docs: v3.5.0 release notes + API documentation across README, deployment guide, FAQ

- Release notes for 95 commits since v3.4.1
- OpenAPI/Swagger docs: /api/spec and /api/docs called out everywhere
- Deployment guide: new API Documentation section
- README: API docs link added
- FAQ: 'Where is the API documentation?' entry
- Test plans for v3.4.2 validation
This commit is contained in:
you
2026-04-08 06:55:25 +00:00
parent 5606bc639e
commit cde62166cb
6 changed files with 764 additions and 0 deletions
+2
View File
@@ -254,6 +254,8 @@ Contributions welcome. Please read [AGENTS.md](AGENTS.md) for coding conventions
**Live instance:** [analyzer.00id.net](https://analyzer.00id.net) — all API endpoints are public, no auth required.
**API Documentation:** CoreScope auto-generates an OpenAPI 3.0 spec. Browse the interactive Swagger UI at [`/api/docs`](https://analyzer.00id.net/api/docs) or fetch the machine-readable spec at [`/api/spec`](https://analyzer.00id.net/api/spec).
## License
MIT
+32
View File
@@ -261,6 +261,38 @@ Caddy handles certificate issuance and renewal automatically.
---
## API Documentation
CoreScope auto-generates an OpenAPI 3.0 specification from its route definitions. The spec is always in sync with the running server — no manual maintenance required.
### Endpoints
| URL | Description |
|-----|-------------|
| `/api/spec` | OpenAPI 3.0 JSON schema — machine-readable API definition |
| `/api/docs` | Interactive Swagger UI — browse and test all 40+ endpoints |
### Usage
**Browse the API interactively:**
```
http://your-instance/api/docs
```
**Fetch the spec programmatically:**
```bash
curl http://your-instance/api/spec | jq .
```
**For bot/integration developers:** The spec includes all request parameters, response schemas, and example values. Import it into Postman, Insomnia, or any OpenAPI-compatible tool.
### Public instance
The live instance at [analyzer.00id.net](https://analyzer.00id.net) has all API endpoints publicly accessible:
- Spec: [analyzer.00id.net/api/spec](https://analyzer.00id.net/api/spec)
- Docs: [analyzer.00id.net/api/docs](https://analyzer.00id.net/api/docs)
---
## Monitoring & Health Checks
### Docker health check
+248
View File
@@ -0,0 +1,248 @@
# CoreScope v3.5.0 Release Notes
**95 commits** since v3.4.1 — the biggest release yet.
---
## 🐳 Pre-built Docker Images — One-Line Deploy
CoreScope now publishes pre-built Docker images to GitHub Container Registry. **No more building from source.**
### Quick start
```bash
docker run -d --name corescope \
-p 80:80 -p 1883:1883 \
-v corescope-data:/app/data \
ghcr.io/kpa-clawbot/corescope:latest
```
### Transitioning from source builds
If you're currently building your own image:
1. **Stop your existing container**`docker stop corescope && docker rm corescope`
2. **Keep your data volume** — your SQLite DB and config.json are preserved in the bind mount
3. **Pull the pre-built image**`docker pull ghcr.io/kpa-clawbot/corescope:latest`
4. **Run with your existing data** — mount the same volume:
```bash
docker run -d --name corescope \
-p 80:80 -p 1883:1883 \
-v /path/to/your/data:/app/data \
ghcr.io/kpa-clawbot/corescope:latest
```
### Image tags
- `ghcr.io/kpa-clawbot/corescope:latest` — latest stable release
- `ghcr.io/kpa-clawbot/corescope:vX.Y.Z` — pinned version (recommended for production)
- `ghcr.io/kpa-clawbot/corescope:edge` — tip of master (unstable, for testing)
### Zero-config defaults
No `config.json` required. The server starts with sensible defaults: HTTP on port 80 (via Caddy), internal MQTT broker on 1883, SQLite in `/app/data/`. Drop a `config.json` into your data volume to customize.
### New environment variables
- `DISABLE_CADDY=true` — skip the Caddy reverse proxy (if you bring your own)
- `DISABLE_MOSQUITTO=true` — skip the internal MQTT broker (if using an external one)
---
## ⚡ Performance
**35 performance optimizations** — CoreScope now serves HTTP within 2 minutes on any database size.
### Headlines
- **Packets page**: p50 83% faster, p95 97% faster
- **Topology analytics**: p50 83% faster
- **Async backfill**: resolved_path backfill runs in the background — server is responsive during startup even with millions of observations
- **Memory**: prefix map capped at 8 chars (10x memory reduction)
### Backend
- Incremental max ID tracking — no more full table scans on ingest
- Batched eviction index cleanup
- Combined filter passes (single scan instead of chained)
- Single sort for RF analytics arrays
- Early time filter in node analytics
- In-memory hop resolution (eliminates N+1 DB queries)
- Incremental distance index updates
- Region observer caching (30s TTL)
- Store stats: 5→2 concurrent queries
- Subpath detail index instead of full packet scan
- Grouped packets: cached observer count, deferred map construction
### Frontend
- Coalesced WebSocket renders with requestAnimationFrame
- Replaced N+1 API calls with bulk `expand=observations` query
- Parallelized `replayRecent()` fetches
- Map markers repositioned on zoom instead of full rebuild
- Network status computed client-side
- Combined 4 subpath API calls into single bulk endpoint
- Batched observation fetching eliminates N+1 on sort
- Timeline updates skipped when tab is hidden
- Incremental DOM diff in virtual scroll
- Lazy observation maps in `txToMap`
- og-image compressed from 1.1MB to 235KB
---
## 🔬 RF Health Dashboard
New **RF Health** tab on the Analytics page with comprehensive observer radio metrics:
- **Small multiples grid** — one card per observer showing key metrics at a glance
- **Noise floor column chart** — color-coded bars (green/yellow/red by threshold) replacing the old line chart. Hover tooltips show exact dBm + timestamp
- **Airtime charts** — TX/RX airtime with delta computation
- **Error rate tracking** — receive errors per observer over time
- **Battery charts** — observer battery level monitoring
- **Region filter** — filter RF metrics by geographic region
- **Observer detail** — click any observer to see detailed metrics in a side panel
---
## 🗺️ Live Map Enhancements
### TRACE packet real path visualization
Trace packets now show where the trace **actually reached** vs the full intended route:
- **Solid animated line** for completed hops
- **Dashed/ghosted line** (25% opacity) for unreached hops
- Helps operators see exactly where traces fail
### Movable panels
- Overlay panels (feed, legend, node detail) have corner-position toggle buttons
- Non-scrolling panel headers keep controls accessible
- Panels remember their position in localStorage
### Collapsible panels
- All live map panels can be collapsed to just their header bar
- Medium breakpoint responsive layout
---
## 📊 Analytics & Data
### Hash Stats accuracy
- "By Repeaters" section now correctly filters to repeater-role nodes only (was including companions/rooms)
- Hash size display correctly shows "Unknown" for zero-hop direct adverts instead of misleading "1 byte"
### Prefix Tool
- New tab on the Analytics page showing prefix collision data
- Text consistency improvements
### Paths through node accuracy
- Fixed prefix collision bug where packets through a different node with the same 2-char prefix were incorrectly shown
- Now uses `resolved_path` (neighbor affinity graph) for accurate disambiguation
### Table sorting
- Sortable column headers on nodes list, neighbor tables, observer tables, and packets table
- Sort indicators (▲/▼) show active sort column and direction
- Stable sort with timestamp tiebreaker
---
## 🔗 Deep Linking
- **Nodes page**: URL updates with node pubkey, tab, and search state
- **Packets page**: URL includes time window and region filter
- **Channels page**: URL reflects selected channel and node
- **Analytics page**: URL includes active tab
*Note: Packet filter expressions and desktop node selection deep links are coming in a follow-up.*
---
## 🎨 UI/UX
### Distance unit preference
- Choose between km, mi, or auto (locale-based) in the Customizer → Display tab
- Small imperial distances show in feet (not "0.0 mi")
- Analytics, node neighbors, and all distance displays respect the setting
### Channel color highlighting
- Assign colors to channels for visual distinction in live feed and packets table
- 8-color palette, left-border encoding
- Color dots on channels page and live feed items for easy access
- Persists in localStorage
### Mobile accessibility
- Touch targets minimum 44×44px
- ARIA labels on interactive elements
- Small viewport support improvements
### Map byte-size filter
- Filter map markers by packet size
### OpenAPI / Swagger
- Auto-generated OpenAPI 3.0 spec at **`/api/spec`** — returns the full API schema as JSON
- Interactive Swagger UI at **`/api/docs`** — browse and test all endpoints from your browser
- Spec is generated from route definitions, always in sync with the running server
- Useful for bot developers, integrators, and anyone building on top of CoreScope
- Example: `curl https://your-instance/api/spec | jq .info`
---
## 🔒 Security
### API key enforcement
- Weak/default API keys (e.g., "test", "admin", "password") are rejected at startup
- Write endpoints blocked when no API key is configured
- Startup warning when API key is missing
### License
- Changed to GPL v3
---
## 🐛 Bug Fixes
- Fixed `originLat` out-of-scope ReferenceError killing live map animations
- Fixed zero-hop direct adverts showing incorrect hash size
- Fixed live map animations stuck on any advert (block-scoping bug)
- Fixed hash stats including non-repeater nodes
- Fixed paths-through-node showing unrelated packets (prefix collision)
- Fixed channel color picker completely non-functional (data shape mismatch)
- Fixed panel corner toggle buttons invisible/scrolling away
- Fixed table sorting broken on all node tables (attribute mismatch)
- Fixed `time.Tick` leak in prune goroutine
- Fixed View Route on Map button on packet detail page
- Fixed virtual scroll row height measurement
- Fixed collision cache not respecting TTL config
- Fixed prefix map unbounded memory growth (capped at 8 chars)
- Fixed memory eviction using inaccurate memory estimates (now uses runtime heap stats)
---
## 🏗️ Infrastructure
- CI pipeline consolidated: build → test → publish to GHCR → deploy staging
- Staging compose: standard ports (80/1883), no memory limit
- CI build time reduced from ~14 min to ~3-4 min (removed arm64 QEMU cross-compilation)
---
## 📖 Documentation
- Deployment guide with Docker quick start, compose examples, and configuration reference
- User guide: getting started, page descriptions, config options, FAQ
- OpenAPI 3.0 spec (auto-generated from route definitions)
- Security analysis: AES-128-ECB block-level vulnerability assessment
- Lock ordering documentation for concurrent access patterns
---
## Upgrading
### From v3.4.1
1. Pull the new image: `docker pull ghcr.io/kpa-clawbot/corescope:v3.5.0`
2. Stop and remove old container: `docker stop corescope && docker rm corescope`
3. Start with the new image (keep your data volume):
```bash
docker run -d --name corescope \
-p 80:80 -p 1883:1883 \
-v /your/data:/app/data \
ghcr.io/kpa-clawbot/corescope:v3.5.0
```
**First startup note:** The server will backfill `resolved_path` data for existing observations in the background. The HTTP server is available within ~2 minutes. Full backfill may take longer on large databases but does not block serving.
### Breaking changes
- None. All API endpoints are backward compatible.
- `resolved_path` is added to API responses (new field, `omitempty` — won't break existing clients).
+162
View File
@@ -0,0 +1,162 @@
# v3.4.2 Manual Validation Checklist
**Tester:** _______________
**Staging:** http://20.109.157.39
**Prod:** https://analyzer.00id.net (READ ONLY — do not deploy until staging passes)
**Browser:** Chrome + Firefox + Safari (mobile for responsive items)
**Time estimate:** ~45 minutes
---
## 🔴 HIGH RISK — Test First
### 1. Zero-hop hash size display (#649, #653)
- [ ] Go to Packets page, find a DIRECT advert (route_type=2, 0 hops)
- [ ] Open packet detail — hash size should say "Unknown (zero-hop)" or be hidden, NOT "1 byte"
- [ ] Check "Path Length" field shows `hash_count=0 (direct advert)`
- [ ] Find a FLOOD advert with 0 hops — it SHOULD show hash size (this is different from DIRECT)
### 2. TRACE packet real path (#651, #656)
- [ ] Send a trace from your companion
- [ ] Watch Live map — the animated dot should only travel along completed hops (solid line)
- [ ] Unreached hops should show as dashed/ghosted line at reduced opacity
- [ ] If trace completes fully, entire path should be solid
- [ ] Ghost line should auto-clean after ~10 seconds
### 3. "Paths through this node" accuracy (#655, #658)
- [ ] Go to: http://20.109.157.39/#/nodes/c0dedad4208acb6cbe44b848943fc6d3c5d43cf38a21e48b43826a70862980e4
- [ ] Check "Packets through this node" — packets should actually have this node in their path
- [ ] Compare with a node that shares a 2-char prefix (e.g. C0ffee SF) — they should show DIFFERENT packets
- [ ] Spot-check 3-4 packets: click through, verify path contains the node
### 4. Hash Stats "By Repeaters" (#652, #654)
- [ ] Go to Analytics → Hash Stats
- [ ] "By Repeaters" section should only show repeater-role nodes
- [ ] Compare count in "Multi-Byte Hash Adopters" vs "By Repeaters" — adopters may include companions, repeaters section should not
- [ ] Check that companions/rooms/sensors are excluded from the repeater distribution
### 5. Noise floor column chart (#600, #659)
- [ ] Go to Analytics → RF Health
- [ ] Noise floor chart should show vertical color-coded bars, NOT a line
- [ ] Green bars (< -100 dBm), yellow (-100 to -85), red (≥ -85)
- [ ] Hover over a bar — tooltip should show exact dBm + timestamp
- [ ] Check with only 1 observer selected — chart should still render (division by zero edge case)
- [ ] Reboot markers (if any) should show as vertical dashed lines
### 6. Async backfill on startup
- [ ] SSH to staging: `ssh -i ~/.ssh/id_ed25519 runner@20.109.157.39`
- [ ] `docker restart corescope-staging-go`
- [ ] Within 30 seconds, hit `curl http://localhost:82/api/stats` — should return data (not hang)
- [ ] Check `backfilling` and `backfillProgress` fields in stats response
- [ ] Server should be serving HTTP while backfill runs in background
---
## 🟡 MEDIUM RISK — Features
### 7. Distance unit preference (#621, #646)
- [ ] Go to Customizer → Display tab
- [ ] Change distance unit to "mi" — all distances should show in miles
- [ ] Change to "km" — all distances should show in km
- [ ] Change to "auto" — should use locale (US = miles, EU = km)
- [ ] Check Analytics page distances update after customizer change (no page reload needed)
- [ ] Check Node detail → Neighbors table distances
- [ ] Very small distances (<0.1 mi) should show in feet, not "0.0 mi"
### 8. Panel corner toggle (#608, #657)
- [ ] Go to Live map page
- [ ] Each panel (feed, legend, node detail) should have a small corner-toggle button
- [ ] Click the button — panel should snap to next corner (TL → TR → BR → BL)
- [ ] Refresh page — panel positions should persist (localStorage)
- [ ] Move two panels to same corner — collision avoidance should skip to next free corner
- [ ] On mobile viewport (<768px?) — toggle buttons should be hidden
### 9. Deep linking (#536, #618)
- [ ] Navigate to Nodes page, click a node → URL should update with pubkey hash
- [ ] Copy URL, open in new tab → should land on same node
- [ ] Apply packet filters → URL hash should include filter params
- [ ] Channels page: select a node → URL should reflect selection
- [ ] Analytics tabs: switch tabs → URL should include tab name
- [ ] Share a deep link with someone — they should see the same view
### 10. Sortable tables (#620, #638, #639)
- [ ] Nodes list: click column headers — should sort ascending/descending
- [ ] Sort indicator (arrow) should be visible on active column
- [ ] Node detail → Neighbors table: sortable
- [ ] Node detail → Observers table: sortable
- [ ] Packets table: sortable by column headers
### 11. Channel color highlighting (#271, #607, #611)
- [ ] Go to Channels page
- [ ] Assign a color to a channel using the color picker
- [ ] Feed rows should highlight with that color
- [ ] Change color — should update immediately
- [ ] Refresh — color assignment should persist
### 12. Collapsible panels (#606)
- [ ] Live map: panels should have collapse/expand toggle
- [ ] Collapsed panel should show just the header
- [ ] State should persist across page navigations
### 13. Mobile accessibility (#630, #633)
- [ ] Open staging on phone (or Chrome DevTools mobile emulation)
- [ ] Touch targets should be at least 44×44px
- [ ] Channel color picker should work on mobile
- [ ] No horizontal scroll on any page
- [ ] ARIA labels present on interactive elements (inspect with accessibility tools)
### 14. Map byte-size filter (#565, #568)
- [ ] Go to Map page
- [ ] Find the byte-size filter control
- [ ] Filter by packet size — map should update to show only matching packets
- [ ] Clear filter — all packets should return
### 15. API key security (#532, #628)
- [ ] Try accessing a write endpoint without API key — should be blocked
- [ ] Try with a weak key (e.g., "test", "admin") — should be rejected at startup
- [ ] Check staging logs for API key warning: `docker logs corescope-staging-go 2>&1 | grep -i "apiKey\|api_key\|security"`
### 16. OpenAPI/Swagger (#530, #632)
- [ ] Hit http://20.109.157.39/api/spec — should return valid OpenAPI 3.0 spec
- [ ] Hit http://20.109.157.39/api/docs — should show Swagger UI
- [ ] Try an endpoint from Swagger UI — should work
---
## 🟢 LOW RISK — Verify Quickly
### 17. View Route on Map button
- [ ] Go to any packet detail page
- [ ] Click "View Route on Map" — should navigate to map with route highlighted
### 18. og-image compression
- [ ] Check page source or network tab — og-image.png should be < 300KB (was 1.1MB)
### 19. Prefix Tool
- [ ] Analytics → Prefix Tool tab should load
- [ ] Should show collision data
### 20. License
- [ ] Check repo footer/LICENSE — should be GPL v3
### 21. Docker DISABLE_CADDY
- [ ] (If testable) Set DISABLE_CADDY=true — Caddy should not start
### 22. Region filter on RF Health
- [ ] RF Health tab: change region filter — charts should update
---
## 🏁 Sign-off
| Section | Status | Notes |
|---------|--------|-------|
| High risk (1-6) | ☐ | |
| Medium risk (7-16) | ☐ | |
| Low risk (17-22) | ☐ | |
| **Overall** | ☐ | |
**Tested by:** _______________
**Date:** _______________
**Staging version:** `curl -s http://20.109.157.39/api/stats | jq .version`
**Ready for release:** ☐ Yes / ☐ No — blockers: _______________
+309
View File
@@ -0,0 +1,309 @@
# v3.4.2 Release Test Plan
**Scope:** 90 commits since v3.4.1 (84 files, +14,931 / -1,005)
**Categories:** 19 perf, 19 feat, 18 fix, 15 docs, 3 chore, 1 test, 1 refactor, 1 ci
**Date:** 2026-04-08
---
## A. Automated Tests — Verify All Pass
### Go Backend
```bash
cd cmd/server && go test -race -count=1 ./...
cd cmd/ingestor && go test -race -count=1 ./...
```
**Test files (27 total):**
| File | Tests For |
|------|-----------|
| `cmd/server/decoder_test.go` | Hash size zero-hop, TRACE hopsCompleted, transport direct |
| `cmd/server/backfill_async_test.go` | **NEW** — Async chunked backfill |
| `cmd/server/eviction_test.go` | Memory eviction with runtime heap stats |
| `cmd/server/apikey_security_test.go` | **NEW** — Weak/default API key rejection |
| `cmd/server/openapi_test.go` | **NEW** — OpenAPI spec generation |
| `cmd/server/routes_test.go` | Batch observations endpoint, subpaths-bulk, expand=observations |
| `cmd/server/cache_invalidation_test.go` | cacheTTL config wiring |
| `cmd/server/config_knobs_test.go` | cacheTTLSec helper |
| `cmd/server/helpers_test.go` | constantTimeEqual, IsWeakAPIKey |
| `cmd/server/obs_dedup_test.go` | UniqueObserverCount tracking |
| `cmd/server/neighbor_*.go` (4 files) | Neighbor graph, affinity, persistence |
| `cmd/server/perfstats_race_test.go` | Perf stats concurrency |
| `cmd/server/resolve_context_test.go` | Resolved path filtering |
| `cmd/server/advert_pubkey_test.go` | Advert pubkey tracking |
| `cmd/server/db_test.go` | SQLite operations |
| `cmd/server/config_test.go` | Config loading |
| `cmd/server/coverage_test.go` | Coverage helpers |
| `cmd/server/parity_test.go` | Go/JS decoder parity |
| `cmd/server/websocket_test.go` | WebSocket broadcast |
| `cmd/ingestor/decoder_test.go` | Ingestor decoder (hash size zero-hop) |
| `cmd/ingestor/db_test.go` | Ingestor DB writes |
| `cmd/ingestor/config_test.go` | Ingestor config |
| `cmd/ingestor/main_test.go` | Ingestor entry |
| `cmd/ingestor/coverage_boost_test.go` | Coverage helpers |
### Frontend Unit Tests
```bash
node test-packet-filter.js
node test-aging.js
node test-frontend-helpers.js
node test-table-sort.js # NEW — shared table sort utility
node test-channel-colors.js # NEW — channel color model
node test-panel-corner.js # NEW — panel corner toggle
node test-packets.js # NEW — packets page logic
node test-hop-resolver-affinity.js
node test-customizer-v2.js
node test-live.js
node test-live-dedup.js
```
### E2E / Playwright
```bash
BASE_URL=http://localhost:13581 node test-e2e-playwright.js
```
**Expected:** All existing tests pass + new tests added for sortable tables, deep linking, collapsible panels.
---
## B. Manual Browser Verification
### B1. HIGH RISK — Data Correctness
| # | Feature | Page | What to Check |
|---|---------|------|---------------|
| 1 | Hash size zero-hop | Packets detail | Find a direct (route_type=0) packet → hash_size should show 0, not a bogus computed value |
| 2 | TRACE hopsCompleted | Packets detail / Live map | Find a TRACE packet → verify `hopsCompleted` shows in decoded JSON, live map shows real path length vs intended |
| 3 | Transport direct hash size | Packets detail | Find route_type=RouteTransportDirect packet → hash_size=0 |
| 4 | resolved_path filtering | Node detail → Paths tab | Verify path-hop candidates use resolved_path, no prefix collision false positives |
| 5 | Hash stats repeater filter | Analytics → Hash Issues | "By Repeaters" should only show nodes with repeater role, not companions/sensors |
| 6 | Async chunked backfill | Server startup | Start server with large DB → verify HTTP serves within 2 minutes, `X-CoreScope-Status: backfilling` header present, then transitions to `ready` |
| 7 | Memory eviction (heap stats) | Admin/stats | Verify `/api/stats` shows realistic memory numbers from runtime heap, not the old estimation |
| 8 | Distance/subpath/path-hop indexes | Analytics → Distances, Subpaths | Verify analytics data matches v3.4.1 output (no missing or extra entries) |
| 9 | cacheTTL config wiring | Config | Set `cacheTTL.analyticsHashSizes: 300` in config → verify collision cache respects it |
### B2. MEDIUM RISK — User-Facing Features
| # | Feature | Page | What to Check |
|---|---------|------|---------------|
| 10 | Distance unit preference | Nodes detail, Map | Toggle km/mi/auto in settings → distances update throughout UI |
| 11 | Panel corner toggle | Live page | Click corner toggle → panel moves to opposite corner, persists on reload |
| 12 | Noise floor column chart | Analytics → RF | Verify column chart renders with color-coded thresholds, hover shows values |
| 13 | Deep linking UI states | All pages | Navigate to `#/nodes?tab=neighbors`, `#/packets?observer=X`, `#/channels?node=Y` → correct state loads. Copy URL, open in new tab → same state |
| 14 | Sortable tables | Nodes list, Neighbors, Observers | Click column headers → sort asc/desc, indicator arrow shows, persists correctly |
| 15 | Channel color highlighting | Channels, Live feed | Assign color to channel → feed rows show that color, persists on reload |
| 16 | Mobile accessibility | All pages (phone viewport) | Touch targets ≥44px, ARIA labels present, small viewport doesn't overflow |
| 17 | Collapsible panels | Live map | Collapse/expand panels, medium breakpoint auto-collapses, state persists |
| 18 | Byte-size map filter | Map page | Filter by byte size → markers update correctly |
| 19 | OpenAPI/Swagger | `/api/spec`, `/api/docs` | Spec loads valid JSON, Swagger UI renders and all endpoints are documented |
| 20 | API key rejection | Protected endpoints | Send weak key (e.g. "changeme", "test123") → 403 forbidden |
| 21 | Channel color picker mobile | Channels (phone viewport) | Color picker usable on touch, doesn't overflow |
| 22 | RF Health dashboard | Analytics → RF Health | Observer metrics grid, airtime charts, battery charts, error rate, region filter |
| 23 | Prefix Tool tab | Analytics → Prefix Tool | Renders correctly, collision data consistent with Hash Issues |
| 24 | View Route on Map | Packet detail page | Button works and shows route on map |
### B3. LOWER RISK — Performance (Verify No Regressions)
| # | Feature | Page | What to Check |
|---|---------|------|---------------|
| 25 | Incremental DOM diff | Packets (30K+) | Virtual scroll renders smoothly, no visible flicker |
| 26 | Coalesced WS renders | Live page | Rapid packets don't cause frame drops (rAF coalescing) |
| 27 | Marker reposition on zoom | Map | Zoom/resize → markers move smoothly, no full rebuild flash |
| 28 | Parallel replay fetches | Live → VCR | Replay loads quickly (parallel observation fetches) |
| 29 | Batch observations API | Packets page (sort change) | Changing sort fetches observations in batch (network tab: 1 POST not N GETs) |
| 30 | Client-side network status | Analytics | No separate API call for network status |
| 31 | og-image compression | `/og-image.png` | Verify loads, ~235KB not ~1.1MB |
---
## C. API Regression Tests
Run against a local server with test-fixture DB:
```bash
BASE=http://localhost:13581
# Core endpoints — verify response shape
curl -s "$BASE/api/stats" | jq '.totalPackets, .backfilling, .backfillProgress'
curl -s "$BASE/api/packets?limit=5" | jq '.packets[0] | keys'
curl -s "$BASE/api/packets?limit=5&expand=observations" | jq '.packets[0].observations | length'
curl -s "$BASE/api/nodes?limit=5" | jq '.[0] | keys'
# New endpoints
curl -s -X POST "$BASE/api/packets/observations" \
-H 'Content-Type: application/json' \
-d '{"hashes":["test123"]}' | jq '.results | keys'
curl -s "$BASE/api/analytics/subpaths-bulk?hops=A,B&hops=B,C" | jq 'keys'
curl -s "$BASE/api/observers/metrics/summary" | jq 'type'
curl -s "$BASE/api/spec" | jq '.openapi'
curl -s "$BASE/api/docs" | head -5 # Should return HTML
# Backfill status header
curl -sI "$BASE/api/stats" | grep X-CoreScope-Status
# API key rejection
curl -s -H 'X-API-Key: changeme' "$BASE/api/debug/vars" | jq '.error'
curl -s -H 'X-API-Key: test' "$BASE/api/debug/vars" | jq '.error'
# Existing endpoints — verify not broken
curl -s "$BASE/api/analytics/rf?timeRange=24h" | jq 'keys'
curl -s "$BASE/api/analytics/hash-sizes" | jq 'type'
curl -s "$BASE/api/analytics/distances" | jq 'type'
curl -s "$BASE/api/analytics/subpaths" | jq 'type'
curl -s "$BASE/api/channels" | jq 'type'
curl -s "$BASE/api/config/client" | jq 'keys'
```
### Expected response shape changes from v3.4.1:
- `/api/stats` now includes `backfilling` (bool) and `backfillProgress` (float 0-1)
- `/api/packets` no longer strips observations by default (lazy via `ExpandObservations` flag) — verify `observations` key absent without `expand=observations`
- Decoded packets with route_type=direct now have `hashSize: 0`
- TRACE packets now have `path.hopsCompleted` field
---
## D. Performance Regression Tests
### D1. Server Startup Time
```bash
# Start server with production-size DB (~30K packets)
# Measure time from process start to first successful HTTP response
time curl -s http://localhost:13581/api/stats > /dev/null
# Target: < 2 minutes (async backfill requirement)
```
### D2. Go Benchmarks
```bash
cd cmd/server && go test -bench=. -benchmem -count=3
```
Key benchmarks to compare with v3.4.1 baseline:
- `BenchmarkQueryPackets` — should not regress with new indexes
- `BenchmarkEvictStale` — batch removal from secondary indexes
- `BenchmarkGetStoreStats` — 2 concurrent queries vs 5 sequential
- `BenchmarkIngestNew` — additional index maintenance overhead
### D3. Frontend Performance
- Open Packets page with 30K+ packets → measure initial render time (DevTools Performance tab)
- Scroll rapidly through virtual scroll → should maintain 60fps
- Switch sort column on packets → single batch POST, not N+1 GETs
- Open Analytics page → no redundant API calls in network tab
### D4. Memory Usage
- After loading 30K packets, check `/api/stats` memory figure
- Compare with v3.4.1 baseline (prefix map cap at 8 chars should reduce ~10x)
- Verify eviction triggers at correct memory threshold using runtime heap stats
---
## E. Infrastructure / Deployment Tests
### E1. Docker Build
```bash
docker build -t corescope:test .
docker run --rm -p 13581:13581 corescope:test
# Verify: container starts, HTTP responds, WebSocket connects
```
### E2. GHCR Publish (CI)
- Verify CI publishes to `ghcr.io/kpa-clawbot/corescope`
- Verify tags: `edge` (master), `vX.Y.Z` (release)
### E3. Staging Deploy
```bash
# Verify staging compose works with standard ports
docker compose -f docker-compose.staging.yml up -d
# Check: no 3GB memory limit, standard port binding
```
### E4. DISABLE_CADDY
```bash
docker run --rm -e DISABLE_CADDY=true corescope:test
# Verify: Caddy not started, Go server serves directly
```
### E5. CI Pipeline
- Verify consolidated pipeline: build → publish GHCR → deploy staging
- Verify runs on `meshcore-runner-2`
---
## F. Edge Cases & Integration Tests
### F1. Cross-Feature Interactions
| Scenario | Risk |
|----------|------|
| Deep link to sorted table → sort state matches URL params | Medium |
| Channel color + deep link → color persists in linked URL | Medium |
| Panel corner toggle + collapsible panels → both states persist independently | Low |
| Distance unit pref + neighbor table sort by distance → sort uses correct unit | Medium |
| Noise floor chart + region filter → chart respects filter | Medium |
| Byte-size map filter + channel color highlighting → both active simultaneously | Low |
### F2. Data Correctness Edge Cases
| Scenario | Risk |
|----------|------|
| Zero-hop TRACE packet (should NOT reset hashSize — TRACE exemption) | **High** |
| Packet with all hops having same 2-char prefix → resolved_path filtering prevents false match | **High** |
| Node that switches role (repeater → companion) → hash stats updates | Medium |
| Backfill interrupted mid-chunk (server restart) → resumes or completes on next start | Medium |
| Empty DB startup → no errors, backfill completes instantly | Low |
| DB with 100K+ packets → async backfill doesn't OOM, progress reported | **High** |
### F3. Concurrency / Race Conditions
| Scenario | Risk |
|----------|------|
| Concurrent API requests during backfill → no deadlock (lock ordering documented) | **High** |
| Eviction running while analytics query in progress → no stale pointer panic | **High** |
| Multiple WebSocket clients during high ingest rate → coalesced broadcasts don't drop | Medium |
| `time.NewTicker` cleanup on graceful shutdown (replaced `time.Tick`) | Low |
### F4. API Key Security
| Scenario | Expected |
|----------|----------|
| No API key configured → write endpoints disabled | 403 "write endpoints disabled" |
| Weak key "changeme" → rejected even if configured | 403 "forbidden" |
| Timing-safe comparison → no timing oracle | Constant-time via `crypto/subtle` |
| Empty string key → rejected | 401 "unauthorized" |
### F5. Browser Compatibility
- Test on Chrome, Firefox, Safari (latest)
- Test on iOS Safari, Android Chrome
- Verify touch targets on mobile (44px minimum)
- Verify ARIA labels with screen reader
---
## G. Test Coverage Gaps — Action Items
| Gap | Priority | Action |
|-----|----------|--------|
| No automated test for distance unit preference rendering | Medium | Add Playwright test |
| No automated test for noise floor column chart | Medium | Add Playwright test |
| No automated test for deep link state restoration | **High** | Add Playwright tests for each deep-linkable state |
| No automated test for channel color persistence | Medium | `test-channel-colors.js` covers model; need Playwright for UI |
| No automated test for mobile viewport behavior | Medium | Add Playwright test with mobile viewport |
| No automated test for backfill progress header | Low | Add to `routes_test.go` |
| No automated test for `time.NewTicker` cleanup | Low | Add to graceful shutdown test |
| Observer metrics endpoints not covered in route tests | Medium | Add to `routes_test.go` |
| Subpaths-bulk endpoint needs test | Medium | Add to `routes_test.go` |
| No load test for batch observations endpoint (200 hash limit) | Low | Add boundary test |
---
## H. Release Checklist
- [ ] All Go tests pass with `-race` flag
- [ ] All frontend unit tests pass
- [ ] Playwright E2E tests pass
- [ ] Manual browser verification (Section B) complete
- [ ] API regression tests (Section C) pass
- [ ] Docker build succeeds
- [ ] Staging deploy verified
- [ ] No console errors on any page
- [ ] Performance spot-checks (Section D) — no regressions
- [ ] Coverage badges updated (backend ≥85%, frontend ≥42%)
- [ ] CHANGELOG updated
- [ ] Tag `v3.4.2` created
+11
View File
@@ -52,3 +52,14 @@ CoreScope uses URL hashes for deep linking. Copy the URL from your browser — i
- `#/packets/abc123` — a specific packet
- `#/analytics?tab=collisions` — the hash issues tab
- `#/nodes/pubkey123` — a specific node's detail page
### Where is the API documentation?
CoreScope auto-generates an OpenAPI 3.0 specification from its route definitions:
- **Interactive docs (Swagger UI):** `/api/docs` — browse and test all 40+ endpoints from your browser
- **Machine-readable spec:** `/api/spec` — import into Postman, Insomnia, or any OpenAPI tool
The spec is always in sync with the running server. No manual maintenance needed.
On the public instance: [analyzer.00id.net/api/docs](https://analyzer.00id.net/api/docs)