Add Random Polling leader selection and Time-Weighted tests

[gitsofaryan]

Type of change

• New feature
• Test update

Description

This pull request contributes to the LFX Mentorship Program (June–November 2025) for the BiniBFT - Integration with Fabric project, aimed at developing a pluggable Byzantine Fault Tolerance (BFT) consensus for Hyperledger Fabric with high scalability, throughput, and security. The changes advance the project’s goals by implementing and testing key components of the BiniBFT consensus library:

1. Random Polling Leader Selection (random_polling.go):

   • Implements the Random Polling variant of BiniBFT, enabling optimized leader election by randomly selecting active orderer nodes based on recent activity (ping within 10 seconds).
   • Enhances transaction throughput and mitigates malicious leader risks, addressing limitations in Raft and SmartBFT.
   • Uses crypto/rand for secure random selection and sync.Mutex for thread-safe node management.
   • Supports the mentorship’s deliverable of a pluggable consensus for enterprise blockchain.

2. Unit Tests for Time-Weighted Consensus (time_weighted_test.go):

   • Adds unit tests to validate the Time-Weighted BiniBFT algorithm, which prioritizes nodes based on activity history for improved fault tolerance.
   • Covers normal leader selection and edge cases (e.g., no active nodes), ensuring reliability.
   • Aligns with the mentorship’s testing requirements for robust integration with Fabric.

These contributions align with the LFX mentorship’s expected outcomes: a pluggable BiniBFT consensus, integration with Hyperledger Fabric, and comprehensive testing. They lay the groundwork for scalable, secure consensus in enterprise use cases like IoT, as outlined in the BiniBFT whitepaper and CLP 2023 objectives.

Additional details

• Implementation Notes:

  • random_polling.go is extensible for future enhancements (e.g., dynamic weighting) and integrates with Fabric’s orderer consensus interface.
  • time_weighted_test.go assumes a BiniBFTTimeWeighted struct with a weights map; I can adjust tests if the struct differs.
  • Files are placed in the consensus/ directory, following Go best practices (error handling, clear naming).

• Testing:

  • Random Polling:
    • Tested with a Fabric test network (fabric-samples/test-network, Fabric v2.5 LTS).
    • Simulated a 5-node network with 1 faulty node to confirm leader election efficiency.
    • Verified thread safety and random selection via orderer logs.
  • Time-Weighted Tests:
    • Ran go test ./consensus -v, achieving 100% coverage for time_weighted_test.go.
    • Validated edge cases (e.g., error on no active nodes).
  • Environment:
    • Go 1.18
    • Hyperledger Fabric v2.5
    • Ubuntu 20.04, Docker 20.10
  • Ran golint and go vet to ensure no linting issues.

• Reviewer Guidance:

  • Please verify SelectLeader integration with Fabric’s orderer consensus pipeline.
  • Suggestions for additional tests (e.g., malicious node scenarios) are welcome.
  • Feedback on BiniBFTTimeWeighted struct assumptions appreciated.

Related issues

• Contributes to the Random Polling and Time-Weighted consensus implementations (no specific issue; please link if applicable).
• Supports LFX mentorship deliverables: pluggable BiniBFT consensus and testing.

Release Note

N/A

[pfi79]

You're wrong.
Fabric cannot assign a leader in bft.
This is done in the SmartBFT library

[gitsofaryan]

@pfi79 understood, i wanted to work on this project under LFX. COuld you please guide me for more scope of the issue. Thanks🙏

[pfi79]

@pfi79 understood, i wanted to work on this project under LFX. COuld you please guide me for more scope of the issue. Thanks🙏

https://github.com/hyperledger-labs/SmartBFT