Identifying Amazon S3 requests using CloudTrail

In Amazon S3, you can identify requests using an Amazon CloudTrail event log. Amazon CloudTrail is the preferred way of identifying Amazon S3 requests, but if you are using Amazon S3 server access logs, see Using Amazon S3 access logs to identify requests.

Identifying requests made to Amazon S3 in a CloudTrail log

Events logged by CloudTrail are stored as compressed, gzipped JSON objects in your S3 bucket. To efficiently find requests, you should use a service like Amazon Athena to index and query the CloudTrail logs. For more information about CloudTrail and Athena, see Querying Amazon CloudTrail Logs in the Amazon Athena User Guide.

Using Athena with CloudTrail logs

After you set up CloudTrail to deliver events to a bucket, you should start to see objects go to your destination bucket on the Amazon S3 console. These are formatted as follows:

s3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/Region/yyyy/mm/dd

Identifying Amazon S3 Signature Version 2 requests by using CloudTrail

You can use a CloudTrail event log to identify which API signature version was used to sign a request in Amazon S3. This capability is important because support for Signature Version 2 will be turned off (deprecated). After that, Amazon S3 will no longer accept requests that use Signature Version 2, and all requests must use Signature Version 4 signing.

We strongly recommend that you use CloudTrail to help determine whether any of your workflows are using Signature Version 2 signing. Remediate them by upgrading your libraries and code to use Signature Version 4 instead to prevent any impact to your business.

For more information, see Announcement: Amazon CloudTrail for Amazon S3 adds new fields for enhanced security auditing in Amazon Web Services re:Post.

Amazon CloudTrail is the preferred method for identifying Signature Version 2 requests. If you're using Amazon S3 server-access logs, see Identifying Signature Version 2 requests using Amazon S3 access logs.

Athena query examples for identifying Amazon S3 Signature Version 2 requests

SELECT EventTime, EventName as S3_Action, requestParameters as Request_Parameters, awsregion as AWS_Region, sourceipaddress as Source_IP, useragent as User_Agent
FROM s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE eventsource='s3.amazonaws.com.cn'
AND json_extract_scalar(additionalEventData, '$.SignatureVersion')='SigV2'
LIMIT 10;
                

SELECT useridentity.arn, Count(requestid) as RequestCount
FROM s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE eventsource='s3.amazonaws.com.cn'
    and json_extract_scalar(additionalEventData, '$.SignatureVersion')='SigV2'
Group by useridentity.arn
                

Partitioning Signature Version 2 data

If you have a large amount of data to query, you can reduce the costs and running time of Athena by creating a partitioned table.

To do this, create a new table with partitions as follows.

   CREATE EXTERNAL TABLE s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table_partitioned(
        eventversion STRING,
        userIdentity STRUCT<
            type:STRING,
            principalid:STRING,
            arn:STRING,
            accountid:STRING,
            invokedby:STRING,
            accesskeyid:STRING,
            userName:STRING,
         sessioncontext:STRUCT<
                    attributes:STRUCT< 
                    mfaauthenticated:STRING,
                    creationdate:STRING>,
                    sessionIssuer:STRUCT<
                    type:STRING,
                    principalId:STRING,
                    arn:STRING,
                    accountId:STRING,
                    userName:STRING>
                >
             >,
        eventTime STRING,
        eventSource STRING,
        eventName STRING,
        awsRegion STRING,
        sourceIpAddress STRING,
        userAgent STRING,
        errorCode STRING,
        errorMessage STRING,
        requestParameters STRING,
        responseElements STRING,
        additionalEventData STRING,
        requestId STRING,
        eventId STRING,
        resources ARRAY<STRUCT<ARN:STRING,accountId: STRING,type:STRING>>, 
        eventType STRING,
        apiVersion STRING,
        readOnly STRING,
        recipientAccountId STRING,
        serviceEventDetails STRING,
        sharedEventID STRING,
        vpcEndpointId STRING
    )   
    PARTITIONED BY (region string, year string, month string, day string)
    ROW FORMAT SERDE 'org.apache.hadoop.hive.ql.io.orc.OrcSerde' 
    STORED AS INPUTFORMAT 'org.apache.hadoop.hive.ql.io.SymlinkTextInputFormat'
    OUTPUTFORMAT 'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat'
    LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/';
                

Then, create the partitions individually. You can't get results from dates that you haven't created.

ALTER TABLE s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table_partitioned ADD
    PARTITION (region= 'us-east-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/us-east-1/2019/02/19/'
    PARTITION (region= 'us-west-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/us-west-1/2019/02/19/'
    PARTITION (region= 'us-west-2', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/us-west-2/2019/02/19/'
    PARTITION (region= 'ap-southeast-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/ap-southeast-1/2019/02/19/'
    PARTITION (region= 'ap-southeast-2', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/ap-southeast-2/2019/02/19/'
    PARTITION (region= 'ap-northeast-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/ap-northeast-1/2019/02/19/'
    PARTITION (region= 'eu-west-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/eu-west-1/2019/02/19/'
    PARTITION (region= 'sa-east-1', year= '2019', month= '02', day= '19') LOCATION 's3://DOC-EXAMPLE-BUCKET1/AWSLogs/111122223333/CloudTrail/sa-east-1/2019/02/19/';
                

You can then make the request based on these partitions, and you don't need to load the full bucket.

SELECT useridentity.arn,
Count(requestid) AS RequestCount
FROM s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table_partitioned
WHERE eventsource='s3.amazonaws.com.cn'
AND json_extract_scalar(additionalEventData, '$.SignatureVersion')='SigV2'
AND region='us-east-1'
AND year='2019'
AND month='02'
AND day='19'
Group by useridentity.arn
                

Identifying access to S3 objects by using CloudTrail

You can use your Amazon CloudTrail event logs to identify Amazon S3 object access requests for data events such as GetObject, DeleteObject, and PutObject, and discover additional information about those requests.

The following example shows how to get all PUT object requests for Amazon S3 from an Amazon CloudTrail event log.

Athena query examples for identifying Amazon S3 object access requests

In the following Athena query examples, replace s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table with your Athena details, and modify the date range as needed.

SELECT
  eventTime, 
  eventName, 
  eventSource, 
  sourceIpAddress, 
  userAgent, 
  json_extract_scalar(requestParameters, '$.bucketName') as bucketName, 
  json_extract_scalar(requestParameters, '$.key') as object,
  userIdentity.arn as userArn
FROM
  s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE
  eventName = 'PutObject'
  AND eventTime BETWEEN '2019-07-05T00:00:00Z' and '2019-07-06T00:00:00Z'
                    

SELECT
  eventTime, 
  eventName, 
  eventSource, 
  sourceIpAddress, 
  userAgent, 
  json_extract_scalar(requestParameters, '$.bucketName') as bucketName, 
  json_extract_scalar(requestParameters, '$.key') as object,
  userIdentity.arn as userArn
FROM
  s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE
  eventName = 'GetObject'
  AND eventTime BETWEEN '2019-07-05T00:00:00Z' and '2019-07-06T00:00:00Z'

SELECT
  eventTime, 
  eventName, 
  eventSource, 
  sourceIpAddress, 
  userAgent, 
  json_extract_scalar(requestParameters, '$.bucketName') as bucketName, 
  userIdentity.arn as userArn,
  userIdentity.accountId
FROM
  s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE
  userIdentity.accountId = 'anonymous'
  AND eventTime BETWEEN '2019-07-05T00:00:00Z' and '2019-07-06T00:00:00Z'

SELECT
  eventTime, 
  eventName, 
  eventSource, 
  sourceIpAddress, 
  userAgent, 
  userIdentity.arn as userArn,
  json_extract_scalar(requestParameters, '$.bucketName') as bucketName,
  json_extract_scalar(requestParameters, '$.key') as object,
  json_extract_scalar(additionalEventData, '$.aclRequired') as aclRequired
FROM 
  s3_cloudtrail_events_db.cloudtrail_DOC-EXAMPLE-BUCKET1_table
WHERE
  json_extract_scalar(additionalEventData, '$.aclRequired') = 'Yes'
  AND eventTime BETWEEN '2022-05-10T00:00:00Z' and '2022-08-10T00:00:00Z'

If you are using Amazon S3 server access logs, see Identifying object access requests using Amazon S3 access logs.