Skip to content
This repository has been archived by the owner on May 27, 2020. It is now read-only.
/ S3 Public archive
forked from scality/cloudserver

An open-source Node.js implementation of a server handling the S3 protocol

License

Notifications You must be signed in to change notification settings

truststamp/S3

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Scality S3 Server

S3 Server logo

CircleCI Scality CI

Learn more at s3.scality.com

Contributing

In order to contribute, please follow the Contributing Guidelines.

Installation

Dependencies

Building and running the Scality S3 Server requires node.js 6.9.5 and npm v3 . Up-to-date versions can be found at Nodesource.

Clone source code

git clone https://github.com/scality/S3.git

Install js dependencies

Go to the ./S3 folder,

npm install

Run it with a file backend

npm start

This starts an S3 server on port 8000. The default access key is accessKey1 with a secret key of verySecretKey1.

By default the metadata files will be saved in the localMetadata directory and the data files will be saved in the localData directory within the ./S3 directory on your machine. These directories have been pre-created within the repository. If you would like to save the data or metadata in different locations of your choice, you must specify them with absolute paths. So, when starting the server:

mkdir -m 700 $(pwd)/myFavoriteDataPath
mkdir -m 700 $(pwd)/myFavoriteMetadataPath
export S3DATAPATH="$(pwd)/myFavoriteDataPath"
export S3METADATAPATH="$(pwd)/myFavoriteMetadataPath"
npm start

Run it with multiple data backends

export S3DATA='multiple'
npm start

This starts an S3 server on port 8000. The default access key is accessKey1 with a secret key of verySecretKey1.

With multiple backends, you have the ability to choose where each object will be saved by setting the following header with a locationConstraint on a PUT request:

'x-amz-meta-scal-location-constraint':'myLocationConstraint'

If no header is sent with a PUT object request, the location constraint of the bucket will determine where the data is saved. If the bucket has no location constraint, the endpoint of the PUT request will be used to determine location.

See the Configuration section below to learn how to set location constraints.

Run it with an in-memory backend

npm run mem_backend

This starts an S3 server on port 8000. The default access key is accessKey1 with a secret key of verySecretKey1.

Setting your own access key and secret key pairs

You can set credentials for many accounts by editing conf/authdata.json but if you want to specify one set of your own credentials, you can use SCALITY_ACCESS_KEY_ID and SCALITY_SECRET_ACCESS_KEY environment variables.

SCALITY_ACCESS_KEY_ID and SCALITY_SECRET_ACCESS_KEY

These variables specify authentication credentials for an account named "CustomAccount".

Note: Anything in the authdata.json file will be ignored.

SCALITY_ACCESS_KEY_ID=newAccessKey SCALITY_SECRET_ACCESS_KEY=newSecretKey npm start

Run it for continuous integration testing or in production with Docker

DOCKER.md

Testing

You can run the unit tests with the following command:

npm test

You can run the multiple backend unit tests with:

npm run multiple_backend_test

You can run the linter with:

npm run lint

Running functional tests locally:

The test suite requires additional tools, s3cmd and Redis installed in the environment the tests are running in.

  • Install s3cmd
  • Install redis and start Redis.
  • Add localCache section to your config.json:
"localCache": {
    "host": REDIS_HOST,
    "port": REDIS_PORT
}

where REDIS_HOST is your Redis instance IP address ("127.0.0.1" if your Redis is running locally) and REDIS_PORT is your Redis instance port (6379 by default)

  • Add the following to the etc/hosts file on your machine:
127.0.0.1 bucketwebsitetester.s3-website-us-east-1.amazonaws.com
  • Start the S3 server in memory and run the functional tests:
npm run mem_backend
npm run ft_test

Configuration

There are three configuration files for your Scality S3 Server:

  1. conf/authdata.json, described above for authentication

  2. locationConfig.json, to set up configuration options for

    where data will be saved

  3. config.json, for general configuration options

Location Configuration

You must specify at least one locationConstraint in your locationConfig.json (or leave as pre-configured).

For instance, the following locationConstraint will save data sent to myLocationConstraint to the file backend:

"myLocationConstraint": {
    "type": "file",
    "legacyAwsBehavior": false,
    "details": {}
},

Each locationConstraint must include the type, legacyAwsBehavior, and details keys. type indicates which backend will be used for that region. Currently, mem, file, and scality are the supported backends. legacyAwsBehavior indicates whether the region will have the same behavior as the AWS S3 'us-east-1' region. If the locationConstraint type is scality, details should contain connector information for sproxyd. If the locationConstraint type is mem or file, details should be empty.

Once you have your locationConstraints in your locationConfig.json, you can specify a default locationConstraint for each of your endpoints.

For instance, the following sets the localhost endpoint to the myLocationConstraint data backend defined above:

"restEndpoints": {
     "localhost": "myLocationConstraint"
},

If you would like to use an endpoint other than localhost for your Scality S3 Server, that endpoint MUST be listed in your restEndpoints. Otherwise if your server is running with a:

  • file backend: your default location constraint will be file

  • memory backend: your default location constraint will be mem

Endpoints

Note that our S3server supports both:

However, hosted-style requests will not hit the server if you are using an ip address for your host. So, make sure you are using path-style requests in that case. For instance, if you are using the AWS SDK for JavaScript, you would instantiate your client like this:

const s3 = new aws.S3({
   endpoint: 'http://127.0.0.1:8000',
   s3ForcePathStyle: true,
});

Getting started: List of applications that have been tested with S3 Server

GUI

Command Line Tools

https://github.com/scality/S3/blob/master/tests/functional/s3curl/s3curl.pl

~/.aws/credentials on Linux, OS X, or Unix or C:\Users\USERNAME\.aws\credentials on Windows

[default]
aws_access_key_id = accessKey1
aws_secret_access_key = verySecretKey1

~/.aws/config on Linux, OS X, or Unix or C:\Users\USERNAME\.aws\config on Windows

[default]
region = us-east-1

Note: us-east-1 is the default region, but you can specify any region.

See all buckets:

aws s3 ls --endpoint-url=http://localhost:8000

Create bucket:

aws --endpoint-url=http://localhost:8000 s3 mb s3://mybucket

If using s3cmd as a client to S3 be aware that v4 signature format is buggy in s3cmd versions < 1.6.1.

~/.s3cfg on Linux, OS X, or Unix or C:\Users\USERNAME\.s3cfg on Windows

[default]
access_key = accessKey1
secret_key = verySecretKey1
host_base = localhost:8000
host_bucket = %(bucket).localhost:8000
signature_v2 = False
use_https = False

See all buckets:

s3cmd ls

~/.rclone.conf on Linux, OS X, or Unix or C:\Users\USERNAME\.rclone.conf on Windows

[remote]
type = s3
env_auth = false
access_key_id = accessKey1
secret_access_key = verySecretKey1
region = other-v2-signature
endpoint = http://localhost:8000
location_constraint =
acl = private
server_side_encryption =
storage_class =

See all buckets:

rclone lsd remote:

JavaScript

const AWS = require('aws-sdk');

const s3 = new AWS.S3({
    accessKeyId: 'accessKey1',
    secretAccessKey: 'verySecretKey1',
    endpoint: 'localhost:8000',
    sslEnabled: false,
    s3ForcePathStyle: true,
});

JAVA

import com.amazonaws.auth.AWSCredentials;
import com.amazonaws.auth.BasicAWSCredentials;
import com.amazonaws.services.s3.AmazonS3;
import com.amazonaws.services.s3.AmazonS3Client;
import com.amazonaws.services.s3.S3ClientOptions;
import com.amazonaws.services.s3.model.Bucket;

public class S3 {

    public static void main(String[] args) {

        AWSCredentials credentials = new BasicAWSCredentials("accessKey1",
        "verySecretKey1");

        // Create a client connection based on credentials
        AmazonS3 s3client = new AmazonS3Client(credentials);
        s3client.setEndpoint("http://localhost:8000");
        // Using path-style requests
        // (deprecated) s3client.setS3ClientOptions(new S3ClientOptions().withPathStyleAccess(true));
        s3client.setS3ClientOptions(S3ClientOptions.builder().setPathStyleAccess(true).build());

        // Create bucket
        String bucketName = "javabucket";
        s3client.createBucket(bucketName);

        // List off all buckets
        for (Bucket bucket : s3client.listBuckets()) {
            System.out.println(" - " + bucket.getName());
        }
    }
}

Ruby

require 'aws-sdk'

s3 = Aws::S3::Client.new(
  :access_key_id => 'accessKey1',
  :secret_access_key => 'verySecretKey1',
  :endpoint => 'http://localhost:8000',
  :force_path_style => true
)

resp = s3.list_buckets
require "fog"

connection = Fog::Storage.new(
{
    :provider => "AWS",
    :aws_access_key_id => 'accessKey1',
    :aws_secret_access_key => 'verySecretKey1',
    :endpoint => 'http://localhost:8000',
    :path_style => true,
    :scheme => 'http',
})

Python

import boto
from boto.s3.connection import S3Connection, OrdinaryCallingFormat


connection = S3Connection(
    aws_access_key_id='accessKey1',
    aws_secret_access_key='verySecretKey1',
    is_secure=False,
    port=8000,
    calling_format=OrdinaryCallingFormat(),
    host='localhost'
)

connection.create_bucket('mybucket')
import boto3
client = boto3.client(
    's3',
    aws_access_key_id='accessKey1',
    aws_secret_access_key='verySecretKey1',
    endpoint_url='http://localhost:8000'
)

lists = client.list_buckets()

PHP

Should use v3 over v2 because v2 would create virtual-hosted style URLs while v3 generates path-style URLs.

use Aws\S3\S3Client;

$client = S3Client::factory([
    'region'  => 'us-east-1',
    'version'   => 'latest',
    'endpoint' => 'http://localhost:8000',
    'credentials' => [
         'key'    => 'accessKey1',
         'secret' => 'verySecretKey1'
    ]
]);

$client->createBucket(array(
    'Bucket' => 'bucketphp',
));

About

An open-source Node.js implementation of a server handling the S3 protocol

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • JavaScript 97.7%
  • Perl 0.7%
  • Shell 0.5%
  • Java 0.4%
  • C 0.3%
  • Ruby 0.3%
  • Other 0.1%