From b42a70ff3bc257f82746a655b6426bf10be87f78 Mon Sep 17 00:00:00 2001
From: wangjia <809946525@qq.com>
Date: Sun, 21 Jun 2026 11:39:28 +0800
Subject: [PATCH] =?UTF-8?q?ci:=20=E4=BF=AE=E5=A4=8D=E6=95=B0=E6=8D=AE?=
 =?UTF-8?q?=E5=BA=93=E5=A4=87=E4=BB=BD=20workflow=E2=80=94=E2=80=94?=
 =?UTF-8?q?=E6=94=B9=E7=94=A8=20mac=20runner=E3=80=81EC2=20=E7=AB=AF?=
 =?UTF-8?q?=E8=AF=BB=E5=AF=86=E7=A0=81=E3=80=81=E6=9C=AC=E6=9C=BA=E8=90=BD?=
 =?UTF-8?q?=E7=9B=98=E3=80=81=E6=AF=8F=E6=97=A5=E5=AE=9A=E6=97=B6?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

backup.yml 原 runs-on ubuntu-latest 无对应 runner、落盘 NAS 路径在 mac runner 不存在，长期跑不起来。
改为 runs-on mac + scripts/ci/backup-db.sh：复用 lib-forgejo 的 ssh 封装连 EC2，
在 EC2 端从 production.env 读 DSN 密码 mysqldump（零额外 secret），流回本机 ~/jiu-db-backups
gzip 落盘、保留 30 天、完整性校验，恢复每日 02:00 定时 + Telegram 成败通知。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YZ4DskSRKsSiheQonFtQvx
---
 .gitea/workflows/backup.yml | 44 ++++++++++++++++++-------------------
 scripts/ci/backup-db.sh     | 41 ++++++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+), 23 deletions(-)
 create mode 100644 scripts/ci/backup-db.sh

diff --git a/.gitea/workflows/backup.yml b/.gitea/workflows/backup.yml
index fc4dbb9..1ffc8b9 100644
--- a/.gitea/workflows/backup.yml
+++ b/.gitea/workflows/backup.yml
@@ -1,37 +1,35 @@
 name: DB Backup
 
 on:
-  # 定时备份已暂停（保留手动触发）。恢复时取消下面 schedule 的注释即可。
-  # schedule:
-  #   - cron: '0 18 * * *'  # UTC 18:00 = 北京时间 02:00
+  # 每日定时备份（北京时间 02:00）。手动触发亦可。
+  schedule:
+    - cron: '0 18 * * *'  # UTC 18:00 = 北京时间 02:00
   workflow_dispatch:
 
+concurrency:
+  group: db-backup
+  cancel-in-progress: false
+
 jobs:
   backup:
-    runs-on: ubuntu-latest
+    runs-on: mac
     steps:
-      - name: Setup SSH
-        run: |
-          mkdir -p ~/.ssh
-          printf '%s' "${{ secrets.EC2_SSH_KEY }}" > ~/.ssh/ec2.pem
-          chmod 600 ~/.ssh/ec2.pem
-          ssh-keyscan -H ${{ secrets.EC2_HOST }} >> ~/.ssh/known_hosts
+      - uses: actions/checkout@v4
 
-      - name: Dump MySQL to NAS
+      - name: Dump MySQL to local backup dir
         env:
+          EC2_SSH_KEY: ${{ secrets.EC2_SSH_KEY }}
           EC2_HOST: ${{ secrets.EC2_HOST }}
           EC2_USER: ${{ secrets.EC2_USER }}
-          DB_PASSWORD: ${{ secrets.DB_PASSWORD }}
-        run: |
-          BACKUP_DIR=/volume1/docker/backups/jiu-db
-          mkdir -p $BACKUP_DIR
-          FILENAME="jiu_db_$(date +%Y%m%d_%H%M%S).sql.gz"
-          ssh -i ~/.ssh/ec2.pem ${EC2_USER}@${EC2_HOST} \
-            "docker exec jiu_mysql mysqldump -uroot -p${DB_PASSWORD} jiu_db" \
-            | gzip > ${BACKUP_DIR}/${FILENAME}
-          find $BACKUP_DIR -name "*.sql.gz" -mtime +30 -delete
-          echo "Saved: ${BACKUP_DIR}/${FILENAME}"
+        run: sh scripts/ci/backup-db.sh
 
-      - name: Cleanup SSH key
+      - name: Notify (Telegram)
         if: always()
-        run: rm -f ~/.ssh/ec2.pem
+        env:
+          TELEGRAM_TOKEN: ${{ secrets.TELEGRAM_TOKEN }}
+          TELEGRAM_CHAT_ID: ${{ secrets.TELEGRAM_CHAT_ID }}
+        run: |
+          if [ "${{ job.status }}" = "success" ]; then ICON="✅"; LABEL="数据库备份成功"; else ICON="❌"; LABEL="数据库备份失败"; fi
+          curl -f -X POST "https://api.telegram.org/bot${TELEGRAM_TOKEN}/sendMessage" \
+            -d "chat_id=${TELEGRAM_CHAT_ID}" \
+            --data-urlencode "text=${ICON} 岩美 ${LABEL}" > /dev/null || true
diff --git a/scripts/ci/backup-db.sh b/scripts/ci/backup-db.sh
new file mode 100644
index 0000000..a016749
--- /dev/null
+++ b/scripts/ci/backup-db.sh
@@ -0,0 +1,41 @@
+#!/usr/bin/env bash
+# backup-db.sh — dump the production MySQL (jiu_db) from EC2 and keep a gzip
+# snapshot on the mac (host) runner's local disk (~/jiu-db-backups), retaining
+# the last 30 days. The DB password is read on EC2 from production.env's
+# DATABASE_DSN, so no DB-password secret is needed.
+#
+# Requires env (same as deploy-server): EC2_SSH_KEY, EC2_HOST, EC2_USER.
+set -euo pipefail
+
+# shellcheck source=scripts/ci/lib-forgejo.sh
+. "$(dirname "$0")/lib-forgejo.sh"
+
+BACKUP_DIR="${HOME}/jiu-db-backups"
+mkdir -p "${BACKUP_DIR}"
+FILENAME="jiu_db_$(date +%Y%m%d_%H%M%S).sql.gz"
+DEST="${BACKUP_DIR}/${FILENAME}"
+
+setup_ssh
+trap teardown_ssh EXIT
+
+echo "==> backup-db: dumping jiu_db from ${EC2_HOST}"
+# On EC2: parse the DSN password from production.env, then mysqldump the
+# container to stdout; stream back over ssh and gzip locally. The heredoc is
+# single-quoted, so it runs verbatim on EC2 (no local expansion). ${SSH} carries
+# no -t, keeping stdout a clean dump stream.
+${SSH} "${EC2_USER}@${EC2_HOST}" 'bash -s' <<'ENDSSH' | gzip > "${DEST}"
+set -euo pipefail
+PW=$(python3 -c 'import re;e=open("/opt/jiu/config/production.env").read();d=re.search(r"DATABASE_DSN=(.*)",e).group(1).strip().strip(chr(34)).strip(chr(39));print(re.match(r"[^:]+:([^@]+)@",d).group(1))')
+exec docker exec -e MYSQL_PWD="${PW}" jiu_mysql mysqldump -uroot --single-transaction --no-tablespaces jiu_db
+ENDSSH
+
+# Integrity + sanity + retention.
+gzip -t "${DEST}"
+SIZE=$(stat -f%z "${DEST}" 2>/dev/null || stat -c%s "${DEST}")
+if [ "${SIZE}" -lt 100000 ]; then
+  echo "==> backup-db: dump suspiciously small (${SIZE} bytes), aborting" >&2
+  rm -f "${DEST}"
+  exit 1
+fi
+find "${BACKUP_DIR}" -name '*.sql.gz' -mtime +30 -delete
+echo "==> backup-db: saved ${DEST} (${SIZE} bytes)"