[jira] [Commented] (HADOOP-12806) Hadoop fs s3a lib not working with temporary credentials in AWS Lambda

["Chris Nauroth (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-12806?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15151086#comment-15151086 ] 

Chris Nauroth commented on HADOOP-12806:
----------------------------------------

[~nicktgr15], improving support with AWS Lambda sound great, but I don't think we'll be able to jump right to {{DefaultAWSCredentialsProviderChain}} due to backwards-compatibility concerns.  We definitely need to be able to keep supporting {{BasicAWSCredentialsProvider}}.  If you want to suggest another way to modify our chain to support AWS Lambda, then we could go in that direction.

We'd also likely need help with testing the various credential sources in different deployments.

> Hadoop fs s3a lib not working with temporary credentials in AWS Lambda 
> -----------------------------------------------------------------------
>
>                 Key: HADOOP-12806
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12806
>             Project: Hadoop Common
>          Issue Type: Improvement
>          Components: fs/s3
>    Affects Versions: 2.7.2
>            Reporter: Nikolaos Tsipas
>              Labels: aws-s3
>
> Trying to use the hadoop fs s3a library in AWS lambda with temporary credentials but it's not possible because of the way the {{AWSCredentialsProviderChain}} is defined under https://github.com/apache/hadoop/blob/29ae25801380b94442253c4202dee782dc4713f5/hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/S3AFileSystem.java
> Specifically the following code is used to initialise the credentials chain
> {code}
> AWSCredentialsProviderChain credentials = new AWSCredentialsProviderChain(
>         new BasicAWSCredentialsProvider(accessKey, secretKey),
>         new InstanceProfileCredentialsProvider(),
>         new AnonymousAWSCredentialsProvider()
>     );
> {code}
> The above works fine when the EC2 metadata endpoint is available (i.e. running on an EC2 instance) however it doesn't work properly when the environment variables  are used to define credentials as it happens in AWS Lambda. Amazon suggests to use the {{EnvironmentVariableCredentialsProvider}} in AWS Lambda. 
> To summarise and suggest an alternative I think that the {{DefaultAWSCredentialsProviderChain}} could be used instead of  the {{InstanceProfileCredentialsProvider}}  and that would cover the following cases: 
> {panel}
> * Environment Variables - AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY (RECOMMENDED since they are recognized by all the AWS SDKs and CLI except for .NET), or AWS_ACCESS_KEY and AWS_SECRET_KEY (only recognized by Java SDK)
> * Java System Properties - aws.accessKeyId and aws.secretKey
> * Credential profiles file at the default location (~/.aws/credentials) shared by all AWS SDKs and the AWS CLI
> * Instance profile credentials delivered through the Amazon EC2 metadata service
> {panel}
> If you think that the above change would be useful I could investigate more about what the required changes would be and submit a patch.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)