|
| 1 | +# Performance profiling workflow - runs every 2 days to detect performance regressions |
| 2 | + |
| 3 | +name: Performance Profiling |
| 4 | + |
| 5 | +on: |
| 6 | + schedule: |
| 7 | + - cron: '0 0 */2 * *' |
| 8 | + workflow_dispatch: |
| 9 | + |
| 10 | +permissions: |
| 11 | + contents: read |
| 12 | + |
| 13 | +jobs: |
| 14 | + profiling: |
| 15 | + runs-on: ubuntu-latest |
| 16 | + strategy: |
| 17 | + fail-fast: false |
| 18 | + matrix: |
| 19 | + benchmark: |
| 20 | + - name: chatagent_overhead |
| 21 | + script: profiling/chatagent_overhead.py |
| 22 | + description: "ChatAgent overhead vs direct API calls" |
| 23 | + num_runs: 20 |
| 24 | + threshold: 2.0 |
| 25 | + env: |
| 26 | + RESULT_FILE: profiling/results/${{ matrix.benchmark.name }}.json |
| 27 | + BASELINE_FILE: profiling/baselines/${{ matrix.benchmark.name }}.json |
| 28 | + |
| 29 | + steps: |
| 30 | + - uses: step-security/harden-runner@20cf305ff2072d973412fa9b1e3a4f227bda3c76 |
| 31 | + with: |
| 32 | + egress-policy: audit |
| 33 | + |
| 34 | + - uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 |
| 35 | + |
| 36 | + - uses: ./.github/actions/camel_install |
| 37 | + with: |
| 38 | + python-version: "3.10" |
| 39 | + |
| 40 | + - run: | |
| 41 | + source .venv/bin/activate |
| 42 | + uv pip install -e ".[all]" |
| 43 | +
|
| 44 | + - uses: actions/cache@v4 |
| 45 | + with: |
| 46 | + path: ${{ env.BASELINE_FILE }} |
| 47 | + key: profiling-${{ matrix.benchmark.name }}-${{ github.ref_name }} |
| 48 | + restore-keys: profiling-${{ matrix.benchmark.name }}- |
| 49 | + |
| 50 | + - name: Run benchmark |
| 51 | + env: |
| 52 | + OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }} |
| 53 | + run: | |
| 54 | + source .venv/bin/activate |
| 55 | + mkdir -p profiling/results profiling/baselines |
| 56 | + python ${{ matrix.benchmark.script }} --num-runs ${{ matrix.benchmark.num_runs }} --output "$RESULT_FILE" --quiet |
| 57 | +
|
| 58 | + - name: Check regression and create summary |
| 59 | + env: |
| 60 | + THRESHOLD: ${{ matrix.benchmark.threshold }} |
| 61 | + run: | |
| 62 | + echo "## ${{ matrix.benchmark.description }}" >> $GITHUB_STEP_SUMMARY |
| 63 | +
|
| 64 | + if [ ! -f "$RESULT_FILE" ]; then |
| 65 | + echo "❌ Benchmark failed" >> $GITHUB_STEP_SUMMARY |
| 66 | + exit 1 |
| 67 | + fi |
| 68 | +
|
| 69 | + # Show current results |
| 70 | + echo "### Results (threshold: ${THRESHOLD}x)" >> $GITHUB_STEP_SUMMARY |
| 71 | + echo "| Mode | Current (ms) | Baseline (ms) | Ratio | API Calls | Status |" >> $GITHUB_STEP_SUMMARY |
| 72 | + echo "|------|--------------|---------------|-------|-----------|--------|" >> $GITHUB_STEP_SUMMARY |
| 73 | +
|
| 74 | + FAILED=false |
| 75 | + for mode in $(jq -r '.metrics | keys[]' "$RESULT_FILE"); do |
| 76 | + CURRENT=$(jq -r ".metrics.\"$mode\".avg_overhead_ms" "$RESULT_FILE") |
| 77 | + EXPECTED_CALLS=$(jq -r ".metrics.\"$mode\".expected_api_calls" "$RESULT_FILE") |
| 78 | + INVALID_RUNS=$(jq -r ".metrics.\"$mode\".invalid_api_call_runs" "$RESULT_FILE") |
| 79 | +
|
| 80 | + # Check API call count |
| 81 | + if [ "$INVALID_RUNS" -gt 0 ]; then |
| 82 | + API_STATUS="❌ ${INVALID_RUNS} invalid" |
| 83 | + FAILED=true |
| 84 | + else |
| 85 | + API_STATUS="✅ ${EXPECTED_CALLS}/call" |
| 86 | + fi |
| 87 | +
|
| 88 | + # Check regression |
| 89 | + if [ -f "$BASELINE_FILE" ]; then |
| 90 | + BASELINE=$(jq -r ".metrics.\"$mode\".avg_overhead_ms // empty" "$BASELINE_FILE") |
| 91 | + if [ -n "$BASELINE" ] && [ "$BASELINE" != "null" ]; then |
| 92 | + RATIO=$(awk "BEGIN {printf \"%.2f\", $CURRENT / $BASELINE}") |
| 93 | + if awk "BEGIN {exit !($RATIO > $THRESHOLD)}"; then |
| 94 | + STATUS="❌ REGRESSION" |
| 95 | + FAILED=true |
| 96 | + else |
| 97 | + STATUS="✅ OK" |
| 98 | + fi |
| 99 | + echo "| $mode | $CURRENT | $BASELINE | ${RATIO}x | $API_STATUS | $STATUS |" >> $GITHUB_STEP_SUMMARY |
| 100 | + continue |
| 101 | + fi |
| 102 | + fi |
| 103 | + echo "| $mode | $CURRENT | - | - | $API_STATUS | 🆕 New |" >> $GITHUB_STEP_SUMMARY |
| 104 | + done |
| 105 | +
|
| 106 | + if [ "$FAILED" = "true" ]; then |
| 107 | + echo "" >> $GITHUB_STEP_SUMMARY |
| 108 | + echo "❌ **Check failed!**" >> $GITHUB_STEP_SUMMARY |
| 109 | + exit 1 |
| 110 | + fi |
| 111 | +
|
| 112 | + - name: Update baseline |
| 113 | + if: success() |
| 114 | + run: cp "$RESULT_FILE" "$BASELINE_FILE" |
| 115 | + |
| 116 | + - uses: actions/cache/save@v4 |
| 117 | + if: success() |
| 118 | + with: |
| 119 | + path: ${{ env.BASELINE_FILE }} |
| 120 | + key: profiling-${{ matrix.benchmark.name }}-${{ github.ref_name }}-${{ github.run_id }} |
| 121 | + |
| 122 | + - uses: actions/upload-artifact@v4 |
| 123 | + if: always() |
| 124 | + with: |
| 125 | + name: profiling-${{ matrix.benchmark.name }} |
| 126 | + path: profiling/results/${{ matrix.benchmark.name }}.json |
| 127 | + retention-days: 90 |
0 commit comments