How qb411/fis-template-library works
Project type and purpose
• This is an open-source repository of AWS Fault Injection Service (FIS) experiments.
• Each subdirectory under the root (e.g., aurora-cluster-failover, ec2-instances-terminate, mysql-rds-loadtest-failover, etc.) represents a distinct FIS experiment: JSON/YAML definitions, IAM policies/trust relationships, and—in some cases—CloudFormation or SSM scripts.
File structure
• Top level:
• README.md, CONTRIBUTING, LICENSE – repo metadata.
• Directories for each experiment (aurora-cluster-failover, ec2-spot-interruption, etc.).
• templates/ – central AWSFIS.json template plus example IAM policy/trust files.
• Per-experiment directories contain:
• AWSFIS.json or FIS experiment templates (defines targets, actions, stop conditions).
• IAM policy files (.json) and trust-relationship definitions.
• In some cases CloudFormation (e.g., mysql- and postgres-rds-loadtest-failover) and SSM scripts.
• Architectural choice: modular “Infrastructure as Code” experiments that plug into AWS FIS; each module is self-contained.
README for additional insights
• Purpose: simulate failure scenarios for resilience testing.
• Usage flow: checkout repo → pick experiment → customize JSON/YAML → run via AWS CLI/SDK/Console → monitor → analyze.
a. Main components
• Developer workstation / CI pipeline (GitHub repo)
• AWS CLI/SDK/Console (invokes experiments)
• AWS FIS (central orchestrator)
• Shared Templates bucket or Code repository (templates/*.json)
• IAM Roles & Policies (experiment-specific)
• AWS Targets:
– EC2 Instances (terminate, spot interruption)
– RDS Instances (MySQL, PostgreSQL failover; Aurora clusters)
– SAP components (EBS volumes, ASCS, DATABASE EC2s)
– SSM agent on targets (for executing load-test scripts or custom actions)
• Optional deployment services: CloudFormation (for load-test stacks), Systems Manager (SSM)
b. Relationships & interactions
1. Developer/CI pushes or updates templates in GitHub.
2. Operator uses AWS CLI/Console to create a new FIS experiment referencing a local or S3-hosted JSON/YAML.
3. AWS FIS service assumes the experiment’s IAM role (trust relationship) and executes defined actions on target resources.
4. FIS actions call AWS APIs to pause volumes, stop instances, inject chaos events, or trigger SSM documents.
5. For load-test experiments, FIS triggers CloudFormation to deploy test harness and SSM to run scripts.
6. Metrics and logs flow into CloudWatch or SSM Session Manager for monitoring and analysis.
c. Architectural patterns & design principles
– Modular IaC: each experiment is self-contained to simplify reuse.
– Principle of least privilege: experiment-specific IAM policies limit FIS actions.
– Decoupling: templates directory provides base definitions; experiment folders override or extend.
– Infrastructure orchestration: FIS coordinates cross-service calls via IAM to minimize direct manual intervention.
d. Relevant technologies
– AWS FIS (service orchestrator)
– AWS IAM (policies & roles)
– AWS CloudFormation (deploy load-test stacks)
– AWS Systems Manager (execute scripts via SSM documents)
– AWS CLI/SDK/Console (trigger experiments)
– JSON/YAML templates (experiment definitions)