Migrating to the AWS SDK for Go V2

How to migrate to the AWS SDK for Go V2 from AWS SDK for Go V1.

Minimum Go Version

The AWS SDK for Go V2 requires a minimum Go version of 1.20. Migration from v1 to v2 The latest version of Go can be downloaded on the Downloads page. See the Release History for more information about each Go version release, and relevant information required for upgrading.

Modularization

The AWS SDK for Go V2 has been updated to take advantage of the Go modules which became the default development mode in Go 1.13. A number of packages provided by the SDK have been modularized and are independently versioned and released respectively. This change enables improved application dependency modeling, and enables the SDK to provide new features and functionality that follows the Go module versioning strategy.

The following list are some Go modules provided by the SDK:

Module Description
github.com/aws/aws-sdk-go-v2 The SDK core
github.com/aws/aws-sdk-go-v2/config Shared Configuration Loading
github.com/aws/aws-sdk-go-v2/credentials AWS Credential Providers
github.com/aws/aws-sdk-go-v2/feature/ec2/imds Amazon EC2 Instance Metadata Service Client

The SDK’s service clients and higher level utilities modules are nested under the following import paths:

Import Root Description
github.com/aws/aws-sdk-go-v2/service/ Service Client Modules
github.com/aws/aws-sdk-go-v2/feature/ High-Level utilities for services, for example the Amazon S3 Transfer Manager

Configuration Loading

The session package and associated functionality are replaced with a simplified configuration system provided by the config package. The config package is a separate Go module, and can be included in your applications dependencies by with go get.

go get github.com/aws/aws-sdk-go-v2/config

The session.New, session.NewSession, NewSessionWithOptions, and session.Must must be migrated to config.LoadDefaultConfig.

The config package provides several helper functions that aid in overriding the shared configuration loading programmatically. These function names are prefixed with With followed by option that they override. Let’s look at some examples of how to migrate usage of the session package.

For more information on loading shared configuration see Configuring the AWS SDK for Go V2.

Examples

Migrating from NewSession to LoadDefaultConfig

The following example shows how usage of session.NewSession without additional argument parameters is migrated to config.LoadDefaultConfig.

// V1 using NewSession

import "github.com/aws/aws-sdk-go/aws/session"

// ...

sess, err := session.NewSession()
if err != nil {
	// handle error
}
// V2 using LoadDefaultConfig

import "context"
import "github.com/aws/aws-sdk-go-v2/config"

// ...

cfg, err := config.LoadDefaultConfig(context.TODO())
if err != nil {
	// handle error
}
Migrating from NewSession with aws.Config options

The example shows how to migrate overriding of aws.Config values during configuration loading. One or more config.With* helper functions can be provided to config.LoadDefaultConfig to override the loaded configuration values. In this example the AWS Region is overridden to us-west-2 using the config.WithRegion helper function.

// V1

import "github.com/aws/aws-sdk-go/aws"
import "github.com/aws/aws-sdk-go/aws/session"

// ...

sess, err := session.NewSession(aws.Config{
	Region: aws.String("us-west-2")
})
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/config"

// ...

cfg, err := config.LoadDefaultConfig(context.TODO(),
	config.WithRegion("us-west-2"),
)
if err != nil {
	// handle error
}
Migrating from NewSessionWithOptions

This example shows how to migrate overriding values during configuration loading. Zero or more config.With* helper functions can be provided to config.LoadDefaultConfig to override the loaded configuration values. In this example we show how to override the target profile that is used when loading the AWS SDK shared configuration.

// V1

import "github.com/aws/aws-sdk-go/aws"
import "github.com/aws/aws-sdk-go/aws/session"

// ...

sess, err := session.NewSessionWithOptions(aws.Config{
	Profile: "my-application-profile"
})
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/config"

// ...

cfg, err := config.LoadDefaultConfig(context.TODO(),
	config.WithSharedConfigProfile("my-application-profile"),
)
if err != nil {
	// handle error
}

Mocking and *iface

The *iface packages and interfaces therein (e.g. s3iface.S3API) have been removed. These interface definitions are not stable since they are broken every time a service adds a new operation.

Usage of *iface should be replaced by scoped caller-defined interfaces for the service operations being used:

// V1

import "io"

import "github.com/aws/aws-sdk-go/service/s3"
import "github.com/aws/aws-sdk-go/service/s3/s3iface"

func GetObjectBytes(client s3iface.S3API, bucket, key string) ([]byte, error) {
    object, err := client.GetObject(&s3.GetObjectInput{
        Bucket: &bucket,
        Key:    &key,
    })
    if err != nil {
        return nil, err
    }
    defer object.Body.Close()

    return io.ReadAll(object.Body)
}
// V2

import "context"
import "io"

import "github.com/aws/aws-sdk-go-v2/service/s3"


type GetObjectAPIClient interface {
    GetObject(context.Context, *s3.GetObjectInput, ...func(*s3.Options)) (*s3.GetObjectOutput, error)
}

func GetObjectBytes(ctx context.Context, client GetObjectAPIClient, bucket, key string) ([]byte, error) {
    object, err := api.GetObject(ctx, &s3.GetObjectInput{
        Bucket: &bucket,
        Key:    &key,
    })
    if err != nil {
        return nil, err
    }
    defer object.Body.Close()

    return io.ReadAll(object.Body)
}

See the testing guide for more information.

Credentials & Credential Providers

The aws/credentials package and associated credential providers have been relocated to the credentials package location. The credentials package is a Go module that you retrieve by using go get.

go get github.com/aws/aws-sdk-go-v2/credentials

The AWS SDK for Go V2 release updates the AWS Credential Providers to provide a consistent interface for retrieving AWS Credentials. Each provider implements the aws.CredentialsProvider interface, which defines a Retrieve method that returns a (aws.Credentials, error). aws.Credentials that is analogous to the AWS SDK for Go credentials.Value type.

You must wrap aws.CredentialsProvider objects with aws.CredentialsCache to allow credential caching to occur. You use NewCredentialsCache to construct a aws.CredentialsCache object. By default, credentials configured by config.LoadDefaultConfig are wrapped with aws.CredentialsCache.

The following table list the location changes of the AWS credential providers from AWS SDK for Go V1 to AWS SDK for Go V2.

Name V1 Import V2 Import
Amazon EC2 IAM Role Credentials github.com/aws/aws-sdk-go/aws/credentials/ec2rolecreds github.com/aws/aws-sdk-go-v2/credentials/ec2rolecreds
Endpoint Credentials github.com/aws/aws-sdk-go/aws/credentials/endpointcreds github.com/aws/aws-sdk-go-v2/credentials/endpointcreds
Process Credentials github.com/aws/aws-sdk-go/aws/credentials/processcreds github.com/aws/aws-sdk-go-v2/credentials/processcreds
AWS Security Token Service github.com/aws/aws-sdk-go/aws/credentials/stscreds github.com/aws/aws-sdk-go-v2/credentials/stscreds

Static Credentials

Applications that use credentials.NewStaticCredentials to construct static credential programmatically must use credentials.NewStaticCredentialsProvider.

Example

// V1

import "github.com/aws/aws-sdk-go/aws/credentials"

// ...

appCreds := credentials.NewStaticCredentials(accessKey, secretKey, sessionToken)
value, err := appCreds.Get()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/aws"
import "github.com/aws/aws-sdk-go-v2/credentials"

// ...

appCreds := aws.NewCredentialsCache(credentials.NewStaticCredentialsProvider(accessKey, secretKey, sessionToken))
value, err := appCreds.Retrieve(context.TODO())
if err != nil {
	// handle error
}

Amazon EC2 IAM Role Credentials

You must migrate usage of NewCredentials and NewCredentialsWithClient to use New.

The ec2rolecreds package’s ec2rolecreds.New takes functional options of ec2rolecreds.Options as input, allowing you override the specific Amazon EC2 Instance Metadata Service client to use, or to override the credential expiry window.

Example

// V1

import "github.com/aws/aws-sdk-go/aws/credentials/ec2rolecreds"

// ...

appCreds := ec2rolecreds.NewCredentials(sess)
value, err := appCreds.Get()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/aws"
import "github.com/aws/aws-sdk-go-v2/credentials/ec2rolecreds"

// ...

// New returns an object of a type that satisfies the aws.CredentialProvider interface
appCreds := aws.NewCredentialsCache(ec2rolecreds.New())
value, err := appCreds.Retrieve(context.TODO())
if err != nil {
	// handle error
}

Endpoint Credentials

You must migrate usage of NewCredentialsClient and NewProviderClient to use New.

The endpointcreds package’s New function takes a string argument containing the URL of an HTTP or HTTPS endpoint to retrieve credentials from, and functional options of endpointcreds.Options to mutate the credentials provider and override specific configuration settings.

Process Credentials

You must migrate usage of NewCredentials, NewCredentialsCommand, and NewCredentialsTimeout to use NewProvider or NewProviderCommand.

The processcreds package’s NewProvider function takes a string argument that is the command to be executed in the host environment’s shell, and functional options of Options to mutate the credentials provider and override specific configuration settings.

NewProviderCommand takes an implementation of the NewCommandBuilder interface that defines more complex process commands that might take one or more command-line arguments, or have certain execution environment requirements. DefaultNewCommandBuilder implements this interface, and defines a command builder for a process that requires multiple command-line arguments.

Example

// V1

import "github.com/aws/aws-sdk-go/aws/credentials/processcreds"

// ...

appCreds := processcreds.NewCredentials("/path/to/command")
value, err := appCreds.Get()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/aws"
import "github.com/aws/aws-sdk-go-v2/credentials/processcreds"

// ...

appCreds := aws.NewCredentialsCache(processcreds.NewProvider("/path/to/command"))
value, err := appCreds.Retrieve(context.TODO())
if err != nil {
	// handle error
}

AWS Security Token Service Credentials

AssumeRole

You must migrate usage of NewCredentials, and NewCredentialsWithClient to use NewAssumeRoleProvider.

The stscreds package’s NewAssumeRoleProvider function must be called with a sts.Client, and the AWS Identity and Access Management Role ARN to be assumed from the provided sts.Client’s configured credentials. You can also provide a set of functional options of AssumeRoleOptions to modify other optional settings of the provider.

Example
// V1

import "github.com/aws/aws-sdk-go/aws/credentials/stscreds"

// ...

appCreds := stscreds.NewCredentials(sess, "arn:aws:iam::123456789012:role/demo")
value, err := appCreds.Get()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/credentials/stscreds"
import "github.com/aws/aws-sdk-go-v2/service/sts"

// ...

client := sts.NewFromConfig(cfg)

appCreds := stscreds.NewAssumeRoleProvider(client, "arn:aws:iam::123456789012:role/demo")
value, err := appCreds.Retrieve(context.TODO())
if err != nil {
	// handle error
}

AssumeRoleWithWebIdentity

You must migrate usage of NewWebIdentityCredentials, NewWebIdentityRoleProvider, and NewWebIdentityRoleProviderWithToken to use NewWebIdentityRoleProvider.

The stscreds package’s NewWebIdentityRoleProvider function must be called with a sts.Client, and the AWS Identity and Access Management Role ARN to be assumed using the provided sts.Client’s configured credentials, and an implementation of a IdentityTokenRetriever for providing the OAuth 2.0 or OpenID Connect ID token. IdentityTokenFile is an IdentityTokenRetriever that can be used to provide the web identity token from a file located on the application’s host file-system. You can also provide a set of functional options of WebIdentityRoleOptions to modify other optional settings for the provider.

Example
// V1

import "github.com/aws/aws-sdk-go/aws/credentials/stscreds"

// ...

appCreds := stscreds.NewWebIdentityRoleProvider(sess, "arn:aws:iam::123456789012:role/demo", "sessionName", "/path/to/token")
value, err := appCreds.Get()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/aws"
import "github.com/aws/aws-sdk-go-v2/credentials/stscreds"
import "github.com/aws/aws-sdk-go-v2/service/sts"

// ...

client := sts.NewFromConfig(cfg)

appCreds := aws.NewCredentialsCache(stscreds.NewWebIdentityRoleProvider(
		client,
		"arn:aws:iam::123456789012:role/demo",
		stscreds.IdentityTokenFile("/path/to/file"),
		func(o *stscreds.WebIdentityRoleOptions) {
			o.RoleSessionName = "sessionName"
		}))
value, err := appCreds.Retrieve(context.TODO())
if err != nil {
	// handle error
}

Service Clients

AWS SDK for Go V2 provides service client modules nested under the github.com/aws/aws-sdk-go-v2/service import path. Each service client is contained in a Go package using each service’s unique identifier. The following table provides some examples of service import paths in the AWS SDK for Go V2.

Service Name V1 Import Path V2 Import Path
Amazon S3 github.com/aws/aws-sdk-go/service/s3 github.com/aws/aws-sdk-go-v2/service/s3
Amazon DynamoDB github.com/aws/aws-sdk-go/service/dynamodb github.com/aws/aws-sdk-go-v2/service/dynamodb
Amazon CloudWatch Logs github.com/aws/aws-sdk-go/service/cloudwatchlogs github.com/aws/aws-sdk-go-v2/service/cloudwatchlogs

Each service client package is an independently versioned Go module. To add the service client as a dependency of your application, use the go get command with the service’s import path. For example, to add the Amazon S3 client to your dependencies use

go get github.com/aws/aws-sdk-go-v2/service/s3

Client Construction

You can construct clients in the AWS SDK for Go V2 using either the New or NewFromConfig constructor functions in the client’s package. When migrating from the AWS SDK for Go we recommend that you use the NewFromConfig variant, which will return a new service client using values from an aws.Config. The aws.Config value will have been created while loading the SDK shared configuration using config.LoadDefaultConfig. For details on creating service clients see Using AWS Services.

Example 1
// V1

import "github.com/aws/aws-sdk-go/aws/session"
import "github.com/aws/aws-sdk-go/service/s3"

// ...

sess, err := session.NewSession()
if err != nil {
	// handle error
}

client := s3.New(sess)
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/config"
import "github.com/aws/aws-sdk-go-v2/service/s3"

// ...

cfg, err := config.LoadDefaultConfig(context.TODO())
if err != nil {
	// handle error
}

client := s3.NewFromConfig(cfg)
Example 2: Overriding Client Settings
// V1

import "github.com/aws/aws-sdk-go/aws"
import "github.com/aws/aws-sdk-go/aws/session"
import "github.com/aws/aws-sdk-go/service/s3"

// ...

sess, err := session.NewSession()
if err != nil {
	// handle error
}

client := s3.New(sess, &aws.Config{
	Region: aws.String("us-west-2"),
})
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/config"
import "github.com/aws/aws-sdk-go-v2/service/s3"

// ...

cfg, err := config.LoadDefaultConfig(context.TODO())
if err != nil {
	// handle error
}

client := s3.NewFromConfig(cfg, func(o *s3.Options) {
	o.Region = "us-west-2"
})

Endpoints

The endpoints package no longer exists in the AWS SDK for Go V2. Each service client now embeds its required AWS endpoint metadata within the client package. This reduces the overall binary size of compiled applications by no longer including endpoint metadata for services not used by your application.

Additionally, each service now exposes its own interface for endpoint resolution in EndpointResolverV2. Each API takes a unique set of parameters for a service EndpointParameters, the values of which are sourced by the SDK from various locations when an operation is invoked.

By default, service clients use their configured AWS Region to resolve the service endpoint for the target Region. If your application requires a custom endpoint, you can specify custom behavior on EndpointResolverV2 field on the aws.Config structure. If your application implements a custom endpoints.Resolver you must migrate it to conform to this new per-service interface.

For more information on endpoints and implementing a custom resolver, see Configuring Client Endpoints.

Authentication

The AWS SDK for Go V2 supports more advanced authentication behavior, which enables the use of newer AWS service features such as codecatalyst and S3 Express One Zone. Additionally, this behavior can be customized on a per-client basis.

Invoking API Operations

The number of service client operation methods have been reduced significantly. The <OperationName>Request, <OperationName>WithContext, and <OperationName> methods have all been consolidated into single operation method, <OperationName>.

Example

The following example shows how calls to the <nil> PutObject operation would be migrated from AWS SDK for Go to AWS SDK for Go V2.

// V1

import "context"
import "github.com/aws/aws-sdk-go/service/s3"

// ...

client := s3.New(sess)

// Pattern 1
output, err := client.PutObject(&s3.PutObjectInput{
	// input parameters
})

// Pattern 2
output, err := client.PutObjectWithContext(context.TODO(), &s3.PutObjectInput{
	// input parameters
})

// Pattern 3
req, output := client.PutObjectRequest(context.TODO(), &s3.PutObjectInput{
	// input parameters
})
err := req.Send()
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/service/s3"

// ...

client : s3.NewFromConfig(cfg)

output, err := client.PutObject(context.TODO(), &s3.PutObjectInput{
	// input parameters
})

Service Data Types

The top-level input and output types of an operation are found in the service client package. The input and output type for a given operation follow the pattern of <OperationName>Input and <OperationName>Output, where OperationName is the name of the operation you are invoking. For example the input and output shape for the Amazon S3 PutObject operation are PutObjectInput and PutObjectOutput respectively.

All other service data types, other than input and output types, have been migrated to the types package located under the service client package import path hierarchy. For example, the s3.AccessControlPolicy type is now located at types.AccessControlPolicy.

Enumeration Values

The SDK now provides a typed experience for all API enumeration fields. Rather than using a string literal value copied from the service API reference documentation, you can now use one of the concrete types found in the service client’s types package. For example, you can provide the Amazon S3 PutObjectInput operation with an ACL to be applied on an object. In the AWS SDK for Go V1, this parameter was a *string type. In the {{& alias sdk-go %}} this parameter is now a types.ObjectCannedACL. The types package provides generated constants for the valid enumeration values that can be assigned to this field. For example types.ObjectCannedACLPrivate is the constant for the “private” canned ACL value. This value can be used in place of managing string constants within your application.

Pointer Parameters

The AWS SDK for Go v1 required pointer references to be passed for all input parameters to service operations. The AWS SDK for Go V2 has simplified the experience with most services by removing the need to pass input values as pointers where possible. This change means that many service clients operations no longer require your application to pass pointer references for the following types: uint8, uint16, uint32, int8, int16, int32, float32, float64, bool. Similarly, slice and map element types have been updated accordingly to reflect whether their elements must be passed as pointer references.

The aws package contains helper functions for creating pointers for the Go built-in types, these helpers should be used to more easily handle creating pointer types for these Go types. Similarly, helper methods are provided for safely de-referencing pointer values for these types. For example, the aws.String function converts from string*string. Inversely, the aws.ToString converts from *stringstring. When upgrading your application from AWS SDK for Go V1 to AWS SDK for Go V2, you must migrate usage of the helpers for converting from the pointer types to the non-pointer variants. For example, aws.StringValue must be updated to aws.ToString.

Errors Types

The AWS SDK for Go V2 takes full advantage of the error wrapping functionality introduced in Go 1.13. Services that model error responses have generated types available in their client’s types package that can be used to test whether a client operation error was caused by one of these types. For example Amazon S3 GetObject operation can return a NoSuchKey error if attempting to retrieve an object key that doesn’t exist. You can use errors.As to test whether the returned operation error is a types.NoSuchKey error. In the event a service does not model a specific type for an error, you can utilize the smithy.APIError interface type for inspecting the returned error code and message from the service. This functionality replaces awserr.Error and the other awserr functionality from the AWS SDK for Go V1. For more details information on handling errors see Handling Errors.

Example

// V1

import "github.com/aws/aws-sdk-go/aws/awserr"
import "github.com/aws/aws-sdk-go/service/s3"

// ...

client := s3.New(sess)

output, err := s3.GetObject(&s3.GetObjectInput{
	// input parameters
})
if err != nil {
	if awsErr, ok := err.(awserr.Error); ok {
		if awsErr.Code() == "NoSuchKey" {
			// handle NoSuchKey
		} else {
			// handle other codes
		}
		return
	}
	// handle a error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/service/s3"
import "github.com/aws/aws-sdk-go-v2/service/s3/types"
import "github.com/aws/smithy-go"

// ...

client := s3.NewFromConfig(cfg)

output, err := s3.GetObject(context.TODO(), &s3.GetObjectInput{
	// input parameters
})
if err != nil {
	var nsk *types.NoSuchKey
	if errors.As(err, &nsk) {
		// handle NoSuchKey error
		return
	}
	var apiErr smithy.APIError
	if errors.As(err, &apiErr) {
		code := apiErr.ErrorCode()
		message := apiErr.ErrorMessage()
		// handle error code
		return
	}
	// handle error
	return
}

Paginators

Service operation paginators are no longer invoked as methods on the service client. To use a paginator for an operation you must construct a paginator for an operation using one of the paginator constructor methods. For example, to use paginate over the Amazon S3 ListObjectsV2 operation you must construct its paginator using the s3.NewListObjectsV2Paginator. This constructor returns a ListObjectsV2Paginator which provides the methods HasMorePages, and NextPage for determining whether there are more pages to retrieve and invoking the operation to retrieve the next page respectively. More details on using the SDK paginators can be found at .

Let’s look at an example of how to migrate from a AWS SDK for Go paginator to the AWS SDK for Go V2 equivalent.

Example

// V1

import "fmt"
import "github.com/aws/aws-sdk-go/service/s3"

// ...

client := s3.New(sess)

params := &s3.ListObjectsV2Input{
	// input parameters
}

totalObjects := 0
err := client.ListObjectsV2Pages(params, func(output *s3.ListObjectsV2Output, lastPage bool) bool {
	totalObjects += len(output.Contents)
	return !lastPage
})
if err != nil {
	// handle error
}
fmt.Println("total objects:", totalObjects)
// V2

import "context"
import "fmt"
import "github.com/aws/aws-sdk-go-v2/service/s3"

// ...

client := s3.NewFromConfig(cfg)

params := &s3.ListObjectsV2Input{
	// input parameters
}

totalObjects := 0
paginator := s3.NewListObjectsV2Paginator(client, params)
for paginator.HasMorePages() {
	output, err := paginator.NextPage(context.TODO())
	if err != nil {
		// handle error
	}
	totalObjects += len(output.Contents)
}
fmt.Println("total objects:", totalObjects)

Waiters

Service operation waiters are no longer invoked as methods on the service client. To use a waiter you first construct the desired waiter type, and then invoke the wait method. For example, to wait for a Amazon S3 Bucket to exist, you must construct a BucketExists waiter. Use the s3.NewBucketExistsWaiter constructor to create a s3.BucketExistsWaiter. The s3.BucketExistsWaiter provides a Wait method which can be used to wait for a bucket to become available.

Presigned Requests

The V1 SDK technically supported presigning any AWS SDK operation, however, this does not accurately represent what is actually supported at the service level (and in reality most AWS service operations do not support presigning).

AWS SDK for Go V2 resolves this by exposing specific PresignClient implementations in service packages with specific APIs for supported presignable operations.

Note: If a service is missing presigning support for an operation that you were successfully using in SDK v1, please let us know by filing an issue on GitHub.

Uses of Presign and PresignRequest must be converted to use service-specific presigning clients.

The following example shows how to migrate presigning of an S3 GetObject request:

// V1

import (
	"fmt"
	"time"

	"github.com/aws/aws-sdk-go/aws"
	"github.com/aws/aws-sdk-go/aws/session"
	"github.com/aws/aws-sdk-go/service/s3"
)

func main() {
	sess := session.Must(session.NewSessionWithOptions(session.Options{
		SharedConfigState: session.SharedConfigEnable,
	}))

	svc := s3.New(sess)
	req, _ := svc.GetObjectRequest(&s3.GetObjectInput{
		Bucket: aws.String("bucket"),
		Key:    aws.String("key"),
	})

	// pattern 1
	url1, err := req.Presign(20 * time.Minute)
	if err != nil {
		panic(err)
	}
	fmt.Println(url1)

	// pattern 2
	url2, header, err := req.PresignRequest(20 * time.Minute)
	if err != nil {
		panic(err)
	}
	fmt.Println(url2, header)
}
// V2

import (
	"context"
	"fmt"
	"time"

	"github.com/aws/aws-sdk-go-v2/aws"
	"github.com/aws/aws-sdk-go-v2/config"
	"github.com/aws/aws-sdk-go-v2/service/s3"
)

func main() {
	cfg, err := config.LoadDefaultConfig(context.Background())
	if err != nil {
		panic(err)
	}

	svc := s3.NewPresignClient(s3.NewFromConfig(cfg))
	req, err := svc.PresignGetObject(context.Background(), &s3.GetObjectInput{
		Bucket: aws.String("bucket"),
		Key:    aws.String("key"),
	}, func(o *s3.PresignOptions) {
		o.Expires = 20 * time.Minute
	})
	if err != nil {
		panic(err)
	}

	fmt.Println(req.Method, req.URL, req.SignedHeader)
}

Request customization

The monolithic request.Request API has been re-compartmentalized.

Operation input/output

The opaque Request fields Params and Data, which hold the operation input and output structures respectively, are now accessible within specific middleware phases as input/output:

Request handlers which reference Request.Params and Request.Data must be migrated to middleware.

migrating Params

// V1

import (
    "github.com/aws/aws-sdk-go/aws"
    "github.com/aws/aws-sdk-go/aws/request"
    "github.com/aws/aws-sdk-go/aws/session"
    "github.com/aws/aws-sdk-go/service/s3"
)

func withPutObjectDefaultACL(acl string) request.Option {
    return func(r *request.Request) {
        in, ok := r.Params.(*s3.PutObjectInput)
        if !ok {
            return
        }

        if in.ACL == nil {
            in.ACL = aws.String(acl)
        }
        r.Params = in
    }
}

func main() {
    sess := session.Must(session.NewSession())
    sess.Handlers.Validate.PushBack(withPutObjectDefaultACL(s3.ObjectCannedACLBucketOwnerFullControl))

    // ...
}
// V2

import (
    "context"

    "github.com/aws/aws-sdk-go-v2/service/s3"
    "github.com/aws/aws-sdk-go-v2/service/s3/types"
    "github.com/aws/smithy-go/middleware"
    smithyhttp "github.com/aws/smithy-go/transport/http"
)

type withPutObjectDefaultACL struct {
    acl types.ObjectCannedACL
}

// implements middleware.InitializeMiddleware, which runs BEFORE a request has
// been serialized and can act on the operation input
var _ middleware.InitializeMiddleware = (*withPutObjectDefaultACL)(nil)

func (*withPutObjectDefaultACL) ID() string {
    return "withPutObjectDefaultACL"
}

func (m *withPutObjectDefaultACL) HandleInitialize(ctx context.Context, in middleware.InitializeInput, next middleware.InitializeHandler) (
    out middleware.InitializeOutput, metadata middleware.Metadata, err error,
) {
    input, ok := in.Parameters.(*s3.PutObjectInput)
    if !ok {
        return next.HandleInitialize(ctx, in)
    }

    if len(input.ACL) == 0 {
        input.ACL = m.acl
    }
    in.Parameters = input
    return next.HandleInitialize(ctx, in)
}

// create a helper function to simplify instrumentation of our middleware
func WithPutObjectDefaultACL(acl types.ObjectCannedACL) func (*s3.Options) {
    return func(o *s3.Options) {
        o.APIOptions = append(o.APIOptions, func (s *middleware.Stack) error {
            return s.Initialize.Add(&withPutObjectDefaultACL{acl: acl}, middleware.After)
        })
    }
}

func main() {
    cfg, err := config.LoadDefaultConfig(context.Background())
    if err != nil {
        // ...
    }

    svc := s3.NewFromConfig(cfg, WithPutObjectDefaultACL(types.ObjectCannedACLBucketOwnerFullControl))
    // ...
}

migrating Data

// V1

import (
    "github.com/aws/aws-sdk-go/aws"
    "github.com/aws/aws-sdk-go/aws/request"
    "github.com/aws/aws-sdk-go/aws/session"
    "github.com/aws/aws-sdk-go/service/s3"
)

func readPutObjectOutput(r *request.Request) {
        output, ok := r.Data.(*s3.PutObjectOutput)
        if !ok {
            return
        }

        // ...
    }
}

func main() {
    sess := session.Must(session.NewSession())
    sess.Handlers.Unmarshal.PushBack(readPutObjectOutput)

    svc := s3.New(sess)
    // ...
}
// V2

import (
    "context"

    "github.com/aws/aws-sdk-go-v2/config"
    "github.com/aws/aws-sdk-go-v2/service/s3"
    "github.com/aws/smithy-go/middleware"
    smithyhttp "github.com/aws/smithy-go/transport/http"
)

type readPutObjectOutput struct{}

var _ middleware.DeserializeMiddleware = (*readPutObjectOutput)(nil)

func (*readPutObjectOutput) ID() string {
    return "readPutObjectOutput"
}

func (*readPutObjectOutput) HandleDeserialize(ctx context.Context, in middleware.DeserializeInput, next middleware.DeserializeHandler) (
    out middleware.DeserializeOutput, metadata middleware.Metadata, err error,
) {
    out, metadata, err = next.HandleDeserialize(ctx, in)
    if err != nil {
        // ...
    }

    output, ok := in.Parameters.(*s3.PutObjectOutput)
    if !ok {
        return out, metadata, err
    }

    // inspect output...

    return out, metadata, err
}

func WithReadPutObjectOutput(o *s3.Options) {
    o.APIOptions = append(o.APIOptions, func (s *middleware.Stack) error {
        return s.Initialize.Add(&withReadPutObjectOutput{}, middleware.Before)
    })
}

func main() {
    cfg, err := config.LoadDefaultConfig(context.Background())
    if err != nil {
        // ...
    }

    svc := s3.NewFromConfig(cfg, WithReadPutObjectOutput)
    // ...
}

HTTP request/response

The HTTPRequest and HTTPResponse fields from Request are now exposed in specific middleware phases. Since middleware is transport-agnostic, you must perform a type assertion on the middleware input or output to reveal the underlying HTTP request or response.

Request handlers which reference Request.HTTPRequest and Request.HTTPResponse must be migrated to middleware.

migrating HTTPRequest

// V1

import (
    "github.com/aws/aws-sdk-go/aws/request"
    "github.com/aws/aws-sdk-go/aws/session"
)

func withHeader(header, val string) request.Option {
    return func(r *request.Request) {
        request.HTTPRequest.Header.Set(header, val)
    }
}

func main() {
    sess := session.Must(session.NewSession())
    sess.Handlers.Build.PushBack(withHeader("x-user-header", "..."))

    svc := s3.New(sess)
    // ...
}
// V2

import (
    "context"
    "fmt"

    "github.com/aws/aws-sdk-go-v2/config"
    "github.com/aws/aws-sdk-go-v2/service/s3"
    "github.com/aws/smithy-go/middleware"
    smithyhttp "github.com/aws/smithy-go/transport/http"
)

type withHeader struct {
    header, val string
}

// implements middleware.BuildMiddleware, which runs AFTER a request has been
// serialized and can operate on the transport request
var _ middleware.BuildMiddleware = (*withHeader)(nil)

func (*withHeader) ID() string {
    return "withHeader"
}

func (m *withHeader) HandleBuild(ctx context.Context, in middleware.BuildInput, next middleware.BuildHandler) (
    out middleware.BuildOutput, metadata middleware.Metadata, err error,
) {
    req, ok := in.Request.(*smithyhttp.Request)
    if !ok {
        return out, metadata, fmt.Errorf("unrecognized transport type %T", in.Request)
    }

    req.Header.Set(m.header, m.val)
    return next.HandleBuild(ctx, in)
}

func WithHeader(header, val string) func (*s3.Options) {
    return func(o *s3.Options) {
        o.APIOptions = append(o.APIOptions, func (s *middleware.Stack) error {
            return s.Build.Add(&withHeader{
                header: header,
                val: val,
            }, middleware.After)
        })
    }
}

func main() {
    cfg, err := config.LoadDefaultConfig(context.Background())
    if err != nil {
        // ...
    }

    svc := s3.NewFromConfig(cfg, WithHeader("x-user-header", "..."))
    // ...
}

Handler phases

SDK v2 middleware phases are the successor to v1 handler phases.

The following table provides a rough mapping of v1 handler phases to their equivalent location within the V2 middleware stack:

v1 handler name v2 middleware phase
Validate Initialize
Build Serialize
Sign Finalize
Send n/a (1)
ValidateResponse Deserialize
Unmarshal Deserialize
UnmarshalMetadata Deserialize
UnmarshalError Deserialize
Retry Finalize, after "Retry" middleware (2)
AfterRetry Finalize, before "Retry" middleware, post-next.HandleFinalize() (2,3)
CompleteAttempt Finalize, end of step
Complete Initialize, start of step, post-next.HandleInitialize() (3)

(1) The Send phase in v1 is effectively the wrapped HTTP client round-trip in v2. This behavior is controlled by the HTTPClient field on client options.

(2) Any middleware after the "Retry" middleware in the Finalize step will be part of the retry loop.

(3) The middleware “stack” at operation time is built into a repeatedly-decorated handler function. Each handler is responsible for calling the next one in the chain. This implicitly means that a middleware step can also take action AFTER its next step has been called.

For example, for the Initialize step, which is at the top of the stack, this means Initialize middlewares that take action after calling the next handler effectively operate at the end of the request:

// V2

import (
    "context"

    "github.com/aws/smithy-go/middleware"
)

type onComplete struct{}

var _ middleware.InitializeMiddleware = (*onComplete)(nil)

func (*onComplete) ID() string {
    return "onComplete"
}

func (*onComplete) HandleInitialize(ctx context.Context, in middleware.InitializeInput, next middleware.InitializeHandler) (
    out middleware.InitializeOutput, metadata middleware.Metadata, err error,
) {
    out, metadata, err = next.HandleInitialize(ctx, in)

    // the entire operation was invoked above - the deserialized response is
    // available opaquely in out.Result, run post-op actions here...

    return out, metadata, err
}

Features

Amazon EC2 Instance Metadata Service

The AWS SDK for Go V2 provides an Amazon EC2 Instance Metadata Service (IMDS) client that you can use to query the local IMDS when executing your application on an Amazon EC2 instance. The IMDS client is a separate Go module that can be added to your application by using

go get github.com/aws/aws-sdk-go-v2/feature/ec2/imds

The client constructor and method operations have been updated to match the design of the other SDK service clients.

Example

// V1

import "github.com/aws/aws-sdk-go/aws/ec2metadata"

// ...

client := ec2metadata.New(sess)

region, err := client.Region()
if err != nil {
	// handle error
}
// V2

import "context"
import "github.com/aws/aws-sdk-go-v2/feature/ec2/imds"

// ...

client := imds.NewFromConfig(cfg)

region, err := client.GetRegion(context.TODO())
if err != nil {
	// handle error
}

Amazon S3 Transfer Manager

The Amazon S3 transfer manager is available for managing uploads and downloads of objects concurrently. This package is located in a Go module outside the service client import path. This module can be retrieved by using go get github.com/aws/aws-sdk-go-v2/feature/s3/manager.

s3.NewUploader and s3.NewUploaderWithClient have been replaced with the constructor method manager.NewUploader for creating an Upload manager client.

s3.NewDownloader and s3.NewDownloaderWithClient have been replaced with a single constructor method manager.NewDownloader for creating a Download manager client.

Amazon CloudFront Signing Utilities

The AWS SDK for Go V2 provides Amazon CloudFront signing utilities in a Go module outside the service client import path. This module can be retrieved by using go get.

go get github.com/aws/aws-sdk-go-v2/feature/cloudfront/sign

Amazon S3 Encryption Client

Starting in AWS SDK for Go V2, the Amazon S3 encryption client is a separate module under AWS Crypto Tools. The latest version of the S3 encryption client for Go, 3.x, is now available at https://github.com/aws/amazon-s3-encryption-client-go. This module can be retrieved by using go get:

go get github.com/aws/amazon-s3-encryption-client-go/v3

The separate EncryptionClient (v1, v2) and DecryptionClient (v1, v2) APIs have been replaced with a single client, S3EncryptionClientV3, that exposes both encrypt and decrypt functionality.

Like other service clients in AWS SDK for Go V2, the operation APIs have been condensed:

  • The GetObject, GetObjectRequest, and GetObjectWithContext decryption APIs are replaced by GetObject.
  • The PutObject, PutObjectRequest, and PutObjectWithContext encryption APIs are replaced by PutObject.

To learn how to migrate to the 3.x major version of the encryption client, see this guide.