Learn LocalStack - Core Service: S3

#Introduction

In episode 4 we managed the emulator lifecycle: start, restart, and restricting services with SERVICES=s3,dynamodb. Now it's time to use real services. We start with S3, the object storage service that underpins so many modern applications.

Why S3 first? Because S3 is the perfect example of near-perfect stateful emulation in LocalStack: what you store is really stored, what you delete is really gone. Every command in this episode can be repeated exactly on real AWS — only the endpoint differs. This is the most concrete proof of the API parity philosophy we discussed in episode 1.

#Creating Buckets and Objects

The most basic lifecycle: create a bucket, store an object, view its contents. Make sure the emulator is running, then run:

awslocal s3 mb s3://assets-dev
awslocal s3 ls
echo "halo dari LocalStack" > hello.txt
awslocal s3 cp hello.txt s3://assets-dev/hello.txt
awslocal s3 ls s3://assets-dev/
awslocal s3 cat s3://assets-dev/hello.txt

A brief explanation of the commands:

#Versioning

S3 versioning lets a single key store multiple object versions, so overwritten objects can be recovered. Enable it, then store two versions:

awslocal s3api put-bucket-versioning --bucket assets-dev --versioning-configuration Status=Enabled
echo "versi satu" > hello.txt
awslocal s3 cp hello.txt s3://assets-dev/hello.txt
echo "versi dua" > hello.txt
awslocal s3 cp hello.txt s3://assets-dev/hello.txt
awslocal s3api list-object-versions --bucket assets-dev

The list-object-versions output will show two versions for the hello.txt key. Each version has a VersionId that can be used for version-specific operations, e.g. awslocal s3api get-object --version-id. Old versions remain retrievable at any time — that's the core value of versioning.

#Presigned URLs

A presigned URL is a temporary URL that grants access to an object without credentials. A classic use case: a web application gives a file download link that expires after a few minutes. Create one with presign:

awslocal s3 presign s3://assets-dev/hello.txt --expires-in 300

The output is a full URL to localhost:4566 with an authentication token. Open it in a browser or via curl and the object will download — valid for 300 seconds. After it expires, access is denied.

#Lifecycle Configuration

A lifecycle rule automates object management, for example deleting old objects after N days. The configuration is JSON:

{
  "Rules": [
    {
      "ID": "cleanup-temp",
      "Filter": { "Prefix": "tmp/" },
      "Status": "Enabled",
      "Expiration": { "Days": 30 }
    }
  ]
}

Apply it to the bucket:

awslocal s3api put-bucket-lifecycle-configuration \
  --bucket assets-dev \
  --lifecycle-configuration file://lifecycle.json

#Advanced Parity Features

#S3 Tables

S3 Tables is a feature that stores tabular data (Parquet) with integrated querying. In LocalStack, S3 Tables supports the create table bucket workflow and table operations, so you can build a local data lake pipeline before going to production.

#S3 Replication

Since the 2026.06+ release, S3 replication is available in LocalStack. You can configure a bucket that replicates objects to another bucket (possibly in a different region), with replication rules that mimic real AWS. Useful for testing architectures that rely on cross-region replication without creating two real AWS buckets.

#Parity with Production SDKs and the CLI

Because LocalStack mimics AWS endpoints, production applications can be pointed at the emulator just by changing the endpoint. Here's an example using boto3:

boto3 mengarah ke LocalStack

import boto3
 
s3 = boto3.client(
    "s3",
    endpoint_url="http://localhost:4566",
    region_name="us-east-1",
    aws_access_key_id="test",
    aws_secret_access_key="test",
)
print(s3.list_buckets())

The only difference from production code is the endpoint_url parameter. The fake test/test credentials are enough for the emulator, and us-east-1 is the default region.

#Closing

• S3 in LocalStack is stateful emulation: buckets, objects, and versions are genuinely stored.
• Master the mb, cp, ls cycle, plus versioning, presign, and lifecycle configuration.
• Advanced parity features like S3 Tables and replication (2026.06+) are available for local experimentation.
• Integration with production SDKs only requires changing the endpoint — concrete proof of API parity.

You've now mastered buckets and objects. In the next episode, episode 6, we move to the database side: mastering DynamoDB — creating tables with a primary key, put, get, query, scan, LSI and GSI secondary indexes, and data modeling best practices. Keep your emulator running!