Commit Graph

7 Commits

Author SHA1 Message Date
Markus Flür
e7f532db93 Implementing an RPC monitoring solution with a web-ui as implementation example. (#244)
* bd sync: 2025-10-30 12:12:27

* Working on frontend

* bd sync: 2025-11-06 16:55:55

* feat: finish bd monitor human viewer

* Merge conflicts resolved and added tests

* bd sync: 2025-11-06 17:23:41

* bd sync: 2025-11-06 17:34:52

* feat: Add reload button and multiselect status filter to monitor

- Changed status filter from single select to multiselect with 'Open' selected by default
- Added reload button with visual feedback (hover/active states)
- Updated filterIssues() to handle multiple selected statuses
- Added reloadData() function that reloads both stats and issues
- Improved responsive design for mobile devices
- Filter controls now use flexbox layout with better spacing

* fix: Update monitor statistics to show Total, In Progress, Open, Closed

- Replaced 'Ready to Work' stat with 'In Progress' stat
- Reordered stats to show logical progression: Total -> In Progress -> Open -> Closed
- Updated loadStats() to fetch in-progress count from stats API
- Removed unnecessary separate API call for ready count

* fix: Correct API field names in monitor stats JavaScript

The JavaScript was using incorrect field names (stats.total, stats.by_status)
that don't match the actual types.Statistics struct which uses flat fields
with underscores (total_issues, in_progress_issues, etc).

Fixed by updating loadStats() to use correct field names:
- stats.total -> stats.total_issues
- stats.by_status?.['in-progress'] -> stats.in_progress_issues
- stats.by_status?.open -> stats.open_issues
- stats.by_status?.closed -> stats.closed_issues

Fixes beads-9

* bd sync: 2025-11-06 17:51:24

* bd sync: 2025-11-06 17:56:09

* fix: Make monitor require daemon to prevent SQLite locking

Implemented Option 1 from beads-eel: monitor now requires daemon and never
opens direct SQLite connection.

Changes:
- Added 'monitor' to noDbCommands list in main.go to skip normal DB initialization
- Added validateDaemonForMonitor() PreRun function that:
  - Finds database path using beads.FindDatabasePath()
  - Validates daemon is running and healthy
  - Fails gracefully with clear error message if no daemon
  - Only uses RPC connection, never opens SQLite directly

Benefits:
- Eliminates SQLite locking conflicts between monitor and daemon
- Users can now close/update issues via CLI while monitor runs
- Clear error messages guide users to start daemon first

Fixes beads-eel

* bd sync: 2025-11-06 18:03:50

* docs: Add bd daemons restart subcommand documentation

Added documentation for the 'bd daemons restart' subcommand across all documentation files:

- commands/daemons.md: Added full restart subcommand section with synopsis, description, arguments, flags, and examples
- README.md: Added restart examples to daemon management section
- AGENTS.md: Added restart examples with --json flag for agents

The restart command gracefully stops and starts a specific daemon by workspace path or PID,
useful after upgrading bd or when a daemon needs refreshing.

Fixes beads-11

* bd sync: 2025-11-06 18:13:16

* Separated the web ui from the general monitoring functionality

---------

Co-authored-by: Steve Yegge <stevey@sourcegraph.com>
2025-11-07 09:49:12 -08:00
Steve Yegge
e2bb4311f1 Improve mutation channel robustness
- Add event type constants (MutationCreate, MutationUpdate, MutationDelete, MutationComment)
- Make buffer size configurable via BEADS_MUTATION_BUFFER (default 512, up from 100)
- Add defensive closed-channel handling in event loop consumer
- Add 1s dropped-events ticker (down from 60s) for faster recovery
- Verify mutationChan is never closed (prevents panic)

Oracle review findings addressed:
- Eliminates panic risk from send-on-closed-channel
- Reduces worst-case recovery latency from 60s to 1s
- Increases buffer capacity for better burst handling
- Type-safe event constants prevent string typos

Related: bd-36320a04, bd-1f4086c5
2025-10-31 19:12:02 -07:00
Steve Yegge
8a7c36fe70 Validation and testing improvements (bd-74, bd-77)
- Remove unreachable function DroppedEventsCount from RPC server
- Add TestMutationToExportLatency for event-driven daemon validation
- Test currently skipped pending full bd-85 implementation
- Create test coverage improvement issues (bd-114 through bd-118)
- All validation checks pass: tests, build, linting baseline

Completed: bd-74, bd-77
Amp-Thread-ID: https://ampcode.com/threads/T-24404401-6c5b-466d-9045-0da3a70cff9a
Co-authored-by: Amp <amp@ampcode.com>
2025-10-29 15:30:47 -07:00
Steve Yegge
fea86f9b31 Implement event-driven daemon improvements for bd-85
- Add mutation events for label/dep/comment operations
- Create separate export-only and import-only functions
- Add dropped events counter with safety net export
- Complete bd-80 mutation channel implementation

Event-driven mode now:
- Emits mutation events for ALL write operations (not just create/update/close)
- Uses createExportFunc() for mutations (export+commit/push only, no pull)
- Uses createAutoImportFunc() for file changes (pull+import only, no export)
- Tracks dropped events and triggers safety export every 60s if any dropped
- Achieves <500ms latency target by avoiding full sync on each trigger

Behind BEADS_DAEMON_MODE=events flag (poll is still default)
2025-10-29 11:22:29 -07:00
Steve Yegge
781e300d33 Add event-driven daemon architecture (Phase 1 foundation)
- Add fsnotify dependency for file watching
- Create daemon_debouncer.go: batch rapid events (500ms window)
- Create daemon_watcher.go: monitor JSONL and git refs changes
- Create daemon_event_loop.go: event-driven sync loop
- Add mutation channel to RPC server (create/update/close events)
- Add BEADS_DAEMON_MODE env var (poll/events, default: poll)

Phase 1 implementation: opt-in via BEADS_DAEMON_MODE=events
Target: <500ms latency (vs 5000ms), ~60% CPU reduction

Related: bd-49 (epic), bd-50, bd-51, bd-53, bd-54, bd-55, bd-56
Amp-Thread-ID: https://ampcode.com/threads/T-35a3d0d7-4e19-421d-8392-63755035036e
Co-authored-by: Amp <amp@ampcode.com>
2025-10-28 13:12:37 -07:00
Steve Yegge
9edcb6f19f Remove cache fields from Server struct (bd-31)
- Removed storageCache, cacheMu, maxCacheSize, cacheTTL, cleanupTicker fields
- Removed cacheHits and cacheMisses metrics
- Removed cache size/TTL parsing from env vars in NewServer()
- Removed cleanup ticker goroutine from Start()
- Removed cache cleanup logic from Stop()
- Part of bd-29 epic to remove daemon storage cache

Amp-Thread-ID: https://ampcode.com/threads/T-239a5531-68a5-4c98-b85d-0e3512b2553c
Co-authored-by: Amp <amp@ampcode.com>
2025-10-27 23:20:10 -07:00
Steve Yegge
0c737025b5 Split internal/rpc/server.go into 8 focused modules (bd-215)
Refactored monolithic 2238-line server.go into 8 files with clear responsibilities:
- server_core.go: Server type, NewServer (115 lines)
- server_lifecycle_conn.go: Start/Stop/connection handling (248 lines)
- server_cache_storage.go: Storage caching and eviction (286 lines)
- server_routing_validation_diagnostics.go: Request routing/validation (384 lines)
- server_issues_epics.go: Issue CRUD operations (506 lines)
- server_labels_deps_comments.go: Labels/deps/comments (199 lines)
- server_compact.go: Compaction operations (287 lines)
- server_export_import_auto.go: Export/import operations (293 lines)

Improvements:
- Replaced RWMutex.TryLock with atomic.Bool for portable single-flight guard
- Added default storage close in Stop() to prevent FD leaks
- All methods remain on *Server receiver (no behavior changes)
- Each file <510 LOC for better maintainability
- All tests pass, daemon verified working

Amp-Thread-ID: https://ampcode.com/threads/T-92d481ad-1bda-4ecd-bcf5-874a1889db30
Co-authored-by: Amp <amp@ampcode.com>
2025-10-27 21:14:34 -07:00