Migrate HyperShift's AWS SDK dependencies from v1 to v2 to ensure continued security support, vendor maintenance, and compliance before the July 31, 2025 end-of-life deadline.

Market Problem

AWS is ending support for AWS SDK for Go v1 on July 31, 2025. After this date, SDK v1 will no longer receive:

• Security patches for discovered vulnerabilities
• Bug fixes for defects
• Updates for new AWS services or features
• Vendor support from AWS

HyperShift currently depends on AWS SDK v1 across 36 production code files representing approximately ~10,000 lines of AWS-dependent code. This creates critical risks:

Who is affected:

• ROSA HCP (Red Hat OpenShift Service on AWS - Hosted Control Planes) customers
• ARO HCP (Azure Red Hat OpenShift - Hosted Control Planes) customers using AWS dependencies
• Self-managed HyperShift deployments on AWS

Impact of not migrating:

• Security risk:* Post EOL vulnerabilities in AWS SDK v1
• Compliance risk:* Security audits will flag unsupported dependencies
• Technical debt:* Migration becomes increasingly costly as codebase evolves
• Support burden:* AWS will not address SDK v1 issues after EOL

This is a time-critical engineering investment required to maintain security posture and vendor support for all AWS-based hosted control plane offerings.

Proposed Solution

Complete migration of HyperShift's AWS SDK usage from v1 to v2 before the July 31, 2025 EOL deadline, including:

Core Migration:

• Migrate all 36 production code files using AWS SDK v1 to v2 equivalents
• Update configuration patterns from session-based to context-based
• Modernize service clients to use v2 modular design
• Refactor error handling to use v2 error types (smithy.APIError)
• Update credential providers to v2 interfaces
• Migrate paginators and waiters to v2 explicit constructors
• Update testing infrastructure and mocks to v2 patterns

Dependency Updates:

• Update Cluster API Provider AWS (CAPA) to v2-compatible version
• Verify controller-runtime AWS SDK compatibility

Quality Assurance:

• Comprehensive unit test updates across all affected components
• Full E2E test validation on AWS platform
• Performance benchmarking to ensure no regressions
• Security scanning of new SDK version

Strategic Value

Customer Value

• Security assurance:* Continued security patching protects customer infrastructure
• Compliance:* Meets security audit requirements for supported dependencies
• Reliability:* Access to AWS bug fixes maintains cluster stability
• Innovation:* Enables future use of new AWS services and features

Business Impact

• Risk mitigation:* Prevents security incidents from unpatched SDK vulnerabilities
• Compliance:* Maintains SOC2, ISO27001, FedRAMP compliance postures
• Support efficiency:* AWS support available for SDK-related issues
• Cost avoidance:* Prevents emergency migration costs if security issue discovered post-EOL

Success Criteria

Completion

• Zero AWS SDK v1 imports remain in HyperShift codebase
• All CAPA dependencies updated to v2-compatible versions
• Migration completed and shipped before July 31, 2025

Quality

• 100% unit test pass rate with AWS SDK v2
• 100% E2E test pass rate for AWS platform (ROSA HCP, standalone)
• Zero performance regression in AWS operations (EC2, S3, Route53, etc.)
• Zero functional regressions reported in QE or production

Adoption

• Target OpenShift release (4.20 or earlier) ships with migrated code
• All customer deployments receive migrated version through normal upgrade path
• Documentation updated with v2 usage patterns

Security

• Security scan shows zero critical/high vulnerabilities in AWS SDK dependency
• Compliance teams confirm supported dependency status
• No security incidents related to AWS SDK dependencies post-migration

Scope

Affected Components (36 files)

Critical Production Components:

1. CLI Infrastructure Management (cmd/infra/aws/)

• Infrastructure creation/destruction workflows
• AWS services: EC2, Route53, S3, IAM, ELB/ELBv2, RAM, STS
• Components: VPC/subnet/security group creation, NAT gateways, DNS zone management, IAM role/policy provisioning, load balancer setup
• Impact: Core hypershift CLI commands for AWS infrastructure provisioning

2. HyperShift Operator - AWS Platform Controller

• hypershift-operator/controllers/platform/aws/ - AWS Endpoint Service management
• hypershift-operator/controllers/hostedcluster/ - S3 OIDC bucket operations
• hypershift-operator/controllers/nodepool/ - EC2 subnet validation and instance metrics
• AWS services: EC2, S3, ELBv2
• Impact: Core reconciliation logic for AWS-hosted clusters (ROSA HCP, standalone AWS)

3. Control Plane Operator - Private Link Controller

• control-plane-operator/controllers/awsprivatelink/ - VPC Endpoint Service management
• control-plane-operator/controllers/awsprivatelink/route53.go - Private DNS integration
• AWS services: EC2, Route53, KMS
• Impact: Critical for Private cluster functionality - affects PrivateLink-based customer deployments

4. Shared AWS Utilities (support/awsutil/)

• STS role assumption with web identity federation
• AWS error code handling and retry logic
• Security group operations
• Impact: Foundation utilities used across all AWS components

5. Karpenter Operator - Machine Approver

• karpenter-operator/controllers/karpenter/machine_approver.go
• EC2 instance verification for auto-scaling
• Impact: NodePool auto-scaling with Karpenter

6. Etcd Backup Service

• etcd-backup/etcdbackup.go - S3 snapshot uploads
• Impact: Cluster backup and disaster recovery

7. CLI Utilities

• Bastion host creation/destruction
• Console log retrieval from EC2 instances
• Impact: Debugging and troubleshooting workflows

8. E2E Test Infrastructure (9 files)

• AWS resource creation/cleanup for tests
• Shared OIDC provider setup
• KMS encryption validation, tag propagation testing
• Impact: CI/CD test reliability and coverage

9. Contrib Tools

• Zone cleanup utilities
• Impact: Operations and maintenance scripts

AWS Services Utilized (7 services)

• EC2 (16 files) - Instance management, VPC/networking, security groups, volumes
• Route53 (6 files) - DNS zones, record sets, private hosted zones
• S3 (5 files) - OIDC provider buckets, etcd backup storage, multipart uploads
• IAM (4 files) - Roles, policies, OIDC identity providers
• ELBv2 (4 files) - Application/Network Load Balancers, Endpoint Services
• ELB Classic (2 files) - Legacy load balancer operations
• Additional: KMS (encryption), RAM (shared VPC), STS (role assumption), Resource Groups Tagging API

Technical Patterns that might require migration

• Session Management:* aws/session package usage across all components
• Credentials:* Static credentials, STS web identity, role assumption patterns
• Error Handling:* awserr package for error code extraction and retry logic
• Testing Interfaces: Heavy use of _iface packages (ec2iface, s3iface, etc.) for mocking
• Request Customization:* Custom user agent headers ("openshift.io/hypershift"), retry handlers
• S3 Multipart:* s3/s3manager for efficient large file uploads

Epics (Planned)

Phase 1: Test Infrastructure & CLI tools

• Epic 1: Migrate E2E test infrastructure to validate patterns and create examples
• Epic 2: Migrate CLI infrastructure commands (cmd/infra/aws/) - largest code surface

Phase 2: Operators & supporting components

• Epic 3: Migrate HyperShift operator AWS controllers
• Epic 4: Migrate Control Plane operator AWS controllers (PrivateLink, Route53)
• Epic 5: etcd-backup, contrib tools, aws-encryption-provider

Phase 3: Docs

• Epic 6: Update dev documentation, migration guides, developer onboarding materials

Out of Scope

• Feature additions beyond migration requirements
• Performance optimizations not directly related to SDK migration
• Changes to non-AWS platform support (Azure, PowerVS, KubeVirt, OpenStack, Agent)
• Refactoring of AWS infrastructure management patterns (keep existing patterns, just update SDK)

Dependencies

Hard Deadline:

• July 31, 2025: AWS SDK v1 end-of-life - MUST ship before this date

References

• AWS SDK v1 EOL Announcement: https://aws.amazon.com/blogs/developer/announcing-end-of-support-for-aws-sdk-for-go-v1-on-july-31-2025/
• AWS SDK v2 Migration Guide: https://docs.aws.amazon.com/sdk-for-go/v2/developer-guide/migrate-gosdk.html
• CAPA Upstream Issue: https://github.com/kubernetes-sigs/cluster-api-provider-aws/issues/2225
• Original Epic: CNTRLPLANE-886
• Current Working-In-Progress Doc