Error Handling:
- Fix silent Write() error in WebSocket (events.go)
- Add error context to transport handshake messages
- Check os.MkdirAll error in zip extraction (miner.go)
- Explicitly ignore io.Copy errors on drain with comments
- Add retry logic (2 attempts) for transient stats collection failures
Resource Lifecycle:
- Add shutdown mechanism to DigestAuth goroutine
- Call Service.Stop() on context cancellation
- Add NodeService transport cleanup to Service.Stop()
- Fix WriteStdin goroutine leak on timeout with non-blocking send
API Design:
- Add profile validation (name, miner type required)
- Return 404 instead of 500 for missing profile PUT
- Make DELETE profile idempotent (return success if not found)
- Standardize error responses in node_service.go handlers
Observability:
- Add logging for P2P GetAllStats failures
- Add request ID correlation helper for handler logs
- Add logging for miner process exits (xmrig_start.go)
- Rate limit debug logs in transport hot path (1 in 100)
- Add metrics infrastructure with /metrics endpoint
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Critical fixes:
- Release mutex before HTTP calls in GetStats() to prevent blocking
- Fix m.cmd race between Stop() and Wait() goroutine by capturing locally
- Add context support to GetStats() for proper request cancellation
High priority fixes:
- Add existence check in collectMinerStats() before operating on miners
- Add mutex-protected httpClient getter/setter for thread-safe test mocking
Changes:
- Miner interface now requires context.Context for GetStats()
- Stats HTTP requests timeout after 5 seconds (was 30s client default)
- All callers updated to pass context (service uses request context)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
