"&gt;]&lt;/span&gt;

&lt;span class="na"&gt;permissions&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;id-token&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;write&lt;/span&gt;
  &lt;span class="na"&gt;contents&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;read&lt;/span&gt;

&lt;span class="na"&gt;jobs&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;deploy&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;runs-on&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;ubuntu-latest&lt;/span&gt;
    &lt;span class="na"&gt;steps&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;uses&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;actions/checkout@v4&lt;/span&gt;

      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;Configure AWS credentials&lt;/span&gt;
        &lt;span class="na"&gt;uses&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;aws-actions/configure-aws-credentials@v4&lt;/span&gt;
        &lt;span class="na"&gt;with&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
          &lt;span class="na"&gt;role-to-assume&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;arn:aws:iam::111122223333:role/GitHubActionsEKSDeployRole&lt;/span&gt;
          &lt;span class="na"&gt;role-session-name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;gha-deploy-${{ github.run_id }}&lt;/span&gt;
          &lt;span class="na"&gt;aws-region&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;ap-south-1&lt;/span&gt;

      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;Update kubeconfig&lt;/span&gt;
        &lt;span class="na"&gt;run&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;aws eks update-kubeconfig --name my-cluster --region ap-south-1&lt;/span&gt;

      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;Deploy&lt;/span&gt;
        &lt;span class="na"&gt;run&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="pi"&gt;|&lt;/span&gt;
          &lt;span class="s"&gt;kubectl set image deployment/frontend \&lt;/span&gt;
            &lt;span class="s"&gt;frontend=111122223333.dkr.ecr.ap-south-1.amazonaws.com/frontend:${{ github.sha }}&lt;/span&gt;
          &lt;span class="s"&gt;kubectl rollout status deployment/frontend --timeout=120s&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Two things to notice.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;permissions: id-token: write&lt;/code&gt; is not optional.&lt;/strong&gt; Without it, GitHub never mints the token and the credentials step fails with something like &lt;code&gt;Could not assume role with OIDC&lt;/code&gt;. This is the single most common failure, and the error message does not point at the missing permission. Also note that declaring a &lt;code&gt;permissions&lt;/code&gt; block resets &lt;em&gt;all&lt;/em&gt; permissions to none, so you have to list &lt;code&gt;contents: read&lt;/code&gt; explicitly if your job checks out code.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Tag images by commit SHA, not &lt;code&gt;latest&lt;/code&gt;.&lt;/strong&gt; &lt;code&gt;${{ github.sha }}&lt;/code&gt; makes every deploy traceable to a commit, and rolling back is just pointing at the previous SHA.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: The part tutorials skip
&lt;/h2&gt;

&lt;p&gt;At this point my pipeline authenticated to AWS perfectly and then died on the &lt;code&gt;kubectl&lt;/code&gt; step:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;error: You must be logged in to the server (Unauthorized)
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The credentials were fine. The problem is that &lt;strong&gt;EKS has two separate layers&lt;/strong&gt;. IAM decides whether you can talk to the cluster's API endpoint. Kubernetes RBAC decides what you can do once you are there. A brand new IAM role has an AWS identity and no Kubernetes identity at all.&lt;/p&gt;

&lt;p&gt;You have to explicitly map the role into the cluster.&lt;/p&gt;

&lt;h3&gt;
  
  
  The modern way: EKS access entries
&lt;/h3&gt;

&lt;p&gt;Access entries are an EKS API, so you manage cluster access with the same tooling as everything else. First check what your cluster is using:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws eks describe-cluster &lt;span class="nt"&gt;--name&lt;/span&gt; my-cluster &lt;span class="nt"&gt;--query&lt;/span&gt; &lt;span class="s1"&gt;'cluster.accessConfig'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If &lt;code&gt;authenticationMode&lt;/code&gt; is &lt;code&gt;CONFIG_MAP&lt;/code&gt;, switch it:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws eks update-cluster-config &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--name&lt;/span&gt; my-cluster &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--access-config&lt;/span&gt; &lt;span class="nv"&gt;authenticationMode&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;API_AND_CONFIG_MAP
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then create the entry and give it a scope:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;aws eks create-access-entry &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--cluster-name&lt;/span&gt; my-cluster &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--principal-arn&lt;/span&gt; arn:aws:iam::111122223333:role/GitHubActionsEKSDeployRole &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--type&lt;/span&gt; STANDARD

aws eks associate-access-policy &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--cluster-name&lt;/span&gt; my-cluster &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--principal-arn&lt;/span&gt; arn:aws:iam::111122223333:role/GitHubActionsEKSDeployRole &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--policy-arn&lt;/span&gt; arn:aws:eks::aws:cluster-access-policy/AmazonEKSEditPolicy &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;--access-scope&lt;/span&gt; &lt;span class="nb"&gt;type&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;namespace,namespaces&lt;span class="o"&gt;=&lt;/span&gt;production
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;AmazonEKSEditPolicy&lt;/code&gt; scoped to one namespace is usually right for a deploy role. It can update workloads and nothing else. Resist reaching for &lt;code&gt;AmazonEKSClusterAdminPolicy&lt;/code&gt; - it is the &lt;code&gt;AdministratorAccess&lt;/code&gt; of EKS.&lt;/p&gt;

&lt;p&gt;One direction-of-travel note: you can move &lt;code&gt;CONFIG_MAP&lt;/code&gt; to &lt;code&gt;API_AND_CONFIG_MAP&lt;/code&gt; to &lt;code&gt;API&lt;/code&gt;, but you cannot go back. Migrate deliberately.&lt;/p&gt;

&lt;h3&gt;
  
  
  The older way: the aws-auth ConfigMap
&lt;/h3&gt;

&lt;p&gt;If you are on an older cluster still using &lt;code&gt;CONFIG_MAP&lt;/code&gt; mode, you edit &lt;code&gt;aws-auth&lt;/code&gt; in &lt;code&gt;kube-system&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;apiVersion&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;v1&lt;/span&gt;
&lt;span class="na"&gt;kind&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;ConfigMap&lt;/span&gt;
&lt;span class="na"&gt;metadata&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;aws-auth&lt;/span&gt;
  &lt;span class="na"&gt;namespace&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;kube-system&lt;/span&gt;
&lt;span class="na"&gt;data&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;mapRoles&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="pi"&gt;|&lt;/span&gt;
    &lt;span class="s"&gt;- rolearn: arn:aws:iam::111122223333:role/GitHubActionsEKSDeployRole&lt;/span&gt;
      &lt;span class="s"&gt;username: github-actions&lt;/span&gt;
      &lt;span class="s"&gt;groups:&lt;/span&gt;
        &lt;span class="s"&gt;- deployers&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then bind &lt;code&gt;deployers&lt;/code&gt; to a Role or ClusterRole with normal Kubernetes RBAC. Two warnings if you go this route: a bad edit here can lock everyone out of the cluster, including you, and if your IAM role has a path (&lt;code&gt;/ci/GitHubActionsEKSDeployRole&lt;/code&gt;), you must strip the path out of the ARN before EKS will match it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Gotchas worth knowing before you start
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;Could not assume role with OIDC&lt;/code&gt;&lt;/strong&gt; - the missing &lt;code&gt;id-token: write&lt;/code&gt; permission, nine times out of ten&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;Not authorized to perform sts:AssumeRoleWithWebIdentity&lt;/code&gt;&lt;/strong&gt; - your &lt;code&gt;sub&lt;/code&gt; claim does not match. Print &lt;code&gt;github.ref&lt;/code&gt; and &lt;code&gt;github.workflow_ref&lt;/code&gt; in a debug step and compare them character by character with the trust policy&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;You must be logged in to the server (Unauthorized)&lt;/code&gt;&lt;/strong&gt; - IAM is fine, the cluster mapping is missing. This is Step 4&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;kubectl&lt;/code&gt; hangs, then times out&lt;/strong&gt; - your cluster API endpoint is private and a hosted runner cannot reach it. You need a self-hosted runner inside the VPC, or a VPC-connected alternative&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The role works from your laptop but not from CI&lt;/strong&gt; - you are probably testing with a user that has broader permissions than the role. Assume the role locally with &lt;code&gt;aws sts assume-role&lt;/code&gt; and retest&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What you actually get
&lt;/h2&gt;

&lt;p&gt;After the migration my repo has zero AWS secrets. The credentials a job holds are valid for that job only, scoped to one namespace, on one branch. If someone gets a malicious workflow merged, the worst case is bounded by an IAM policy and a Kubernetes RBAC binding instead of by whatever those static keys happened to be allowed to do.&lt;/p&gt;

&lt;p&gt;That is a much better position to defend, and it took an afternoon.&lt;/p&gt;

&lt;p&gt;Next in this series: cutting that same pipeline from five minutes to two and a half with self-hosted runners and parallel stages.&lt;/p&gt;




</description>
      <category>aws</category>
      <category>cicd</category>
      <category>devops</category>
      <category>kubernetes</category>
    </item>
  </channel>
</rss>
