chore(hermes): p2p testing is pre publish

[cong-or]

---

Fix: Restore is_pre_publish_completed for P2P Testing

Summary

Restore the original is_pre_publish_completed behavior to unblock P2P testing (just quickstart). The regression introduced in ff8b6d2d caused publishing to wait indefinitely for DHT providers that never appear in gossipsub-only test environments.

---

Problem

The P2P testing infrastructure (just quickstart) broke after commit ff8b6d2d:

• ❌ HTTP POST requests timing out (exit code 52)
• ❌ Publishing hangs at “Step 4/5: DHT not ready”
• ❌ Error: Other DHT providers not found, aborting
• ❌ Zero message propagation (0/5 nodes receiving)

---

Root Cause

Commit ff8b6d2d changed is_pre_publish_completed to require at least one other peer (not ourselves) to be registered as a DHT provider before completing a publish.

This assumption is invalid in P2P testing environments due to the difference between gossipsub and DHT provider announcements.

Content Distribution Flow

1. ✅ Publishing node adds content to IPFS and announces itself as a DHT provider

2. ✅ Publishing node broadcasts the message via gossipsub (PubSub)

3. ✅ Subscriber nodes receive the message via gossipsub

4. ❌ Subscriber nodes do not announce themselves as DHT providers

   • Nodes only become DHT providers when they explicitly fetch content from IPFS
   • Receiving content via gossipsub does not trigger provider announcements

Result:
The publishing node waits indefinitely for another DHT provider that never appears, blocking the HTTP handler for ~50 seconds before timing out.

---

Solution

1. Restore is_pre_publish_completed Logic

File: hermes/bin/src/runtime_extensions/hermes/doc_sync/host.rs

Reverted to the original behavior:

• ✅ Returns true if we are found as a DHT provider
• ✅ Returns true if other peers are found as providers
• ❌ Returns false only when no providers exist

Finding ourselves as a provider confirms that:

• DHT queries are functioning
• Our provider announcement succeeded
• The content is available on the network (at minimum from ourselves)

This matches the function’s documented contract and supports isolated P2P test networks.

---

3. Update Test Expectations

Updated the test case to reflect intended behavior:

#[test_case("OUR", &["OUR"] => true; "our peer is sufficient for P2P testing")]

Previously, this case incorrectly expected false.

---

4. Increase Bootstrap Initialization Wait

File: p2p-testing/justfile

Increased init-bootstrap peer ID discovery wait from 15s → 30s.

The original timeout was insufficient for nodes to fully initialize and log peer IDs, causing intermittent bootstrap failures on first run.

---

Testing Results

Before Fix

• ❌ Exit code 52 (timeout)
• ❌ 0/5 nodes receiving messages (0% propagation)
• ❌ 1/30 peer connections

After Fix

• ✅ Both health checks pass
• ✅ 100% message propagation (5/5 subscribers)
• ✅ Full mesh connectivity: 30/30 peer connections
• ✅ Low latency: 72–347ms propagation
• ✅ Reliable bootstrap initialization

---

[bkioshn]

This passes when run locally 👍🏼

[github-actions]

📚 Docs Preview

The docs for this PR can be previewed at the following URL:

https://docs.dev.projectcatalyst.io/hermes/fix-p2p-testing-is-pre-publish