I am trying to access AWS s3 bucket from Azure HDInsight Cluster VM. I generated new keys and added to .aws/credentials. "aws s3 ls" is working fine in Azure VM. If I do hadoop distcp or read a s3 file in spark-shell, I am getting 403 error "AWS Access Key Id you provided does not exist in our records.". I did try exporting AWS_SESSION_TOKEN,AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. but no luck. Please help me resolve this issue.
1 Answer
I found a fix for this issue. I added the following properties in core-site.xml.
Property1:
Name: fs.s3a.aws.credentials.provider
value: org.apache.hadoop.fs.s3a.TemporaryAWSCredentialsProvider
Property2:
Name: fs.s3a.access.key
Value: XXXXXXXXX
Property3:
Name: fs.s3a.secret.key
Value: XXXXXXX
Property4:
Name:fs.s3a.session.token
Value: XXXXX
Initially I was trying to access s3 without the property "fs.s3a.aws.credentials.provider" from spark-shell