k8s备份

Ethereal Lv5
  2025-03-26 16:09:53 2025-03-26 16:09 Created   2025-07-04 23:45:34 2025-07-04 23:45 Updated  

1. 安装etcd客户端

1
apt install etcd-client

2. 创建备份

1
2
3
4
5
6
7
mkdir ./etcd-backup

ETCDCTL_API=3 etcdctl --endpoints=https://127.0.0.1:2379 \
                      --cacert=/etc/kubernetes/pki/etcd/ca.crt \
                      --cert=/etc/kubernetes/pki/etcd/server.crt \
                      --key=/etc/kubernetes/pki/etcd/server.key \
snapshot save ./etcd-backup/etcdbackup.db

其他备份

1
2
3
4
5
6
7
8
mkdir ~/confirm                               #备份源集群文件
cp -rf /etc/kubernetes/ ~/confirm/    
ll ~/confirm/kubernetes/

mkdir confirm/data_etcd                       #备份ETCD
cp -rf /var/lib/etcd/* ~/confirm/data_etcd  

cp /usr/bin/kubeadm /usr/bin/kubeadm.bak

3. 测试备份是否成功

1
2
3
4
ETCDCTL_API=3 etcdctl --write-out=table snapshot status ./etcd-backup/etcdbackup.db 


ETCDCTL_API=3 etcdctl snapshot restore etcd-backup/etcdbackup.db

4. 恢复

  1. 解码备份
1
ETCDCTL_API=3 etcdctl snapshot restore etcd-backup/etcdbackup.db
  1. 停止整个集群并保存原始文件
1
2
3
mkdir system_yaml_files
mv /etc/kubernetes/manifests/* system_yaml_files/ # 此时k8s将会无法连接
mv /var/lib/etcd/member/ /var/lib/etcd/member.bak
  1. 恢复数据
1
2
cd default.etcd/
mv  member/ /var/lib/etcd/
  1. 重启集群
1
2
3
systemctl stop kubelet
cp system_yaml_files/* /etc/kubernetes/manifests/
systemctl start kubelet

5. pod日志备份

5.1 日志目录

日志目录位于/var/log下面

5.2 方案一：周期性

Dockerfile

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
# 使用一个轻量的基础镜像
FROM alpine:3.18

# 安装必要的工具 (wget, tar) 并清理缓存
RUN apk add --no-cache wget tar

# 下载并安装 obsutil
# 前往华为云官网获取最新的 obsutil 下载链接
RUN wget https://obs-community.obs.cn-north-1.myhuaweicloud.com/obsutil/current/obsutil_linux_amd64.tar.gz -O /tmp/obsutil.tar.gz && \
    tar -C /usr/local/bin -xzf /tmp/obsutil.tar.gz && \
    chmod +x /usr/local/bin/obsutil_linux_amd64_*/obsutil && \
    rm /tmp/obsutil.tar.gz

# 创建一个符号链接，方便使用
RUN ln -s /usr/local/bin/obsutil_linux_amd64_*/obsutil /usr/local/bin/obsutil

# 验证安装
CMD ["obsutil", "version"]

cronjob

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
apiVersion: batch/v1
kind: CronJob
metadata:
  name: node-var-log-backup-to-obs
spec:
  # Cron 格式：分 时 日 月 周。 "0 3 * * *" 表示每天凌晨3点执行
  schedule: "0 3 * * *"
  jobTemplate:
    spec:
      # Job 在 10 分钟后如果还没完成，就超时
      activeDeadlineSeconds: 600
      template:
        spec:
          # 关键：选择要在哪个节点上运行备份。
          # 您需要为每个希望备份的节点创建一个类似的 CronJob，或者使用更高级的控制器。
          nodeSelector:
            kubernetes.io/hostname: <target-node-name> # <-- 修改为您的目标节点名
          containers:
          - name: log-backup-uploader
            # 使用您在步骤3中构建的镜像
            image: your-docker-repo/obsutil-alpine:latest # <-- 修改为您的镜像
            # 以 root 用户运行，以便有权限读取 /var/log
            securityContext:
              runAsUser: 0
            command: ["/bin/sh", "-c"]
            args:
            - |
              set -ex

              # --- 1. 配置 obsutil ---
              # 从 Secret 获取凭证，并配置 obsutil
              # OBS_ENDPOINT 是您 OBS 桶所在区域的入口地址
              OBS_BUCKET="k8s-log-backups" # <-- 修改为您的桶名
              OBS_ENDPOINT="obs.cn-north-4.myhuaweicloud.com" # <-- 修改为您的桶 Endpoint
              
              echo "Configuring obsutil..."
              obsutil config -i=$HUAWEI_AK -k=$HUAWEI_SK -e=$OBS_ENDPOINT

              # --- 2. 打包日志文件 ---
              NODE_NAME=$(cat /etc/nodename)
              BACKUP_FILE="/tmp/log-backup-${NODE_NAME}-$(date +%Y-%m-%d-%H%M).tar.gz"
              
              echo "Creating backup archive: ${BACKUP_FILE}"
              # -C /host/var/log . 表示切换到 /host/var/log 目录进行打包，避免包含绝对路径
              tar -czf ${BACKUP_FILE} -C /host/var/log .
              
              # --- 3. 上传到 OBS ---
              S3_PATH="obs://${OBS_BUCKET}/node-logs/${NODE_NAME}/"
              echo "Uploading ${BACKUP_FILE} to ${S3_PATH}"
              
              # 使用 obsutil 上传文件。-f 表示强制覆盖（如果需要）
              obsutil cp ${BACKUP_FILE} ${S3_PATH}
              
              # --- 4. 清理本地临时文件 ---
              echo "Cleaning up local backup file..."
              rm ${BACKUP_FILE}
              
              echo "Backup to OBS successfully completed."

            env:
            # 从我们创建的 Secret 中注入环境变量
            - name: HUAWEI_AK
              valueFrom:
                secretKeyRef:
                  name: huawei-obs-credentials
                  key: access_key_id
            - name: HUAWEI_SK
              valueFrom:
                secretKeyRef:
                  name: huawei-obs-credentials
                  key: secret_access_key
            volumeMounts:
            # 将节点的 /var/log 目录以只读方式挂载到 Pod 的 /host/var/log
            - name: node-var-log
              mountPath: /host/var/log
              readOnly: true
            # 挂载节点的 hostname 文件，方便在备份文件名中标识节点
            - name: node-name-file
              mountPath: /etc/nodename
              readOnly: true
          volumes:
          # 定义 hostPath 卷，指向节点的真实目录
          - name: node-var-log
            hostPath:
              path: /var/log
              type: Directory
          - name: node-name-file
            hostPath:
              path: /etc/hostname
          # 任务失败后不重启 Pod
          restartPolicy: OnFailure

5.3 方案二：实时备份

编写脚本到文件/usr/local/bin/sync_script.sh

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#!/bin/sh
set -e

# 定义源目录和目标目录
SOURCE_DIR="/var/log/pods"
DEST_DIR="/opt/k8s-log"

# 确保目标目录存在
mkdir -p "$DEST_DIR"

echo "Starting initial sync from $SOURCE_DIR to $DEST_DIR"
# 第一次启动时，先完整同步一次
rsync -av --delete "$SOURCE_DIR" "$DEST_DIR"

echo "Watching for changes in $SOURCE_DIR..."
# 使用 inotifywait 持续监控源目录中的创建、修改、删除、移动事件
# -m 表示持续监控，-r 表示递归，-q 表示静默模式
# --format '%w%f' 输出变化的文件路径
inotifywait -m -r -q -e create,modify,delete,move --format '%w%f' "$SOURCE_DIR" | while read FILE
do
    echo "Change detected. Syncing..."
    # 每次检测到变化，就执行一次 rsync
    # --delete 参数确保源目录删除的文件在目标目录也删除
    rsync -av --delete "$SOURCE_DIR" "$DEST_DIR"
    echo "Sync complete. Waiting for next change."
done

编写服务配置到文件/etc/systemd/system/k8s-log-sync.service

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
[Unit]
Description=Real-time Log Sync Service for K8s Pods
Documentation=https://your-documentation-link-here # (Optional)
After=network.target

[Service]
# User that will run the script. Use root if you need to access /var/log.
User=root
# Group for the process.
Group=root

# Path to the executable script.
ExecStart=/usr/local/bin/sync_script.sh

# Restart policy: always restart the service if it fails.
Restart=always
# Time to wait before restarting the service.
RestartSec=10

# Standard output and error logging configuration.
# We are redirecting output to a file within the script,
# but systemd can also handle it via journald.
StandardOutput=journal
StandardError=journal

[Install]
# Defines the target to link this service to when enabling it.
# multi-user.target is a standard target for services that should start at boot.
WantedBy=multi-user.target

应用

1
2
3
4
sudo chmod +x /usr/local/bin/sync_script.sh
sudo systemctl daemon-reload
sudo systemctl start k8s-log-sync.service
sudo systemctl enable k8s-log-sync.service

可以使用华为云obs做备份

首先安装s3fs

1
2
3
apt-get install -y s3fs
echo "YOUR_AK:YOUR_SK" | sudo tee /etc/passwd-s3fs
sudo chmod 600 /etc/passwd-s3fs

挂载

1
2
3
4
5
6
7
# --- 请将下面的占位符替换为您的实际信息 ---
# <your-bucket-name>: 您的 OBS 桶名称
# <your-mount-point>: 您创建的本地挂载点
# <obs-endpoint>: 您的桶所在区域的 Endpoint

sudo s3fs <your-bucket-name> <your-mount-point> -o url=https://<obs-endpoint> -o use_path_request_style
s3fs hskj-aimage-log-backups /opt/k8s-log -o url=https://obs.cn-south-1.myhuaweicloud.com -o use_path_request_style

添加开机自动挂载：

在文件/etc/fstab末尾添加：

1
s3fs#hskj-aimage-log-backups /opt/k8s-log fuse _netdev,allow_other,url=https://obs.cn-south-1.myhuaweicloud.com,use_path_request_style 0 0

也可以安装fluent-bit（ubuntu20版本过低无法使用）

1
curl https://raw.githubusercontent.com/fluent/fluent-bit/master/install.sh | sh

5. 参考

实践：Kubernetes环境中Etcd集群的备份与恢复-腾讯云开发者社区-腾讯云

linux下上传华为云OBS—obsutil使用指南_obsutil命令参数-CSDN博客

以华为云OBS为例，记录s3fs挂载对象存储的避坑指南 - 维简网

1. 使用 fluent-bit 采集文件 - 程序阳YY - 博客园

Fluent-bit 配置参考 | ClickVisual

Fluent Bit配置与使用——基于版本V2.2.2-CSDN博客

轻量级日志收集转发 | fluent-bit配置详解（二）-腾讯云开发者社区-腾讯云

pod日志采集（Fluent Bit方案）-崔亮的博客

Ubuntu | Fluent Bit: Official Manual

Install Fluent Bit on Ubuntu 24.04 | Lindevs

如何在Debian使用inotifywait - 问答 - 亿速云

  • Title: k8s备份
  • Author: Ethereal
  • Created at: 2025-03-26 16:09:53
  • Updated at: 2025-07-04 23:45:34
  • Link: https://ethereal-o.github.io/2025/03/26/k8s备份/
  • License: This work is licensed under CC BY-NC-SA 4.0.
 Comments
On this page
k8s备份
  1. 1. 安装etcd客户端
  2. 2. 创建备份
  3. 3. 测试备份是否成功
  4. 4. 恢复
  5. 5. pod日志备份
    1. 5.1 日志目录
    2. 5.2 方案一：周期性
    3. 5.3 方案二：实时备份
  6. 5. 参考