Indexer improvements: Multi-network, Monitoring, Graceful Shutdown & Replay CLI

[kris-nana]

Summary

This PR implements comprehensive improvements to the AudioBlock indexer subsystem, addressing four related issues that enhance reliability, observability, and debuggability.

Closes #255, Closes #256, Closes #257, Closes #258

What's Changed

1. Multi-Network Support (#255)

• ✅ Both networks index independently with separate cursors
• ✅ Failures on one network don't affect the other
• ✅ Documented in ADR-010

Implementation:

• Each (contract, network) pair runs in its own async poll loop
• Independent error handling and cursor management
• Configurable via environment variables (10 contracts: 5×mainnet + 5×testnet)

2. Replay/Reindex CLI (#256)

• ✅ Does not mutate live cursor
• ✅ Idempotent event upserts
• ✅ Documented with example invocations

Usage:

npm run cli:reindex -- \
  --contract CXXX... \
  --network mainnet \
  --from 1000000 \
  --to 1001000 \
  [--dry-run]

3. RPC Monitoring (#257)

• ✅ RPC call count metrics exposed
• ✅ Expected monthly volume estimated in ADR (5.2M calls/month)
• ✅ Alert thresholds proposed

Metrics:

• soroban_rpc_calls_total{network, method, status}
• soroban_rpc_latency_seconds{network, method}

4. Graceful Shutdown (#258)

• ✅ Cursor only advanced after batch fully persisted
• ✅ Verified with test suite
• ✅ No orphaned partial writes

Behavior:

• Listens for SIGTERM/SIGINT
• Finishes in-flight batch
• Updates cursor
• Exits cleanly

Files Changed

New Files

• src/workers/IndexerWorker.ts - Core indexer worker (300+ lines)
• src/cli/reindex.ts - Replay/reindex CLI tool (250+ lines)
• docs/adrs/010-indexer-architecture.md - Architecture decision record
• src/__tests__/IndexerWorker.test.ts - Worker tests
• src/__tests__/ReindexCLI.test.ts - CLI tests

Modified Files

• src/services/MetricsService.ts - Added RPC metrics
• src/services/Soroban/SorobanService.ts - Instrumented for metrics
• .env.example - Added indexer configuration
• package.json - Added worker:indexer and cli:reindex scripts

Configuration

New environment variables:

# Per-network RPC URLs
SOROBAN_RPC_URL_MAINNET=https://soroban-mainnet.stellar.org
SOROBAN_RPC_URL_TESTNET=https://soroban-testnet.stellar.org

# Mainnet contracts
ARTIST_FACET_MAINNET_CONTRACT_ID=
SONG_FACET_MAINNET_CONTRACT_ID=
ALBUM_FACET_MAINNET_CONTRACT_ID=
MARKETPLACE_FACET_MAINNET_CONTRACT_ID=
ROYALTY_FACET_MAINNET_CONTRACT_ID=

# Testnet contracts (same pattern)
ARTIST_FACET_TESTNET_CONTRACT_ID=
# ... etc

# Indexer tuning
INDEXER_POLL_INTERVAL_MS=5000  # Default: 5 seconds
INDEXER_BATCH_SIZE=100         # Default: 100 events
INDEXER_OVERLAP_WINDOW=10      # Reorg protection

Running the Indexer

# Start the indexer worker
npm run worker:indexer

# Reindex a specific range (debugging)
npm run cli:reindex -- \
  --contract <CONTRACT_ID> \
  --network mainnet \
  --from 1000000 \
  --to 1001000

# Dry-run mode (read-only)
npm run cli:reindex -- \
  --contract <CONTRACT_ID> \
  --network testnet \
  --from 500000 \
  --to 501000 \
  --dry-run

Testing

# Run indexer tests
npm test -- IndexerWorker.test.ts

# Run CLI tests
npm test -- ReindexCLI.test.ts

Architecture

IndexerWorker
├── ArtistFacet (mainnet)  → Independent poll loop
├── SongFacet (mainnet)    → Independent poll loop
├── AlbumFacet (mainnet)   → Independent poll loop
├── MarketplaceFacet (mainnet) → Independent poll loop
├── RoyaltyFacet (mainnet) → Independent poll loop
├── ArtistFacet (testnet)  → Independent poll loop
├── SongFacet (testnet)    → Independent poll loop
├── AlbumFacet (testnet)   → Independent poll loop
├── MarketplaceFacet (testnet) → Independent poll loop
└── RoyaltyFacet (testnet) → Independent poll loop
    ↓
SorobanRpcClient (instrumented)
    ↓
Soroban RPC
    ↓
Database (indexed_events, indexer_cursors)

Deployment Checklist

• Add contract IDs to production environment variables
• Deploy indexer worker alongside API server
• Set up Grafana dashboard for new RPC metrics
• Configure alerts for lag and error rate
• Test graceful shutdown in staging
• Monitor RPC call volume for first 24h

Notes

• No Claude/AI co-author in commits ✅
• All commits made with kris-nana account ✅
• TypeScript diagnostics: Clean ✅
• Follows existing codebase patterns ✅

Breaking Changes

None. This is additive functionality.

Next Steps

After merge:

1. Populate actual contract IDs in production
2. Deploy indexer worker
3. Monitor metrics and logs
4. Document any RPC provider quotas encountered

[drips-wave]

@kris-nana Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits

[kris-nana]

Closing this PR. The implementation will be added to the existing plan PR #583 instead.