Proxmox VE 8 备份恢复最佳实践 2026：从快照到 PBS异地灾备完全指南

# 备份指定虚拟机（ID 100）
vzdump 100 --storage local-btrfs --mode snapshot --compress zstd

# 备份多个虚拟机
vzdump 100 101 102 --storage local-btrfs --mode snapshot --compress zstd

# 备份所有运行中的虚拟机
vzdump --storage local-btrfs --mode snapshot --compress zstd --stdout

# 指定备份保留策略
vzdump 100 --storage local-btrfs --mode snapshot --compress zstd \
  --retain-last 3 --retain-daily 7 --retain-weekly 4 --retain-monthly 3

# 查看定时备份任务
cat /etc/pve/jobs.cfg

# 查看备份日志
cat /var/log/vzdump.log
# 或
journalctl -u vzdump

# 恢复 KVM 虚拟机
qmrestore /var/lib/vz/dump/vzdump-qemu-100-2026_07_04.vma.zst 100 --storage local-lvm

# 恢复 LXC 容器
pct restore /var/lib/vz/dump/vzdump-lxc-101-2026_07_04.tar.zst 101 --storage local-lvm

# 恢复到新的 VM ID（不影响原虚拟机）
qmrestore /var/lib/vz/dump/vzdump-qemu-100-2026_07_04.vma.zst 200 --storage local-lvm

# 1. 解压备份文件
zstd -d vzdump-qemu-100-2026_07_04.vma.zst -o backup.vma

# 2. 使用 vma 工具提取磁盘
vma extract backup.vma /tmp/extracted/

# 3. 挂载提取的磁盘镜像
modprobe nbd
qemu-nbd -c /dev/nbd0 /tmp/extracted/disk-0.raw
mount /dev/nbd0p1 /mnt/recovery

# 4. 复制需要的文件
cp /mnt/recovery/path/to/file /restore/

# 5. 清理
umount /mnt/recovery
qemu-nbd -d /dev/nbd0

docker run -d \
  --name pbs \
  --restart always \
  -p 8007:8007 \
  -v /data/pbs:/var/lib/proxmox-backup \
  -v /data/pbs-etc:/etc/proxmox-backup \
  ghcr.io/proxmox/proxmox-backup-server:latest

# 在 PBS 上执行
proxmox-backup-manager cert info | grep Fingerprint

# 通过 PBS 备份
vzdump 100 --storage pbs-storage --mode snapshot

# 定时备份到 PBS（Web UI 配置）
# Datacenter → Backup → Add → Storage 选 PBS

# 手动触发垃圾回收
proxmox-backup-manager garbage-collection start pbs-datastore

# 查看垃圾回收状态
proxmox-backup-manager garbage-collection status pbs-datastore

# 每周日凌晨 3 点垃圾回收
echo "0 3 * * 0 root proxmox-backup-manager garbage-collection start pbs-datastore" >> /etc/cron.d/pbs-gc

# 创建快照
zfs snapshot rpool/data/vm-100-disk-0@backup-20260704

# 列出快照
zfs list -t snapshot

# 回滚到快照（⚠️ 会丢弃快照后的数据）
zfs rollback rpool/data/vm-100-disk-0@backup-20260704

# 删除快照
zfs destroy rpool/data/vm-100-disk-0@backup-20260704

#!/bin/bash
# ZFS 自动快照脚本
# 保留策略：每小时1个保留24个，每天1个保留7个，每周1个保留4个

# 按小时快照
for dataset in $(zfs list -H -o name | grep "vm-"); do
    zfs snapshot ${dataset}@hourly-$(date +%Y%m%d%H)
done

# 清理过期快照
# 删除24小时以上的 hourly 快照
zfs list -t snapshot -o name -H | grep "hourly-" | while read snap; do
    snap_date=$(echo $snap | grep -oP 'hourly-\K\d+')
    if [ $(date -d "${snap_date:0:8} ${snap_date:8:2}:00:00" +%s 2>/dev/null) ]; then
        age=$(( $(date +%s) - $(date -d "${snap_date:0:8} ${snap_date:8:2}:00:00" +%s) ))
        if [ $age -gt 86400 ]; then
            zfs destroy $snap
        fi
    fi
done

# 每小时快照
0 * * * * /opt/zfs-snapshot.sh

# 安装 sanoid
apt install sanoid -y

# 配置
cp /etc/sanoid/sanoid.defaults.conf /etc/sanoid/sanoid.conf
nano /etc/sanoid/sanoid.conf

[template production]
    frequently = 0
    hourly = 24
    daily = 7
    weekly = 4
    monthly = 3
    yearly = 1
    autosnap = yes
    autoprune = yes

[rpool/data/vm-100-disk-0]
    use_template = production

# 启用定时任务
systemctl enable sanoid.timer
systemctl start sanoid.timer

# 添加远程 PBS
proxmox-backup-manager remote add pbs-remote 远程PBS地址 \
  --auth-id root@pam \
  --password '远程PBS密码'

# 创建同步任务
proxmox-backup-manager sync-job create \
  --remote pbs-remote \
  --remote-store main-datastore \
  --store local-datastore \
  --schedule "03:00"

#!/bin/bash
# ZFS 远程备份脚本
LOCAL_POOL="rpool/data"
REMOTE_HOST="user@backup-server"
REMOTE_POOL="backup/data"

# 增量发送所有变化的快照
zfs list -t snapshot -o name -H -s creation | grep "@daily-" | while read snap; do
    prev_snap=$(zfs list -t snapshot -o name -H -s creation | grep "@daily-" | grep -B1 "$snap" | head -1)
    
    if [ -z "$prev_snap" ] || [ "$prev_snap" = "$snap" ]; then
        # 首次全量发送
        zfs send "$snap" | ssh "$REMOTE_HOST" zfs recv "$REMOTE_POOL/${snap#*/}"
    else
        # 增量发送
        zfs send -i "$prev_snap" "$snap" | ssh "$REMOTE_HOST" zfs recv "$REMOTE_POOL/${snap#*/}"
    fi
done

# 安装 rclone
curl https://rclone.org/install.sh | bash

# 配置云存储
rclone config
# 选择 n (new remote) → 命名 → 选择存储类型 → 填入密钥

# 同步备份目录到云端
rclone sync /var/lib/vz/dump/ remote:pbs-backup/ --transfers 4 --checkers 8

# 每天备份完成后同步到云端（4点开始）
0 4 * * * rclone sync /var/lib/vz/dump/ remote:pbs-backup/ --transfers 4 --checkers 8 --log-file /var/log/rclone.log

#!/bin/bash
# PVE 备份恢复演练脚本
# 用法：./backup-drill.sh 100

VM_ID=$1
DRILL_ID=9900  # 演练用 VM ID
STORAGE="local-lvm"

if [ -z "$VM_ID" ]; then
    echo "用法：$0 <VM_ID>"
    exit 1
fi

echo "=== 备份恢复演练 ==="
echo "源虚拟机：$VM_ID"
echo "演练虚拟机：$DRILL_ID"

# 1. 查找最新备份
LATEST_BACKUP=$(ls -t /var/lib/vz/dump/vzdump-qemu-${VM_ID}-*.vma.zst 2>/dev/null | head -1)
if [ -z "$LATEST_BACKUP" ]; then
    echo "❌ 未找到虚拟机 $VM_ID 的备份文件"
    exit 1
fi
echo "最新备份：$LATEST_BACKUP"

# 2. 删除旧的演练虚拟机（如果存在）
if qm status $DRILL_ID &>/dev/null; then
    echo "清理旧的演练虚拟机..."
    qm stop $DRILL_ID 2>/dev/null
    qm destroy $DRILL_ID --purge
fi

# 3. 恢复备份
echo "开始恢复..."
qmrestore "$LATEST_BACKUP" $DRILL_ID --storage $STORAGE

if [ $? -ne 0 ]; then
    echo "❌ 恢复失败！"
    exit 1
fi

# 4. 启动虚拟机
echo "启动演练虚拟机..."
qm start $DRILL_ID

# 5. 等待启动
echo "等待虚拟机启动（60秒）..."
sleep 60

# 6. 检查状态
STATUS=$(qm status $DRILL_ID --raw | grep status | awk -F'"' '{print $4}')
if [ "$STATUS" = "running" ]; then
    echo "✅ 演练成功！虚拟机 $DRILL_ID 已启动"
    echo "请手动验证服务和数据，然后运行："
    echo "  qm stop $DRILL_ID && qm destroy $DRILL_ID --purge"
else
    echo "❌ 虚拟机未正常运行，状态：$STATUS"
fi

echo ""
echo "=== 演练完成 ==="
echo "$(date '+%Y-%m-%d %H:%M:%S') 演练结果：$STATUS" >> /var/log/backup-drill.log

# 查看备份任务状态
pvesh get /cluster/backup-info

# 查看最近备份
find /var/lib/vz/dump/ -name "*.vma.zst" -mtime -1 -ls

# 检查 PBS 备份状态
proxmox-backup-client snapshot list --repository root@pam@localhost:pbs-datastore

#!/bin/bash
# 备份监控告警脚本
# 检查最近 24 小时内是否有备份失败

LOG="/var/log/vzdump.log"
ALERT_EMAIL="yourname@localhost"

# 检查 vzdump 日志中的错误
ERRORS=$(grep -i "error\|failed\|warning" "$LOG" 2>/dev/null | tail -20)

if [ -n "$ERRORS" ]; then
    echo "PVE 备份告警：发现备份错误" | mail -s "⚠️ PVE Backup Alert" "$ALERT_EMAIL"
    echo "$ERRORS" | mail -s "PVE Backup Error Details" "$ALERT_EMAIL"
fi

# 检查是否超过 24 小时没有备份
LAST_BACKUP=$(find /var/lib/vz/dump/ -name "*.vma.zst" -mtime -1 2>/dev/null | head -1)
if [ -z "$LAST_BACKUP" ]; then
    echo "PVE 备份告警：超过 24 小时没有新备份！" | mail -s "🚨 PVE No Backup Alert" "$ALERT_EMAIL"
fi

# 验证 vzdump 备份
zstd -t /var/lib/vz/dump/vzdump-qemu-100-*.vma.zst

# 验证 PBS 备份
proxmox-backup-client verify snapshot.vma /path/to/snapshot --repository root@pam@localhost:pbs-datastore