20260326 完成数据库model迁移；20260327 增加一键replay脚本

2026-03-27 11:24:34 +08:00
parent fe8b89d829
commit cd0ba15286
28 changed files with 1956 additions and 930 deletions
--- a/scripts/replay_verify/README.md
+++ b/scripts/replay_verify/README.md
@@ -0,0 +1,45 @@
+# Replay Verify Scripts
+
+## `run_multi_rir_ccr_replay_verify.sh`
+
+用途：
+- 通用 multi-RIR CCR replay verify 入口
+- 通过 `--rir` 指定一个或多个 RIR，按顺序执行
+- 通过 `--mode` 指定 `snapshot`、`delta` 或 `both`
+- 默认每个 RIR 的 RocksDB 目录在 compare/verify 结束后自动删除；传 `--keep-db` 才保留
+- 同一次执行的所有产物都会先落到 `rpki/target/replay/<timestamp>/`
+- 该时间戳目录下再按 RIR 分目录：
+  - `<rir>_ccr_replay_<timestamp>`
+
+默认输入：
+- bundle root: `/home/yuyr/dev/rust_playground/routinator/bench/multi_rir_demo/runs/20260316-112341-multi-final3`
+- 每个 RIR 的 TAL / TA / validation time / record CSV 由 `scripts/payload_replay/multi_rir_case_info.py` 解析
+
+用法：
+- 单个 RIR：
+  - `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic --mode both`
+  - `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic --mode snapshot`
+  - `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic --mode delta`
+- `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic --mode both`
+- `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic,ripe --mode snapshot`
+- `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir afrinic,apnic,arin,lacnic,ripe --mode both`
+- `./scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh --rir apnic --mode delta --keep-db`
+
+可覆盖环境变量：
+- `BUNDLE_ROOT`
+- `OUT_ROOT`（默认：`rpki/target/replay`）
+- `RUN_TAG`
+
+主要产物：
+- 单次执行根目录：
+  - `rpki/target/replay/<timestamp>/`
+- 每个 RIR 子目录下：
+  - `<rir>_snapshot.ccr`
+  - `<rir>_delta.ccr`
+  - `<rir>_*_report.json`
+  - `<rir>_*_ccr_vrps.csv`
+  - `<rir>_*_ccr_compare_summary.md`
+  - `<rir>_*_ccr_verify.json`
+- 同次执行总汇总：
+  - `multi_rir_ccr_replay_verify_<timestamp>_summary.md`
+  - `multi_rir_ccr_replay_verify_<timestamp>_summary.json`
--- a/scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh
+++ b/scripts/replay_verify/run_multi_rir_ccr_replay_verify.sh
@@ -0,0 +1,284 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
+cd "$ROOT_DIR"
+
+usage() {
+  cat <<'USAGE'
+Usage:
+  run_multi_rir_ccr_replay_verify.sh --rir <rir[,rir...]> [--mode snapshot|delta|both] [--keep-db]
+
+Options:
+  --rir <list>       Comma-separated RIR list, e.g. apnic or apnic,ripe
+  --mode <mode>      snapshot | delta | both (default: both)
+  --keep-db          Keep per-run RocksDB directories (default: remove after verify)
+  --bundle-root <p>  Override bundle root
+  --out-root <p>     Override output root (default: rpki/target/replay)
+  --run-tag <tag>    Override timestamp suffix for all RIR runs
+USAGE
+}
+
+MODE="both"
+KEEP_DB=0
+RIR_LIST=""
+BUNDLE_ROOT="${BUNDLE_ROOT:-/home/yuyr/dev/rust_playground/routinator/bench/multi_rir_demo/runs/20260316-112341-multi-final3}"
+OUT_ROOT="${OUT_ROOT:-$ROOT_DIR/target/replay}"
+RUN_TAG="${RUN_TAG:-$(date -u +%Y%m%dT%H%M%SZ)}"
+
+while [[ $# -gt 0 ]]; do
+  case "$1" in
+    --rir)
+      shift
+      RIR_LIST="${1:-}"
+      ;;
+    --mode)
+      shift
+      MODE="${1:-}"
+      ;;
+    --keep-db)
+      KEEP_DB=1
+      ;;
+    --bundle-root)
+      shift
+      BUNDLE_ROOT="${1:-}"
+      ;;
+    --out-root)
+      shift
+      OUT_ROOT="${1:-}"
+      ;;
+    --run-tag)
+      shift
+      RUN_TAG="${1:-}"
+      ;;
+    -h|--help)
+      usage
+      exit 0
+      ;;
+    *)
+      echo "unknown argument: $1" >&2
+      usage >&2
+      exit 2
+      ;;
+  esac
+  shift || true
+done
+
+if [[ -z "$RIR_LIST" ]]; then
+  echo "--rir is required" >&2
+  usage >&2
+  exit 2
+fi
+case "$MODE" in
+  snapshot|delta|both) ;;
+  *)
+    echo "invalid --mode: $MODE" >&2
+    usage >&2
+    exit 2
+    ;;
+esac
+
+CASE_INFO_SCRIPT="$ROOT_DIR/scripts/payload_replay/multi_rir_case_info.py"
+mkdir -p "$OUT_ROOT"
+RUN_ROOT="$OUT_ROOT/$RUN_TAG"
+mkdir -p "$RUN_ROOT"
+
+cargo build --release --bin rpki --bin ccr_to_routinator_csv --bin ccr_verify >/dev/null
+
+summary_md="$RUN_ROOT/multi_rir_ccr_replay_verify_${RUN_TAG}_summary.md"
+summary_json="$RUN_ROOT/multi_rir_ccr_replay_verify_${RUN_TAG}_summary.json"
+
+python3 - <<'PY' >/dev/null
+PY
+
+summary_json_tmp="$(mktemp)"
+printf '[]' > "$summary_json_tmp"
+
+run_one_mode() {
+  local rir="$1"
+  local mode="$2"
+  local run_dir="$3"
+  local trust_anchor="$4"
+  local tal_path="$5"
+  local ta_path="$6"
+  local base_archive="$7"
+  local base_locks="$8"
+  local base_csv="$9"
+  local delta_archive="${10}"
+  local delta_locks="${11}"
+  local delta_csv="${12}"
+  local snapshot_validation_time="${13}"
+  local delta_validation_time="${14}"
+
+  local db_dir="$run_dir/${rir}_${mode}_db"
+  local report_json="$run_dir/${rir}_${mode}_report.json"
+  local run_log="$run_dir/${rir}_${mode}_run.log"
+  local ccr_path="$run_dir/${rir}_${mode}.ccr"
+  local csv_path="$run_dir/${rir}_${mode}_ccr_vrps.csv"
+  local compare_md="$run_dir/${rir}_${mode}_ccr_compare_summary.md"
+  local only_ours="$run_dir/${rir}_${mode}_ccr_only_in_ours.csv"
+  local only_record="$run_dir/${rir}_${mode}_ccr_only_in_record.csv"
+  local verify_json="$run_dir/${rir}_${mode}_ccr_verify.json"
+  local meta_json="$run_dir/${rir}_${mode}_meta.json"
+
+  rm -rf "$db_dir"
+
+  local -a cmd=(target/release/rpki --db "$db_dir" --tal-path "$tal_path" --ta-path "$ta_path")
+  if [[ "$mode" == "snapshot" ]]; then
+    cmd+=(--payload-replay-archive "$base_archive" --payload-replay-locks "$base_locks" --validation-time "$snapshot_validation_time")
+  else
+    cmd+=(
+      --payload-base-archive "$base_archive"
+      --payload-base-locks "$base_locks"
+      --payload-base-validation-time "$snapshot_validation_time"
+      --payload-delta-archive "$delta_archive"
+      --payload-delta-locks "$delta_locks"
+      --validation-time "$delta_validation_time"
+    )
+  fi
+  cmd+=(--report-json "$report_json" --ccr-out "$ccr_path")
+
+  local start_s end_s duration_s
+  start_s="$(date +%s)"
+  (
+    echo "# ${rir} ${mode} command:"
+    printf '%q ' "${cmd[@]}"
+    echo
+    echo
+    "${cmd[@]}"
+  ) 2>&1 | tee "$run_log" >/dev/null
+  end_s="$(date +%s)"
+  duration_s="$((end_s - start_s))"
+
+  target/release/ccr_to_routinator_csv \
+    --ccr "$ccr_path" \
+    --out "$csv_path" \
+    --trust-anchor "$trust_anchor" >/dev/null
+
+  local record_csv
+  if [[ "$mode" == "snapshot" ]]; then
+    record_csv="$base_csv"
+  else
+    record_csv="$delta_csv"
+  fi
+  ./scripts/payload_replay/compare_with_routinator_record.sh \
+    "$csv_path" \
+    "$record_csv" \
+    "$compare_md" \
+    "$only_ours" \
+    "$only_record" >/dev/null
+
+  target/release/ccr_verify \
+    --ccr "$ccr_path" \
+    --db "$db_dir" > "$verify_json"
+
+  python3 - "$report_json" "$meta_json" "$mode" "$duration_s" <<'PY'
+import json, sys
+from pathlib import Path
+report = json.loads(Path(sys.argv[1]).read_text(encoding='utf-8'))
+meta = {
+  'mode': sys.argv[3],
+  'duration_seconds': int(sys.argv[4]),
+  'validation_time': report.get('validation_time_rfc3339_utc'),
+  'publication_points_processed': report['tree']['instances_processed'],
+  'publication_points_failed': report['tree']['instances_failed'],
+  'vrps': len(report['vrps']),
+  'aspas': len(report['aspas']),
+}
+Path(sys.argv[2]).write_text(json.dumps(meta, ensure_ascii=False, indent=2)+'\n', encoding='utf-8')
+PY
+
+  if [[ "$KEEP_DB" -eq 0 ]]; then
+    rm -rf "$db_dir"
+  fi
+}
+
+IFS=',' read -r -a RIRS <<< "$RIR_LIST"
+for rir in "${RIRS[@]}"; do
+  rir="$(echo "$rir" | xargs)"
+  [[ -n "$rir" ]] || continue
+  eval "$(python3 "$CASE_INFO_SCRIPT" --bundle-root "$BUNDLE_ROOT" --rir "$rir" --format env)"
+
+  run_dir="$RUN_ROOT/${rir}_ccr_replay_${RUN_TAG}"
+  mkdir -p "$run_dir"
+
+  if [[ "$MODE" == "snapshot" || "$MODE" == "both" ]]; then
+    run_one_mode \
+      "$rir" snapshot "$run_dir" "$TRUST_ANCHOR" "$TAL_PATH" "$TA_PATH" \
+      "$PAYLOAD_REPLAY_ARCHIVE" "$PAYLOAD_REPLAY_LOCKS" "$ROUTINATOR_BASE_RECORD_CSV" \
+      "$PAYLOAD_DELTA_ARCHIVE" "$PAYLOAD_DELTA_LOCKS" "$ROUTINATOR_DELTA_RECORD_CSV" \
+      "$SNAPSHOT_VALIDATION_TIME" "$DELTA_VALIDATION_TIME"
+  fi
+  if [[ "$MODE" == "delta" || "$MODE" == "both" ]]; then
+    run_one_mode \
+      "$rir" delta "$run_dir" "$TRUST_ANCHOR" "$TAL_PATH" "$TA_PATH" \
+      "$PAYLOAD_REPLAY_ARCHIVE" "$PAYLOAD_REPLAY_LOCKS" "$ROUTINATOR_BASE_RECORD_CSV" \
+      "$PAYLOAD_DELTA_ARCHIVE" "$PAYLOAD_DELTA_LOCKS" "$ROUTINATOR_DELTA_RECORD_CSV" \
+      "$SNAPSHOT_VALIDATION_TIME" "$DELTA_VALIDATION_TIME"
+  fi
+
+  python3 - "$summary_json_tmp" "$run_dir" "$rir" "$MODE" <<'PY'
+import json, sys
+from pathlib import Path
+summary_path = Path(sys.argv[1])
+run_dir = Path(sys.argv[2])
+rir = sys.argv[3]
+mode = sys.argv[4]
+rows = json.loads(summary_path.read_text(encoding='utf-8'))
+for submode in ['snapshot','delta']:
+    if mode not in ('both', submode):
+        continue
+    compare = run_dir / f'{rir}_{submode}_ccr_compare_summary.md'
+    meta = run_dir / f'{rir}_{submode}_meta.json'
+    verify = run_dir / f'{rir}_{submode}_ccr_verify.json'
+    if not compare.exists() or not meta.exists() or not verify.exists():
+        continue
+    compare_text = compare.read_text(encoding='utf-8')
+    meta_obj = json.loads(meta.read_text(encoding='utf-8'))
+    verify_obj = json.loads(verify.read_text(encoding='utf-8'))
+    def metric(name):
+        prefix = f'| {name} | '
+        for line in compare_text.splitlines():
+            if line.startswith(prefix):
+                return int(line.split('|')[2].strip())
+        raise SystemExit(f'missing metric {name} in {compare}')
+    rows.append({
+        'rir': rir,
+        'mode': submode,
+        'run_dir': str(run_dir),
+        'duration_seconds': meta_obj['duration_seconds'],
+        'vrps': meta_obj['vrps'],
+        'aspas': meta_obj['aspas'],
+        'only_in_ours': metric('only_in_ours'),
+        'only_in_record': metric('only_in_record'),
+        'intersection': metric('intersection'),
+        'state_hashes_ok': verify_obj.get('state_hashes_ok'),
+    })
+summary_path.write_text(json.dumps(rows, ensure_ascii=False, indent=2)+'\n', encoding='utf-8')
+PY
+
+done
+
+python3 - "$summary_json_tmp" "$summary_json" "$summary_md" "$RUN_TAG" <<'PY'
+import json, sys
+from pathlib import Path
+rows = json.loads(Path(sys.argv[1]).read_text(encoding='utf-8'))
+out_json = Path(sys.argv[2])
+out_md = Path(sys.argv[3])
+run_tag = sys.argv[4]
+out_json.write_text(json.dumps(rows, ensure_ascii=False, indent=2)+'\n', encoding='utf-8')
+parts = []
+parts.append('# Multi-RIR CCR Replay Verify Summary\n\n')
+parts.append(f'- run_tag: `{run_tag}`\n\n')
+parts.append('| rir | mode | duration_s | vrps | aspas | only_in_ours | only_in_record | state_hashes_ok |\n')
+parts.append('|---|---|---:|---:|---:|---:|---:|---|\n')
+for row in rows:
+    parts.append(f"| {row['rir']} | {row['mode']} | {row['duration_seconds']} | {row['vrps']} | {row['aspas']} | {row['only_in_ours']} | {row['only_in_record']} | {row['state_hashes_ok']} |\n")
+out_md.write_text(''.join(parts), encoding='utf-8')
+PY
+
+rm -f "$summary_json_tmp"
+
+echo "== multi-rir replay verify complete ==" >&2
+echo "- summary: $summary_md" >&2
+echo "- summary json: $summary_json" >&2