AWS-LA (AWS Log Analyzer)

Description

This script allows you to easily import various AWS log types into an Elasticsearch cluster running locally on your computer in a docker container.

Supported AWS Log Types

• ELB access logs
• ALB access logs
• VPC flow logs
• Route53 query logs
• Apache access logs

Future Log Types Supported

• Cloudtrail audit logs
• Cloudfront access logs
• S3 access logs
• Others to come!

Steps Automated

The script configures everything that is needed in the ELK stack:

• Elasticsearch:

  • indices
  • mappings
  • ingest pipelines

• Kibana:

  • index-patterns
  • field formatting for index-pattern fields
  • importing dashboards, visualizations, and dashboards
  • custom link directly to the newly created dashboard

Installation Steps

• Install Docker for Windows or Docker for Mac

• Clone this git repository:

  git clone https://github.com/mike-mosher/aws-la.git && cd aws-la

• Install requirements:

  pip install -r ./requirements.txt

Running the Script

• Bring the docker environment up:

  docker-compose up -d

• Verify that the containers are running:

  docker ps

• Verify that Elasticsearch is running:

  curl -XGET localhost:9200/_cluster/health?pretty

• To run the script, specify the log type and directory containing the logs. For example, you could run the following command to import ELB Access Logs

  python importLogs.py --logtype elb --logdir ~/logs/elblogs/

• Valid log types are specified by running the --help argument. Currently, the valid logtypes are the following:

  elb                 # ELB access logs
  alb                 # ALB access logs
  vpc                 # VPC flow logs
  r53                 # Route53 query logs
  apache              # apache access log ('access_log')
  apache_archives     # apache access logs (gunzip compressed with logrotate)

• Browse to the link provided in the output by using cmd + double-click, or browse directly to the default Kibana page:

  http://localhost:5601

• You can import multiple log types in the same ELK cluster. Just run the command again with the new log type and log directory:

  python importLogs.py --logtype vpc --logdir ~/logs/vpc-flowlogs/

• When done, you can shutdown the containers:

  docker-compose down -v

Screenshots / Examples:

• Python output:

  • As you can see, I was able to import 12.5 million VPC flowlogs in around 2 hours

• Searching for traffic initiated by RFC1918 (private) IP addresses:

  • Browse to Discover tab, and enter the following query in the query bar:

  source_ip_address:"10.0.0.0/8" OR source_ip_address:"172.16.0.0/12" OR source_ip_address:"192.168.0.0/16"

  

  • Alternately, you can search for all traffic initiated by Public IP addresses in the logs:

  NOT (source_ip_address:"10.0.0.0/8" OR source_ip_address:"172.16.0.0/12" OR source_ip_address:"192.168.0.0/16")

  

  • Search for a specific flow to/from a specific ENI:

  interface-id:<eni-name> AND (source_port:<port> OR dest_port:<port>)

  

  • Note: VPC Flow Logs split a flow into two log entries, so the above search will find both sides of the flow and show packets / bytes for each

• Dashboard imported for VPC Flow Logs:

• Dashboard imported for ALB Access Logs: