Skip to content

AWS Lambda script to ship data from S3 to other data stores, like Elasticsearch

License

Notifications You must be signed in to change notification settings

arithmetric/lambda-stash

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

61 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Lambda Stash

npm version Travis CI test status Coverage Status

lambda-stash is an AWS Lambda script for shipping data from S3 or other cloud data sources to data stores, like Elasticsearch.

Features

  • Supported input formats: AWS Cloudfront access logs, AWS Cloudtrail API logs, AWS CloudWatch logs, AWS Config change logs, AWS Elastic Load Balancer access logs, AWS S3 access logs, and other formats by implementing a custom handler.

  • Supported output formats: JSON, plain text key-value pairs.

  • Supported shipping methods: Elasticsearch (includes signing for AWS domains), HTTP/S POST, TCP socket.

Set Up

  1. Set up a Lambda deployment package that includes lambda-stash and a script with your configuration. See the example included (under example/) and the configuration documentation below to get started.

  2. Use the AWS Management Console to create a Lambda function using the Node.js 12.x runtime. Upload your package, configure it with event sources as desired (S3 buckets or CloudWatch logs). Be sure the Lambda function has an IAM role with any necessary permissions, like getting data from an S3 bucket or accessing an AWS Elasticsearch domain.

  3. Check CloudWatch Logs, where a Log Group should be created once the Lambda script runs the first time. Review the logs to make sure the script finishes successfully within the time allowed. You can set up a CloudWatch Alarm to be notified if an error occurs and to track for how long the function is running. The elasticsearch client timeout default is 30 seconds. You can set config.elasticsearch.requestTimeout to milliseconds or 'Infinity'.

Configuration

lambda-stash is intended to be implemented by a script that provides the configuration, such as what processors to use for specific events.

See the included example in example/ and the handlers documentation below for details on how to implement the handlers.

Handlers

convertString

Converts an array of objects to key-value strings with the format: prefix key1="value1" key2="value2" ... suffix

Inputs:

  • config.data should be an array of data objects.
  • config.string.prefix (optional) is a string to add at the beginning of each output string.
  • config.string.suffix (optional) is a string to add at the end of each output string.

Outputs:

  • config.data is transformed to a string with the result of the handler.

decodeBase64

Decodes a Base64 encoded string.

Inputs:

  • config.data should be a string in Base64 format.

Outputs:

  • config.data is transformed to a buffer with the decoded data.

decompressGzip

Decompresses gzipped data.

Inputs:

  • config.data should be a buffer in gzip format.

Outputs:

  • config.data is transformed to a string with the decompressed data.

formatCloudfront

Processes parsed data from Cloudfront access logs and normalizes it in key-value objects.

Raw Cloudfront access log files in S3 should be processed with decompressGzip and parseTabs before using this format handler.

Inputs:

  • config.data should be an array of log records, each of which is an array of field values.

  • config.dateField (optional) is the key name in the output that should contain the log's date time value (combined from the date and time fields).

Outputs:

  • config.data is transformed to an array of objects with log data.

formatCloudtrail

Processes parsed data from Cloudtrail logs and normalizes it in key-value objects.

Raw Cloudtrail access log files in S3 should be processed with decompressGzip and parseJson before using this format handler.

Inputs:

  • config.data should have a Records property that is an array of objects for each log item.

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

formatCloudwatchLogs

Processes parsed data from CloudWatch logs and normalizes it in key-value objects. Includes handling for Lambda function start, end, and report logs.

This handler is meant to be used with data provided through a CloudWatch Logs subscription event, which is automatically processed with decodeBase64, decompressGzip, and parseJson.

Inputs:

  • config.data should have a logEvents property that is an array of objects for each log item.

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

formatConfig

Processes parsed data from AWS Config logs and normalizes it in key-value objects.

Raw Config log files in S3 should be processed with decompressGzip and parseJson before using this format handler.

Inputs:

  • config.data should have a configurationItems property that is an array of objects for each log item.

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

formatELBv1

Processes parsed data from AWS Elastic Load Balancer (ELB) Classic Load Balancer (ELB version 1) logs and normalizes it in key-value objects.

Raw ELB log files in S3 should be processed with parseSpaces before using this format handler.

Inputs:

  • config.data should be an array (one item per log record) of arrays (one item per record field).

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

formatELBv2

Processes parsed data from AWS Elastic Load Balancer (ELB) Application Load Balancer (ELB version 2) logs and normalizes it in key-value objects.

Raw ELB log files in S3 should be processed with parseSpaces before using this format handler.

Inputs:

  • config.data should be an array (one item per log record) of arrays (one item per record field).

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

formatS3Access

Processes parsed data from AWS Simple Storage Service (S3) access logs and normalizes it in key-value objects.

Raw S3 log files should be processed with parseSpaces before using this format handler.

Inputs:

  • config.data should be an array (one item per log record) of arrays (one item per record field).

  • config.dateField (optional) is the key name in the output that should contain the log's date time value.

Outputs:

  • config.data is transformed to an array of objects with log data.

getS3Object

Fetches an object from S3.

Inputs:

  • config.S3.srcBucket is the S3 bucket name for the object to fetch.

  • config.S3.srcKey is the S3 key name for the object to fetch.

Outputs:

  • config.data is set to the buffer with the S3 object data.

outputJsonLines

Transforms output data into a series of JSON strings delimited by a newline. This output format is compatible with the Loggly HTTP/S bulk endpoint.

Inputs:

  • config.data is an array of objects with log data.

Outputs:

  • config.data is transformed into JSON strings delimited by newlines.

parseCsv

Parses CSV data into an array of records, each of which is an array of column values.

Inputs:

  • config.data is a string with CSV data.

Outputs:

  • config.data is transformed into an array of records, each of which is an array of strings with column values.

parseJson

Parses a JSON string into its equivalent JavaScript object.

Inputs:

  • config.data is a JSON string.

Outputs:

  • config.data is the equivalent JavaScript object.

parseTabs

Parses tab separated value data into an array of records, each of which is an array of column values.

Inputs:

  • config.data is a string with tab separated data.

Outputs:

  • config.data is transformed into an array of records, each of which is an array of strings with column values.

shipElasticsearch

Ships the data to an Elasticsearch index. The Elasticsearch _index and _type values are configurable, and support for the AWS Elasticsearch request signing process is provided.

Inputs:

  • config.data is an array of objects with log data.

  • config.elasticsearch.host is an HTTP/S URL to the Elasticsearch domain to which to send data.

  • config.elasticsearch.index is a string with the _index value to add to log objects.

  • config.elasticsearch.type is a string with the _type value to add to log objects.

  • config.elasticsearch.useAWS should be set to true to connect with an AWS Elasticsearch domain and automatically sign requests for AWS.

  • config.elasticsearch.region is the AWS region where the AWS Elasticsearch domain is hosted.

  • config.elasticsearch.accessKey (optional) is the AWS access key to use for signing the request. If an access key and secret key are not provided, the environment credentials are used instead (i.e. the IAM role under which the Lambda function is running).

  • config.elasticsearch.secretKey (optional) is the AWS secret key to use for signing the request.

  • config.elasticsearch.maxChunkSize (optional) is the maximum number of log items to ship in one request. By default, set to 1000.

  • config.elasticsearch.requestTimeout (optional) is the Elasticsearch client timeout in milliseconds or can be set to Infinity for no timeout. The Elasticsearch client's default is 30000 (30 seconds).

shipHttp

Ships the data to an HTTP/S endpoint.

Inputs:

  • config.data is the string data to ship. An alternate key can be configured.

  • config.http.url is an HTTP/S URL to which to send data.

  • config.http.keyData (optional) is the key name in the config object of the data to ship. By default, set to data.

shipTcp

Ships the data to a TCP service.

Inputs:

  • config.data is the string data to ship. An alternate key can be configured.

  • config.tcp.host is a hostname to which to open a TCP socket.

  • config.tcp.port is a port on which to open the socket.

  • config.tcp.keyData (optional) is the key name in the config object of the data to ship. By default, set to data.

Recipes

Ship Cloudtrail logs to AWS Elasticsearch

The following script can be used to ship Cloudtrail logs from S3 to an AWS Elasticsearch instance.

var shipper = require('lambda-stash');
var config = {
  elasticsearch: {
    host: 'https://search-test-es-abcdef...',
    region: 'us-east-1',
    useAWS: true
  },
  mappings: [
    {
      bucket: 'my-cloudtrail-logs',
      processors: [
        'decompressGzip',
        'parseJson',
        'formatCloudtrail',
        'shipElasticsearch'
      ],
      elasticsearch: {
        index: 'logs',
        type: 'cloudtrail'
      },
      dateField: 'date'
    }
  ]
};
shipper.handler(config, event, context, callback);

About

AWS Lambda script to ship data from S3 to other data stores, like Elasticsearch

Resources

License

Stars

Watchers

Forks