From e1a17acbe41e762975395cbedadfa887b37677cf Mon Sep 17 00:00:00 2001
From: wangjia <809946525@qq.com>
Date: Mon, 6 Jul 2026 14:55:11 +0800
Subject: [PATCH] =?UTF-8?q?feat(ci):=20SQLite=20=E6=AF=8F=E6=97=A5?=
 =?UTF-8?q?=E5=A4=87=E4=BB=BD=E5=88=B0=20NAS(=E5=BC=82=E5=9C=B0=E5=AE=B9?=
 =?UTF-8?q?=E7=81=BE,#26/F2)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

backup-db.sh:ssh pangolin1 → sqlite3 online .backup 取 WAL 一致快照 + 远端
integrity_check → gzip 流回 → 存 NAS /volume1/docker/backups/pangolin,保留 30 天。
backup.yml:每日 02:00(北京)cron + 手动,ubuntu-latest,复用 DEPLOY_SSH_KEY,
Telegram 通知。异地=家里 NAS≠VPS 机房。不在 pangolin1 装/常驻任何东西。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 .gitea/workflows/backup.yml | 36 +++++++++++++++++++++++++
 .gitea/workflows/ci.yml     |  3 ++-
 scripts/ci/backup-db.sh     | 53 +++++++++++++++++++++++++++++++++++++
 3 files changed, 91 insertions(+), 1 deletion(-)
 create mode 100644 .gitea/workflows/backup.yml
 create mode 100644 scripts/ci/backup-db.sh

diff --git a/.gitea/workflows/backup.yml b/.gitea/workflows/backup.yml
new file mode 100644
index 0000000..8cab5dc
--- /dev/null
+++ b/.gitea/workflows/backup.yml
@@ -0,0 +1,36 @@
+name: DB Backup
+
+on:
+  # 每日北京 02:00(= UTC 18:00)自动备份;也可手动触发。
+  schedule:
+    - cron: '0 18 * * *'
+  workflow_dispatch:
+
+concurrency:
+  group: db-backup
+  cancel-in-progress: false
+
+jobs:
+  backup:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Backup pangolin1 SQLite → NAS
+        env:
+          DEPLOY_SSH_KEY: ${{ secrets.DEPLOY_SSH_KEY }}
+        run: bash scripts/ci/backup-db.sh
+
+      - name: Notify (Telegram)
+        if: always()
+        env:
+          TELEGRAM_BOT_TOKEN: ${{ secrets.TELEGRAM_BOT_TOKEN }}
+          TELEGRAM_CHAT_ID: ${{ secrets.TELEGRAM_CHAT_ID }}
+        run: |
+          . scripts/ci/notify.sh
+          if [ "${{ job.status }}" = "success" ]; then
+            notify_ok "pangolin1 SQLite 备份成功(NAS)"
+          else
+            notify_fail "pangolin1 SQLite 备份失败"
+          fi
diff --git a/.gitea/workflows/ci.yml b/.gitea/workflows/ci.yml
index 99e5673..1c227b8 100644
--- a/.gitea/workflows/ci.yml
+++ b/.gitea/workflows/ci.yml
@@ -54,7 +54,8 @@ jobs:
               /mnt/scripts/ci/compile-backend.sh \
               /mnt/scripts/ci/release-server.sh \
               /mnt/scripts/ci/deploy-server.sh \
-              /mnt/scripts/ci/test.sh
+              /mnt/scripts/ci/test.sh \
+              /mnt/scripts/ci/backup-db.sh
 
   # ── Job 2: OpenAPI Sync Check ────────────────────────────────────────────
   openapi-check:
diff --git a/scripts/ci/backup-db.sh b/scripts/ci/backup-db.sh
new file mode 100644
index 0000000..b1b4ea7
--- /dev/null
+++ b/scripts/ci/backup-db.sh
@@ -0,0 +1,53 @@
+#!/usr/bin/env bash
+# backup-db.sh — pangolin1 控制面 SQLite 每日备份到 NAS(异地容灾,#26 / F2)。
+#
+# 经 ssh 在 pangolin1 上用 sqlite3 online `.backup` 取 WAL 一致快照(不阻塞服务、
+# 不像裸 cp 那样可能拿到半写状态),远端先跑 PRAGMA integrity_check 确认没坏,再把
+# 快照 gzip 流回 runner,落到 NAS 备份盘(默认 /volume1/docker/backups/pangolin,
+# 该盘已挂进 runner 容器)。保留 30 天。异地 = 家里 NAS ≠ VPS 机房,构成容灾。
+#
+# 不在 pangolin1 上安装/常驻任何东西:调度在 Gitea Actions(backup.yml),每次经
+# ssh 临时跑一次 sqlite3。需环境:DEPLOY_SSH_KEY(secret)。从 repo 根调用。
+set -euo pipefail
+
+# shellcheck source=scripts/ci/lib-ssh.sh
+. scripts/ci/lib-ssh.sh
+
+BACKUP_DIR="${BACKUP_DIR:-/volume1/docker/backups/pangolin}"
+mkdir -p "$BACKUP_DIR"
+
+# 时间戳文件名(bash 内建 strftime,免 $() 命令替换)。
+printf -v TS '%(%Y%m%d_%H%M%S)T' -1
+DEST="${BACKUP_DIR}/pangolin_${TS}.db.gz"
+
+# setup_ssh 内部先注册 EXIT trap 清理临时私钥,再写 key/known_hosts,导出 $SSH。
+setup_ssh
+
+echo "==> backup-db: sqlite3 .backup pangolin1(${DEPLOY_HOST})-> ${DEST}"
+# 远端单引号 heredoc:$BK/$$ 在 pangolin1 上展开。以 pangolin 用户(DB 属主)跑
+# online .backup + integrity_check;校验通过才 cat 回流;临时文件用完即删。
+# integrity_check 失败 → 远端非零退出 → 本地 pipefail 令整条命令失败 → set -e 中止。
+$SSH "root@${DEPLOY_HOST}" 'bash -s' <<'ENDSSH' | gzip > "$DEST"
+set -euo pipefail
+BK="/tmp/pangolin-bk-$$.db"
+trap 'rm -f "$BK"' EXIT
+runuser -u pangolin -- sqlite3 "/var/lib/pangolin/pangolin.db" ".backup ${BK}"
+if ! runuser -u pangolin -- sqlite3 "$BK" "PRAGMA integrity_check;" | grep -q '^ok$'; then
+  echo "backup-db(remote): integrity_check 失败" >&2
+  exit 1
+fi
+cat "$BK"
+ENDSSH
+
+# 本地二次校验:gzip 完好 + 大小 sanity(空/损坏库会很小)。
+gzip -t "$DEST"
+if ! find "$DEST" -size +50k -print -quit | grep -q .; then
+  echo "==> backup-db: 备份异常小(<50k),中止" >&2
+  rm -f "$DEST"
+  exit 1
+fi
+
+# 保留最近 30 天。
+find "$BACKUP_DIR" -name 'pangolin_*.db.gz' -mtime +30 -delete
+
+echo "==> backup-db: 备份完成 ${DEST}"