GitHub 稳定性崩塌：AI 流量 3.5 倍暴增与 squash merge 数据损坏的教训

#!/usr/bin/env bash
# check_squash_integrity.sh
# 检查指定仓库在 2025-04-20 ~ 2025-04-25 期间合并的 PR 是否存在异常
# 使用前需要设置 GH_TOKEN 和 REPO

set -euo pipefail

REPO="${REPO:-owner/repo}"       # 改成你的仓库，格式 owner/repo
START_DATE="2025-04-20"
END_DATE="2025-04-25"
GH_TOKEN="${GH_TOKEN:-}"         # 设置你的 GitHub token

if [ -z "$GH_TOKEN" ]; then
  echo "请设置 GH_TOKEN 环境变量"
  exit 1
fi

API="https://api.github.com/repos/${REPO}/pulls"

# 拉取指定时间范围内已合并的 PR
echo "=== 检查 ${REPO} 在 ${START_DATE} ~ ${END_DATE} 期间合并的 PR ==="

PR_LIST=$(curl -s -H "Authorization: token ${GH_TOKEN}" \
  "${API}?state=closed&per_page=100&sort=updated&direction=desc" \
  | jq -r --arg start "$START_DATE" --arg end "$END_DATE" \
    '.[] | select(.merged_at != null 
                  and .merged_at >= $start 
                  and .merged_at <= $end) 
     | .number')

SUSPECT_COUNT=0

for pr_num in $PR_LIST; do
  # 获取 PR 详情，检查 merge_commit_sha 是否有效
  pr_detail=$(curl -s -H "Authorization: token ${GH_TOKEN}" \
    "${API}/${pr_num}")

  merge_sha=$(echo "$pr_detail" | jq -r '.merge_commit_sha')
  merged_via=$(echo "$pr_detail" | jq -r '.merged_via // "unknown"')

  # 尝试在本地验证该 SHA 是否存在于主分支
  # 如果 merge_commit_sha 为 null 或在远程找不到，标记为可疑
  if [ "$merge_sha" = "null" ] || [ -z "$merge_sha" ]; then
    echo "[可疑] PR #${pr_num}: merge_commit_sha 为空，合并方式=${merged_via}"
    SUSPECT_COUNT=$((SUSPECT_COUNT + 1))
  else
    # 在远程检查该 commit 是否存在
    commit_check=$(curl -s -o /dev/null -w "%{http_code}" \
      -H "Authorization: token ${GH_TOKEN}" \
      "https://api.github.com/repos/${REPO}/commits/${merge_sha}")

    if [ "$commit_check" != "200" ]; then
      echo "[可疑] PR #${pr_num}: SHA ${merge_sha} 在远程不存在 (HTTP ${commit_check})，合并方式=${merged_via}"
      SUSPECT_COUNT=$((SUSPECT_COUNT + 1))
    fi
  fi
done

echo ""
echo "=== 总结：发现 ${SUSPECT_COUNT} 个可疑 PR ==="
if [ "$SUSPECT_COUNT" -gt 0 ]; then
  echo "建议：对这些 PR 重新检查代码完整性，必要时重新 squash merge"
fi

export GH_TOKEN="ghp_你的token"
export REPO="your-org/your-repo"
bash check_squash_integrity.sh

"""
github_availability_monitor.py
每 5 分钟探测 GitHub API，记录可用性数据到本地 CSV。
适合小团队快速建立基线，不依赖外部监控服务。
"""

import csv, time, requests, datetime, os

# ---- 配置 ----
GITHUB_TOKEN = os.getenv("GH_TOKEN", "")  # 可选，有 token 限流更宽松
INTERVAL_SECONDS = 300                     # 5 分钟
CSV_FILE = "github_availability.csv"
ENDPOINTS = [
    ("api_root",   "https://api.github.com"),
    ("user_api",   "https://api.github.com/user"),
    ("git_push",   "https://github.com"),   # 基础可达性
]

def probe(name: str, url: str) -> dict:
    headers = {"Authorization": f"token {GITHUB_TOKEN}"} if GITHUB_TOKEN else {}
    try:
        r = requests.get(url, headers=headers, timeout=10)
        return {
            "timestamp": datetime.datetime.utcnow().isoformat(),
            "endpoint": name,
            "status_code": r.status_code,
            "latency_ms": round(r.elapsed.total_seconds() * 1000),
            "ok": r.status_code < 500,
        }
    except requests.RequestException as e:
        return {
            "timestamp": datetime.datetime.utcnow().isoformat(),
            "endpoint": name,
            "status_code": 0,
            "latency_ms": -1,
            "ok": False,
            "error": str(e),
        }

def main():
    with open(CSV_FILE, "a", newline="") as f:
        writer = csv.DictWriter(f, fieldnames=["timestamp", "endpoint", "status_code", "latency_ms", "ok", "error"])
        if f.tell() == 0:
            writer.writeheader()

        while True:
            for name, url in ENDPOINTS:
                row = probe(name, url)
                writer.writerow(row)
                f.flush()
                status = "✓" if row["ok"] else "✗"
                print(f"{row['timestamp']}  {name}: {status}  {row['latency_ms']}ms")
            print(f"--- 等待 {INTERVAL_SECONDS}s ---")
            time.sleep(INTERVAL_SECONDS)

if __name__ == "__main__":
    main()