二进制部署k8s教程10 - 部署apiserver高可用

2023-03-08 21:12:45 浏览数 (1)

!TIP 二进制部署 k8s - 部署 apiserver 高可用


转载请注明出处:https://janrs.com/clw9 有任何问题欢迎在底部评论区发言。

!WARNING 如果该环节在阿里云 ecs 部署,因为阿里云现在已经不支持 vip 了。 HA 部署环节可以在本地测试。或者线上不使用 keepalived 做高可用。 直接用 nginxtcp 反向代理也是可以。 主要要注意学习的地方就是:kube-apiserverserver 证书地址需要把 nginx 的服务器 ip 写进 hosts 参数。 否则会报无权限错误。

部署 HA 高可用

1.k8s 的 HA

!NOTE k8smaster 的三大组件,其中 kube-controller-managerkube-scheduler 有提供高可用机制。 kube-apiserver 没有,需要自己实现。

1-1.kube-controlle 与 kube-scheduler 的 HA

master 中的 kube-apiserver , kube-controller-manager , kube-scheduler 这三大组件中。

kube-controller-managerkube-scheduler 有自带的 leader 选举机制。

通过部署的时候设置参数:--leader-elect=true 启动 leader 自选举。

在部署完三台 master 服务器,可以在任意一台 master 输入以下命令查看 leader

代码语言:shell复制
kubectl get leases -n kube-system

显示如下

可以看出,kube-controller-managerkube-schedulerleader 是在 master-01 服务器上

代码语言:text复制
NAME                      HOLDER                                              AGE
kube-controller-manager   k8s-master01_e0f4cfd5-1190-4f79-9ee5-a2063eb3ca16   156m
kube-scheduler            k8s-master01_54254610-53a8-4c3a-b3ea-a4fa5f549119   99s

在三台 master 任意一台停止这个三个组件,然后再次查看。

master-01 服务器停止三个组件的服务

代码语言:shell复制
systemctl stop kube-scheduler && 
systemctl stop kube-controller-manager && 
systemctl stop kube-apiserver

master-02 服务器查看 leader 信息

代码语言:shell复制
kubectl get leases -n kube-system

显示如下

可以看出,leaderHOLDER 服务器已经改变了

代码语言:text复制
NAME                      HOLDER                                              AGE
kube-controller-manager   k8s-master02_e0f4cfd5-1190-4f79-9ee5-a2063eb3ca16   3h5m
kube-scheduler            k8s-master03_e350060d-68ad-4f59-82a8-456f835b7f3d   30m

1-2.kube-apiserver 的 HA

k8s 没有提供 kube-apiserverHA,需要手动实现。

因为 kube-apiserver 是无状态的应用,并且对外提供 http/https 的接口调用方式提供服务,所以可以用 nginx 来做负载均衡达到高可用。

2.初始化系统环境

!NOTE 不需要像 master 节点和 node 节点那样初始化。

2-1.安装 epel 以及依赖软件

代码语言:shell复制
dnf install epel-release vim iptables jq ipvsadm ipset curl net-tools rsyslog -y

2-2.关闭防火墙

代码语言:shell复制
systemctl stop firewalld && systemctl disable firewalld

2-3.清空 iptables 规则链

代码语言:shell复制
iptables -F && iptables -X && iptables -F -t nat && iptables -X -t nat

2-4.关闭 swap 分区

代码语言:shell复制
swapoff -a && 
sed -i '/ swap / s/^(.*)$/#1/g' /etc/fstab

2-5.关闭 selinux

代码语言:shell复制
setenforce 0 && 
sed -i 's/^SELINUX=.*/SELINUX=disabled/' /etc/selinux/config

2-6.设置时间同步

设置时区

代码语言:shell复制
timedatectl set-timezone Asia/Shanghai

同步时区

代码语言:shell复制
systemctl enable chronyd && 
systemctl start chronyd

查看

代码语言:shell复制
timedatectl status

显示

代码语言:text复制
               Local time: Sun 2022-10-02 13:30:23 CST
           Universal time: Sun 2022-10-02 05:30:23 UTC
                 RTC time: Sun 2022-10-02 05:30:23
                Time zone: Asia/Shanghai (CST,  0800)
System clock synchronized: yes
              NTP service: active
          RTC in local TZ: no

将当前的 UTC 时间写入硬件时钟

代码语言:shell复制
timedatectl set-local-rtc 0

重启依赖于系统时间的服务

代码语言:shell复制
systemctl restart rsyslog && 
systemctl restart crond

2-7.设置 systemd journald

创建持久化保存日志的目录以及添加配置并生效

代码语言:shell复制
mkdir /var/log/journal && 
mkdir /etc/systemd/journald.conf.d && 
cat > /etc/systemd/journald.conf.d/99-prophet.conf <<EOF
[Journal]
# 持久化保存到磁盘
Storage=persistent
# 压缩历史日志
Compress=yes
SyncIntervalSec=5m
RateLimitInterval=30s
RateLimitBurst=1000
# 最大占用空间 10G
SystemMaxUse=10G
# 单日志文件最大 200M
SystemMaxFileSize=200M
# 日志保存时间 2 周
MaxRetentionSec=2week
# 不将日志转发到 syslog
ForwardToSyslog=no
EOF
systemctl restart systemd-journald

3.查看网卡是否开启多播

查看

代码语言:shell复制
ip a

有显示 MULTICAST 即表示有打开网卡多播

代码语言:text复制
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:e6:09:4c brd ff:ff:ff:ff:ff:ff
    inet 172.16.222.201/24 brd 172.16.222.255 scope global noprefixroute ens160
       valid_lft forever preferred_lft forever
    inet 172.16.222.110/24 scope global secondary ens160
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fee6:94c/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

如果没有开启,执行以下命令打开

ens160 为网卡名称

代码语言:shell复制
ip link set multicast on dev ens160

4.重启

代码语言:shell复制
shutdown -r now

5.安装

两台 HA 服务器都安装

代码语言:shell复制
dnf install keepalived nginx -y

6.创建配置文件

6-1.创建 nginx.conf 配置文件

!NOTE 在两台 HA 服务器都创建。 这里使用的是 8443 端口代理 kube-apiserver6443 端口。

代码语言:shell复制
#备份nginx.conf
mv /etc/nginx/nginx.conf /etc/nginx/nginx.conf.bak && 

#新建nginx.conf
cat > /etc/nginx/nginx.conf << 'EOF'
user nginx;
worker_processes auto;
error_log /var/log/nginx/error.log;
pid /run/nginx.pid;

include /usr/share/nginx/modules/*.conf;

events {
    worker_connections 1024;
    use epoll;
}

# 四层负载均衡,为三台Master apiserver组件提供负载均衡
stream {

    log_format  main  '$remote_addr $upstream_addr - [$time_local] $status $upstream_bytes_sent';

    access_log  /var/log/nginx/k8s-access.log  main;

    upstream k8s-apiserver {
       server 172.16.222.121:6443;   # Master1 APISERVER IP:PORT
       server 172.16.222.122:6443;   # Master2 APISERVER IP:PORT    
       server 172.16.222.123:6443;   # Master3 APISERVER IP:PORT    
    }
    
    server {
       listen 8443;
       proxy_pass k8s-apiserver;
    }
}

http {
    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;

    sendfile            on;
    tcp_nopush          on;
    tcp_nodelay         on;
    keepalive_timeout   65;
    types_hash_max_size 2048;

    include             /etc/nginx/mime.types;
    default_type        application/octet-stream;

    server {
        listen       80 default_server;
        server_name  _;

        location / {
        }
    }
}
EOF

6-2.创建 keepalived 配置文件

!NOTE 其中: k8s-ha-master 作为 nginx master 主负载均衡服务器, k8s-ha-backup 作为 nginx slave 备用负载均衡服务器。

6-2-1.在 k8s-ha-master 服务器创建

创建 keepalived.conf

代码语言:shell复制
#备份
mv /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf.bak && 
cat > /etc/keepalived/keepalived.conf << EOF
global_defs { 

   notification_email { 

     acassen@firewall.loc 
     failover@firewall.loc 
     sysadmin@firewall.loc 

   } 

   script_user root
   
   enable_script_security 
   
   notification_email_from Alexandre.Cassen@firewall.loc  

   smtp_server 127.0.0.1 

   smtp_connect_timeout 30 

   router_id K8S_HA_NGINX_MASTER
} 

vrrp_script nginx_heartbeat {
    script "/etc/keepalived/nginx_heartbeat.sh"
}

vrrp_instance VI_1 { 

    # 定义初始状态,这里是 MASTER 备用的 nginx 设置为 BACKUP
    state MASTER

    # 工作接口,通告选举使用哪个接口进行
    interface ens160

    # VRRP 路由 ID实例,每个实例是唯一的 
    # 虚拟路由ID,如果是一组虚拟路由就定义一个ID,如果是多组就要定义多个,而且这个虚拟
    # ID还是虚拟MAC最后一段地址的信息,取值范围0-255
    virtual_router_id 51

    # MASTER 的优先级最高,BACKUP 的优先级低一点,下面设置为 90
    priority 100

    # 通告频率,单位为秒
    advert_int 1

    # 通信认证机制,这里是明文认证还有一种是加密认证
    authentication { 
        auth_type PASS      
        auth_pass 123456
    }  
    # 虚拟 ip 地址。即 vip
    virtual_ipaddress { 
        172.16.222.110/24
    } 
    track_script {
        nginx_heartbeat
    } 
}
EOF

创建 nginx_heartbeat.sh 脚本

代码语言:shell复制
cat > /etc/keepalived/nginx_heartbeat.sh << 'EOF'
#!/bin/bash
count=$(ps -ef |grep nginx |egrep -cv "grep|$$")
if [ "$count" -eq 0 ];
then
    exit 1
else
    exit 0
fi
EOF
chmod  x /etc/keepalived/nginx_heartbeat.sh

6-3.在 k8s-ha-backup 服务器创建

创建 keepalived.conf

代码语言:shell复制
mv /etc/keepalived/keepalived.conf /etc/keepalived/keepalived.conf.bak && 
cat > /etc/keepalived/keepalived.conf << EOF
global_defs { 

   notification_email { 

     acassen@firewall.loc 
     failover@firewall.loc 
     sysadmin@firewall.loc 

   } 

   script_user root
   
   notification_email_from Alexandre.Cassen@firewall.loc  

   smtp_server 127.0.0.1 

   smtp_connect_timeout 30 

   router_id K8S_HA_NGINX_SLAVE
} 

vrrp_script nginx_heartbeat {
    script "/etc/keepalived/nginx_heartbeat.sh"}

vrrp_instance VI_1 { 

    # 定义初始状态,可以是MASTER或者BACKUP
    # 这里是 nginx 备用,设置为 BACKUP
    state BACKUP

    # 工作接口,通告选举使用哪个接口进行
    interface ens160

    # VRRP 路由 ID实例,每个实例是唯一的 
    # 虚拟路由ID,如果是一组虚拟路由就定义一个ID,如果是多组就要定义多个,而且这个虚拟
    # ID还是虚拟MAC最后一段地址的信息,取值范围0-255
    virtual_router_id 51

    # BACKUP 的优先级低一点
    priority 90

    # 通告频率,单位为秒
    advert_int 1

    # 通信认证机制,这里是明文认证还有一种是加密认证
    authentication { 
        auth_type PASS      
        auth_pass 123456
    }  
    
    # 虚拟 ip。即 vip
    virtual_ipaddress { 
        172.16.222.110/24
    } 
    track_script {
        nginx_heartbeat
    } 
}
EOF

创建 nginx_heartbeat.sh 脚本

代码语言:shell复制
cat > /etc/keepalived/nginx_heartbeat.sh  << 'EOF'
#!/bin/bash
count=$(ps -ef |grep nginx |egrep -cv "grep|$$")
if [ "$count" -eq 0 ];
then
    exit 1
else
    exit 0
fi
EOF
chmod  x /etc/keepalived/nginx_heartbeat.sh

7.启动

启动

代码语言:shell复制
systemctl daemon-reload && 
systemctl start nginx && 
systemctl start keepalived

查看服务状态

查看状态是否都为 Active

代码语言:shell复制
systemctl status nginx && 
systemctl status keepalived

设置开机启动

代码语言:shell复制
systemctl enable nginx && 
systemctl enable keepalived

8.验证

8-1.验证是否可以访问到 kube-apiserver

在任意一台 k8s 服务器访问

代码语言:shell复制
curl -k https://172.16.222.110:8443/version

显示如下表示可以访问到 kube-apiserver

没有详细信息和 401 是因为没有 ssl 验证。此时已经可以访问 kube-apiserver

代码语言:text复制
{
  "kind": "Status",
  "apiVersion": "v1",
  "metadata": {},
  "status": "Failure",
  "message": "Unauthorized",
  "reason": "Unauthorized",
  "code": 401
}

8-2.验证 vip 是否有漂移

在以上两台服务器都启动服务后,测试一下高可用是否正常。

8-2-1.查看 k8s-ha-master 的信息

查看 vip 地址

代码语言:shell复制
ip a

显示如下

可以看到有 vip ip 地址:172.16.222.110

代码语言:text复制
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:e6:09:4c brd ff:ff:ff:ff:ff:ff
    inet 172.16.222.201/24 brd 172.16.222.255 scope global noprefixroute ens160
       valid_lft forever preferred_lft forever
    inet 172.16.222.110/24 scope global secondary ens160
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fee6:94c/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

查看keepalived 状态

代码语言:shell复制
systemctl status keepalived

显示如下

可以看到此时 vip ip 地址 172.16.222.110 在此服务器上

代码语言:text复制
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
   Active: active (running) since Sun 2022-10-02 14:34:04 CST; 14min ago
  Process: 7339 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 7342 (keepalived)
    Tasks: 2 (limit: 23462)
   Memory: 1.9M
   CGroup: /system.slice/keepalived.service
           ├─7342 /usr/sbin/keepalived -D
           └─7343 /usr/sbin/keepalived -D

Oct 02 14:41:51 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:51 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:51 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:51 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: (VI_1) Sending/queueing gratuitous ARPs on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
8-2-2.查看 k8s-ha-backup 的状态
代码语言:shell复制
systemctl status keepalived

显示如下

从最后三条信息中可以看出,BACKUP 的权重比较低

而且 MASTER 有在工作,所以进入 BACKUP 状态,并且移除了 vip 地址

代码语言:text复制
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
   Active: active (running) since Sun 2022-10-02 14:35:52 CST; 21min ago
 Main PID: 5599 (keepalived)
    Tasks: 2 (limit: 23462)
   Memory: 9.7M
   CGroup: /system.slice/keepalived.service
           ├─5599 /usr/sbin/keepalived -D
           └─5600 /usr/sbin/keepalived -D

Oct 02 14:50:01 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: (VI_1) Sending/queueing gratuitous ARPs on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:06 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:55:49 k8s-master-ha-02 Keepalived_vrrp[5600]: (VI_1) Master received advert from 172.16.222.201 with higher priority 100, ours 90
Oct 02 14:55:49 k8s-master-ha-02 Keepalived_vrrp[5600]: (VI_1) Entering BACKUP STATE
Oct 02 14:55:49 k8s-master-ha-02 Keepalived_vrrp[5600]: (VI_1) removing VIPs.
8-2-3.测试转移 vip

停掉 k8s-ha-masternginx

代码语言:shell复制
systemctl stop nginx

再到 k8s-ha-backup 查看的 keepalived 状态

代码语言:shell复制
systemctl status keepalived

显示如下

可以看到通过 nginx_heartbeat.sh 脚本检测不到 nginx 进程

从而停止了 vip 的创建

代码语言:text复制
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:41:56 k8s-master-ha-01 Keepalived_vrrp[7343]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 14:50:00 k8s-master-ha-01 Keepalived_vrrp[7343]: Script `nginx_heartbeat` now returning 1
Oct 02 14:50:00 k8s-master-ha-01 Keepalived_vrrp[7343]: VRRP_Script(nginx_heartbeat) failed (exited with status 1)
Oct 02 14:50:00 k8s-master-ha-01 Keepalived_vrrp[7343]: (VI_1) Entering FAULT STATE
Oct 02 14:50:00 k8s-master-ha-01 Keepalived_vrrp[7343]: (VI_1) sent 0 priority
Oct 02 14:50:00 k8s-master-ha-01 Keepalived_vrrp[7343]: (VI_1) removing VIPs.

再到 k8s-ha-backup 服务器查看 ip 地址

代码语言:shell复制
ip a

显示如下

可以看到在 k8s-ha-backup 服务器上已经有创建 vip 地址

代码语言:text复制
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:2d:ad:07 brd ff:ff:ff:ff:ff:ff
    inet 172.16.222.202/24 brd 172.16.222.255 scope global noprefixroute ens160
       valid_lft forever preferred_lft forever
    inet 172.16.222.110/24 scope global secondary ens160
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe2d:ad07/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

查看 k8s-ha-backupkeepalived 状态

代码语言:shell复制
systemctl status keepalived

显示如下

可以看出此时已经在 k8s-ha-backup 服务器创建了 vip 地址了

代码语言:text复制
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
   Active: active (running) since Sun 2022-10-02 14:35:52 CST; 35min ago
 Main PID: 5599 (keepalived)
    Tasks: 2 (limit: 23462)
   Memory: 9.7M
   CGroup: /system.slice/keepalived.service
           ├─5599 /usr/sbin/keepalived -D
           └─5600 /usr/sbin/keepalived -D

Oct 02 15:11:02 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:02 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:02 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:02 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: (VI_1) Sending/queueing gratuitous ARPs on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110
Oct 02 15:11:07 k8s-master-ha-02 Keepalived_vrrp[5600]: Sending gratuitous ARP on ens160 for 172.16.222.110

到任意一台 k8s 服务器使用 vip 再次访问 kube-apiserver

代码语言:shell复制
curl -k https://172.16.222.110:8443/version

显示如下

可以看到一样能够访问到 kube-apiserver

代码语言:text复制
{
  "kind": "Status",
  "apiVersion": "v1",
  "metadata": {},
  "status": "Failure",
  "message": "Unauthorized",
  "reason": "Unauthorized",
  "code": 401
}

至此。kube-apiserver 高可用已经部署成功。也就是 k8s 高可用部署成功。

转载请注明出处:https://janrs.com/clw9 有任何问题欢迎在底部评论区发言。

0 人点赞