Discussion: Test Coverage for GossipSub 1.1 Upgrade

[seetadev]

@Winter-Soren : Great work on pushing the upgrade! To make sure we have full coverage on GossipSub 1.1 features, here’s a checklist of scenarios to validate. Some may already be in your test cases, but this ensures completeness:

---

Core Functionality

• Message propagation works as expected under normal mesh conditions.
• Interop with GossipSub 1.0 peers works, with graceful fallback.

---

🛠️ GossipSub 1.1 Specific

• Peer Scoring

  • Spamming invalid/duplicate messages decreases score and results in pruning.
  • Honest peers maintain stable/positive scores.

• IHAVE/IWANT Adaptive Gossip

  • Dropping gossip triggers IWANT requests for missing messages.
  • No infinite gossip loops observed.

• Opportunistic Grafting

  • Simulate degraded mesh quality (e.g. high-latency peers) → node grafts better peers into mesh.

• Flood Publishing

  • Publishing from a non-mesh peer still delivers messages reliably to connected peers.

• Invalid Behavior Rejection

  • Invalid signatures / malformed payloads are ignored and penalized.
  • Excessive IHAVE/IWANT spam penalized in scoring.

---

📊 Network-Level Scenarios

• Partition & Reconnect

  • Partition network, publish on both sides, reconnect → eventual mesh healing and message consistency.

• High Churn

  • Frequent connect/disconnect events → mesh stability (GRAFT/PRUNE) maintained.

• Large-scale Fanout

  • Stress test with 100+ peers to confirm gossip propagation, mesh stability, and scoring efficiency.

---

If we cover all of the above, py-libp2p’s GossipSub 1.1 behavior should align with the spec and be resilient under real-world conditions.

Ccing @pacrob and @acul71 to keep him in the loop of our discussion.