codex-genesis-harness 0.1.1 → 0.1.5

package/.codex/skills/genesis-performance-profiling/observability/performance-tracking.md

@@ -0,0 +1,202 @@
+# Performance Tracking Log
+
+This file tracks performance measurements over time. Append a new entry after every profiling cycle. Never delete old entries — they form the historical record. Archive entries older than 6 months to `observability/archive/PERF_LOG_[year].md`.
+
+---
+
+## PERF_LOG.md Format
+
+Each entry must follow this exact format:
+
+```
+---
+date:        2026-05-31
+version:     1.4.2
+commit:      abc1234def5678
+environment: staging-isolated
+test_type:   baseline | regression | optimization | incident
+triggered_by: PR #123 | scheduled | incident INC-456 | manual
+---
+
+### Summary
+
+[One paragraph describing what was measured, why, and the outcome]
+
+### Metrics (averaged across N runs)
+
+| Endpoint | p50 ms | p95 ms | p99 ms | RPS | Error % | vs previous p95 | Status |
+|----------|--------|--------|--------|-----|---------|-----------------|--------|
+| GET /api/users | 42 | 165 | 380 | 312 | 0.02 | -8% | ✅ |
+| POST /api/orders | 89 | 240 | 510 | 156 | 0.05 | +2% | ✅ |
+
+### Memory profile
+- Heap at start: X MB
+- Heap at end: X MB  
+- Delta: +X MB over Y minutes (rate: Z MB/min)
+- Leak assessment: NONE / SUSPECT / CONFIRMED
+
+### Regressions detected
+- [none] OR [list with PERF-ID reference]
+
+### Optimizations implemented
+- [none] OR [description with before/after]
+
+### Next actions
+- [list of follow-up items]
+```
+
+---
+
+## Example Entry: Initial Baseline
+
+```
+---
+date:        2026-05-31
+version:     1.4.2
+commit:      abc1234def5678
+environment: staging-isolated
+test_type:   baseline
+triggered_by: release-1.4.2 preparation
+---
+
+### Summary
+
+Initial baseline captured for release 1.4.2 ahead of production deployment. Five-run protocol with 
+warm-up discarded. Environment was fully isolated with background CPU < 2%. All endpoints pass their 
+p95 SLA targets. Memory growth was minimal (< 5 MB over 10 minutes). No regressions or anomalies 
+detected. System is ready for production release.
+
+### Metrics (averaged across 4 measurement runs)
+
+| Endpoint | p50 ms | p95 ms | p99 ms | RPS | Error % | SLA p95 | Status |
+|----------|--------|--------|--------|-----|---------|---------|--------|
+| GET /api/users | 42 | 165 | 380 | 312 | 0.02 | 200 ms | ✅ |
+| GET /api/users/:id | 18 | 72 | 145 | 489 | 0.01 | 200 ms | ✅ |
+| POST /api/orders | 89 | 195 | 420 | 156 | 0.05 | 300 ms | ✅ |
+| GET /api/products | 25 | 88 | 180 | 401 | 0.00 | 150 ms | ✅ |
+| POST /auth/login | 145 | 320 | 580 | 89 | 0.10 | 500 ms | ✅ |
+
+### Memory profile
+- Heap at start: 256 MB
+- Heap at end: 261 MB
+- Delta: +5 MB over 10 minutes (rate: 0.5 MB/min)
+- Leak assessment: NONE (well within acceptable < 5 MB/min threshold)
+
+### Regressions detected
+- none (first baseline — no previous baseline to compare)
+
+### Optimizations implemented
+- none
+
+### Next actions
+- Set this baseline as the production reference. Store at: observability/baselines/v1.4.2-production.json
+- Schedule next baseline review for v1.5.0 release
+```
+
+---
+
+## Performance Trend Tracking
+
+Use this table to track p95 across releases at a glance:
+
+| Date | Version | GET /api/users p95 | POST /api/orders p95 | Auth p95 | Notes |
+|------|---------|--------------------|---------------------|---------|-------|
+| 2026-01-15 | 1.0.0 | 450 ms | 380 ms | 520 ms | Initial release |
+| 2026-02-20 | 1.1.0 | 420 ms | 350 ms | 500 ms | Minor improvements |
+| 2026-03-10 | 1.2.0 | 380 ms | 310 ms | 490 ms | DB indexes added |
+| 2026-04-05 | 1.3.0 | 320 ms | 280 ms | 480 ms | Query optimization |
+| 2026-05-01 | 1.4.0 | 290 ms | 260 ms | 470 ms | Cache added |
+| 2026-05-31 | 1.4.2 | 165 ms | 195 ms | 320 ms | N+1 fix + new indexes |
+
+**Trend visualization (ASCII chart — p95 GET /api/users):**
+```
+450 ─ ●
+420 ─   ●
+380 ─     ●
+320 ─       ●
+290 ─         ●
+165 ─             ●   ← Current (v1.4.2)
+    ─────────────────
+    1.0  1.1  1.2  1.3  1.4  1.4.2
+```
+
+---
+
+## Regression Detection Queries
+
+### Find all regressions in the log
+```bash
+grep -n "REGRESSION" observability/PERF_LOG.md
+```
+
+### Find all entries for a specific endpoint
+```bash
+grep -A 20 "GET /api/users" observability/PERF_LOG.md | grep "p95"
+```
+
+### Find entries where memory leak was suspected
+```bash
+grep -B 2 "Leak assessment: SUSPECT\|CONFIRMED" observability/PERF_LOG.md
+```
+
+### Extract p95 trend for charting (requires jq and baseline JSONs)
+```bash
+ls observability/baselines/*.json | sort | while read f; do
+  version=$(jq -r '.version' "$f")
+  p95=$(jq -r '.endpoints["GET /api/users"].p95_ms' "$f")
+  echo "$version: $p95 ms"
+done
+```
+
+---
+
+## SLA Budget Tracking
+
+Track how much of the SLA "budget" each service version consumes:
+
+| Endpoint | SLA (p95) | Latest p95 | Budget used | Budget remaining | Risk level |
+|----------|-----------|-----------|------------|-----------------|-----------|
+| GET /api/users | 200 ms | 165 ms | 82.5% | 35 ms (17.5%) | 🟡 MEDIUM |
+| GET /api/users/:id | 200 ms | 72 ms | 36% | 128 ms (64%) | 🟢 LOW |
+| POST /api/orders | 300 ms | 195 ms | 65% | 105 ms (35%) | 🟢 LOW |
+| GET /api/products | 150 ms | 88 ms | 58.7% | 62 ms (41.3%) | 🟢 LOW |
+| POST /auth/login | 500 ms | 320 ms | 64% | 180 ms (36%) | 🟢 LOW |
+
+**Risk levels:**
+- 🔴 HIGH: > 90% budget used (< 10% headroom). Any regression will breach SLA.
+- 🟡 MEDIUM: 75-90% budget used (10-25% headroom). Watch carefully.
+- 🟢 LOW: < 75% budget used (> 25% headroom). Comfortable buffer.
+
+---
+
+## Incident Correlation
+
+When a performance incident occurs, link it here for post-mortem reference:
+
+| Incident | Date | Affected endpoints | p95 during incident | Root cause | Resolution | PERF_LOG entry |
+|----------|------|-------------------|--------------------|-----------|-----------|----|
+| INC-001 | 2026-03-15 | GET /api/users | 3200 ms (16× SLA) | Missing index after schema migration | Added index CONCURRENTLY | 2026-03-15 entry |
+| INC-002 | 2026-04-22 | ALL endpoints | 1500 ms | Memory exhaustion (OOM) on 2 of 4 pods | Increased memory limit; fixed memory leak | 2026-04-23 entry |
+
+### Incident-to-optimization pipeline
+
+When an incident reveals a performance root cause:
+1. Add entry to this incident correlation table.
+2. Create an `OPTIMIZATION RECOMMENDATION` entry in `PERF_LOG.md`.
+3. After fix is deployed, add an after-baseline entry showing recovery.
+4. Close the incident with a link to the PERF_LOG entry.
+
+---
+
+## Observability Maturity Indicators
+
+Track improvement over time in observability completeness:
+
+| Capability | Current status | Target | Progress |
+|-----------|---------------|--------|----------|
+| Baseline automation | Manual | Automated on merge | 60% |
+| Regression gating in CI | Partial (p95 only) | All metrics | 50% |
+| Real-time performance dashboard | Grafana dashboard exists | Alert on breach | 70% |
+| Incident → optimization pipeline | Manual | Semi-automated | 30% |
+| Soak test automation | Manual | Weekly scheduled | 10% |
+| Load test data realism | Synthetic | Production traffic replay | 25% |