test: Add large-scale network tests (50 and 200 peers) for topology, subscriptions, and seeding

[sanity]

Problem

Current test coverage focuses on small networks (2-3 peers). The recent #2164 refactor series (PRs #2167-2172) changes core routing and connection handling, but we lack tests that verify behavior at scale.

What Needs Testing

1. Network Topology Formation

   • Peers should form a mesh topology, not a star (all connected to gateway only)
   • Each peer should have multiple connections (min_connections target)
   • Connection acquisition should complete in reasonable time

2. Subscription Tree Formation

   • When multiple peers subscribe to a contract, proper tree structure should form
   • Updates should propagate through the tree efficiently
   • Unsubscribe should prune branches correctly

3. Seeding Mechanism

   • Contract state should be retrievable from any peer in the network
   • Put operations should seed to appropriate locations
   • Get operations should find seeded state

4. NAT Traversal at Scale

   • Multiple NAT peers should successfully join
   • Peers behind NAT should be reachable for subscriptions
   • Connection handoff between peers should work

Current Best Test

The 6-peer River message flow test with Docker NAT simulation (docker-nat-sim work) is our most comprehensive test. We should expand on this pattern.

Proposed Tests

Phase 1: 50-Peer Network

1. Topology Health Test

   • Start 50-peer network, wait for stabilization
   • Verify each peer has >= min_connections
   • Verify no star topology (peers connected to more than just gateway)
   • Measure time to reach stable state

2. Subscription Propagation Test

   • Put contract at one peer
   • Subscribe from ~20 peers at different network locations
   • Update contract and verify all subscribers receive update
   • Measure propagation latency

3. Seeding Distribution Test

   • Put contracts targeting different ring locations
   • Verify state is retrievable from peers that didn't originate the put
   • Test get from peers far from contract location

Phase 2: 200-Peer Network

1. Scale Topology Test

   • Verify topology formation at 200 peers
   • Measure connection acquisition time distribution
   • Identify bottlenecks

2. Subscription Tree Depth Test

   • With 200 peers, subscription trees should have meaningful depth
   • Verify update propagation through deep trees
   • Test branch pruning behavior

3. Mixed NAT Network Test (100 public, 100 NAT peers)

   • Verify NAT peers can subscribe and receive updates
   • Verify operations route correctly through NAT peers

Implementation Notes

• Use freenet_test macro framework or standalone harness
• May need Docker/container orchestration for 200 peers
• Should run as separate CI job (not blocking normal CI)
• Consider metrics collection for performance analysis

Related

• Issue fix: Star topology formation due to serialized connection acquisition #2173 (star topology due to serial connection acquisition)
• PR refactor: Restructure peer identity and address handling #2164 series (connection-based routing refactor)
• Existing six-peer-regression test

[AI-assisted - Claude]

[github-actions]

🤖 Auto-labeled

Applied labels:

• T-feature (90% confidence)
• P-high (85% confidence)
• E-hard (80% confidence)
• A-networking (95% confidence)
• A-developer-xp (90% confidence)

Reasoning: The issue requests adding large-scale automated network tests (50 and 200 peers) covering topology, subscriptions, seeding, and NAT behavior. This is a request to add new test functionality to the project (T-feature). It impacts many developers and CI trustworthiness, so it should be high priority (P-high). Implementing and orchestrating hundreds of peers with Docker/CI and collecting metrics is non-trivial, so estimate as E-hard. The tests primarily exercise networking (topology, NAT, routing) and also involve developer tooling/CI/test harness (A-networking and A-developer-xp).

Previous labels: E-medium, P-high

If these labels are incorrect, please update them. This helps improve the auto-labeling system.