Consul 服务注册与发现 DNS SRV 解析 python 发现服务示例

安装

官方: https://developer.hashicorp.com/consul/install
参考: https://gitcode.csdn.net/65ed701d1a836825ed798f7a.html?dp_token=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpZCI6NjYxNjIyLCJleHAiOjE3MTE0NDgzMjIsImlhdCI6MTcxMDg0MzUyMiwidXNlcm5hbWUiOiJxcV80NDg3Njg5NCJ9.ZLQ5xpdAILQgxM21FUCcRB0LojuAfU8J_xdZRmZ72wY

简单启动

wget https://releases.hashicorp.com/consul/1.18.0/consul_1.18.0_linux_amd64.zip
unzip consul_1.18.0_linux_amd64.zip
sudo mv consul /usr/bin/
# 启动
consul agent -dev -bind=10.0.0.1 -http-port=8500

分布式方式启动

# 启动3个服务
consul agent -server -bootstrap-expect=3 -node=sv-v -data-dir=/tmp/consul -bind=10.0.0.1 -client=0.0.0.0 -datacenter=dc -ui

consul agent -server -bootstrap-expect=3 -node=sv-v2 -data-dir=/tmp/consul -bind=10.0.0.32 -client=0.0.0.0 -datacenter=dc -ui -retry-join-wan=10.0.0.1
consul agent -server -bootstrap-expect=3 -node=ps1 -data-dir=/tmp/consul -bind=10.0.0.2 -client=0.0.0.0 -datacenter=dc -ui -retry-join=10.0.0.1

启用分布式 agent server 和 client 监听地址 0.0.0.0,使用 consul members 命令行访问 consul ,登录 ui 页面是 bind 地址加上 8500 ,由于是分布式启动任何一个掉线,依然能从别的地址上访问到 consul。

生产部署

https://developer.hashicorp.com/consul/tutorials/production-deploy/deployment-guide

从配置文件启动

命令行启动: https://developer.hashicorp.com/consul/docs/agent/config/cli-flags#_config_file

consul agent -config-file=/home/mrh/Sync/sv-v-config/consul/sv-v-server.json

有关 json 配置文件的介绍:
https://developer.hashicorp.com/consul/docs/agent/config/config-files

配置文件示例:

{
    "datacenter": "dc",
    "data_dir": "/home/mrh/Sync/sv-v-config/consul/data",
    "log_level": "INFO",
    "node_name": "sv-v",
    "server": true,
    "bootstrap_expect": 3,
    "bind_addr": "10.0.0.1",
    "ui_config": {
      "enabled": true
    },
    "client_addr":["10.0.0.1", "127.0.0.1"],
    "connect": {
        "enabled": true
    },
    "acl":{
        "enabled": true,
        "default_policy": "allow",
        "enable_token_persistence":true
    },
    "retry_join": ["sv-v2", "ps1", "sv-v"],
    "watches": [
      {
        "type": "checks",
        "handler": ""
      }
    ]
  }
  

配置了 client_addr 之后就必须使用该地址访问 ui 页面,或者 http 请求。例如

# 正确
consul members -http-addr=http://10.0.0.1:8500
# 错误,因为它使用了默认 http://127.0.0.1:8500 地址请求,由于 sv-v-server.json 已经配置 "client_addr":"10.0.0.1" ,因此默认地址无法访问
consul members 

ACL 秘钥访问控制

在这条consul agent -config-file=/home/mrh/Sync/sv-v-config/consul/sv-v-server.json启动指令中,配置文件 sv-v-server.json 启动 ACL 的配置项:

    "acl":{
        "enabled": true,
        "default_policy": "allow",
        "enable_token_persistence":true
    },

但是目前来说, “client_addr”:“10.0.0.1” 设置为 10 私有网段,实测只有加入这个网段的服务器才可以相互访问。用 192.168.2.31 也访问不了。

"default_policy": "allow" 是为了能够创建一个管理员秘钥,创建完管理员秘钥后,后续可以将它改为 "default_policy": "deny" ,以后用管理员 token 来进行后续操作。

生成 ACL 秘钥

# 生成一个令牌,默认的令牌是不受限制的管理令牌
consul acl bootstrap -http-addr=http://10.0.0.1:8500 
AccessorID:       229b0b4e-748e-0859-48a0-b3b89bce4a5f
SecretID:         b69317df-62be-59a3-9d98-21743514ed9d
Description:      Bootstrap Token (Global Management)
Local:            false
Create Time:      2024-03-20 19:53:46.124615905 +0800 CST
Policies:
   00000000-0000-0000-0000-000000000001 - global-management

http://10.0.0.1:8500/ui/dc/acls/policies 查看秘钥所属读写策略

由于默认的管理秘钥具有非常多的权限,为了避免应用耦合,点击网页 http://10.0.0.1:8500 Policy - 右上角 - create - 新增一个仅用于注册和服务发现的策略:

acl = "write"
service_prefix "" {
	policy = "write"
	intentions = "write"
}
key_prefix "" {
	policy = "write"
}

然后新建另一个 token 秘钥,将 Policy 应用到这个秘钥中:
http://10.0.0.1:8500/ui/dc/acls/tokens

注册服务

定义 service.json 配置文件
https://developer.hashicorp.com/consul/docs/services/usage/define-services#define-multiple-services-in-a-single-file

https://developer.hashicorp.com/consul/commands/services/register

# 注册服务
consul services register -token=a657c158-d4ac-459d-cd7c-8379a2a7e0ae -http-addr=http://10.0.0.1:8500 /home/mrh/Sync/sv-v-config/consul/config/service.json
# 移除服务
consul services deregister -token=a657c158-d4ac-459d-cd7c-8379a2a7e0ae -http-addr=http://10.0.0.1:8500 /home/mrh/Sync/sv-v-config/consul/config/service.json

健康检查

http
https://developer.hashicorp.com/consul/docs/services/usage/checks#http-checks

{
    "services": [
        {
            "id": "video-get-pc",
            "name": "video-get",
            "address": "10.0.0.12",
            "port": 9082,
            "check": {
                "http": "http://10.0.0.12:9082/",
                "method": "GET",
                "interval": "25s",
                "timeout": "1s"
            }
        },
        {
            "id": "video-get-pc2",
            "name": "video-get",
            "address": "10.0.0.24",
            "port": 9082,
            "check": {
                "http": "http://10.0.0.24:9082/",
                "method": "GET",
                "interval": "25s",
                "timeout": "1s"
            }
        }
]
}
# 注册服务
consul services register -token=a657c158-d4ac-459d-cd7c-8379a2a7e0ae -http-addr=http://10.0.0.1:8500 /home/mrh/Sync/sv-v-config/consul/config/service.json
# 注销指定服务
consul services deregister -token=a657c158-d4ac-459d-cd7c-8379a2a7e0ae -http-addr=http://10.0.0.1:8500 -id=vector-sv-v2

请注意,从上述指令我在 10.0.0.1 服务器注册了两个服务,分别是 10.0.0.12 和 10.0.0.24,这两个服务并不是在 10.0.0.1 主机上(此处为了方便而简单配置)。在 consul 正确实践中应当注册的服务在同一台主机上,否则会有歧义。

也就是说,只能在 consul 主机上,只能注册主机自己的服务,如果你有别的 consul 集群主机 10.0.0.2 ,则应该注册自己的服务如 10.0.0.2:9000,10.0.0.1:9002 。

gRPC

{
    "services": [
        {
            "id": "vector-sv-v2",
            "name": "vector",
            "check": {
                "grpc": "10.0.0.32:18600",
                "interval": "25s",
                "timeout": "1s"
            }
        }
]
}

proto: https://github.com/grpc/grpc/blob/master/doc/health-checking.md
python server

class HealthService(HealthBase):
    async def Check(self, stream):
        request:vector_pb2.SaveDocToVectorRequest = await stream.recv_message()
        try:
            engine = create_engine(DATABASE_URL)  
            connection = engine.connect()  
            # 如果能成功连接到数据库,那么数据库就是健康的  
            connection.close()
            responses = health_pb2.HealthCheckResponse(status=health_pb2.HealthCheckResponse.SERVING)
        except psycopg2.Error as e:
            responses = health_pb2.HealthCheckResponse(status=health_pb2.HealthCheckResponse.NOT_SERVING)
            logger.error(f"Database is not available: {str(e)}")
        await stream.send_message(responses) 
async def main():
    server = Server([VectorService(),HealthService()])
    host = "0.0.0.0"
    port = 18600
    with graceful_exit([server]):
        await server.start(host, port)
        logger.info(f'Serving on {host}:{port}')
        await server.wait_closed()    

if __name__ == '__main__':
    asyncio.run(main())

服务方式运行

https://developer.hashicorp.com/consul/tutorials/production-deploy/deployment-guide#configure-the-consul-process

ln /home/mrh/Sync/sv-v-config/consul/consul.service /etc/systemd/system/consul.service
sudo systemctl enable consul
sudo systemctl start consul
sudo systemctl status consul


sudo journalctl -u consul.service --since "1 hour ago"

服务发现

https://developer.hashicorp.com/consul/docs/concepts/service-discovery
服务的消费者通过 DNS 与服务通信。与 SRV 不同的是,consul 是一个中介服务,可以直接将请求路由到目标服务,而不用DNS获得实际IP然后才主动访问IP和端口。

https://developer.hashicorp.com/consul/docs/services/discovery/dns-overview
默认情况下,consul 服务发现使用 DNS 解析。

创建DNS策略

参考 https://developer.hashicorp.com/consul/tutorials/security/access-control-setup-production#token-for-dns

可以在命令行创建,也可以在 ui 页面创建。最终得到 dns token 2782c566-2efd-9b1b-e9fc-e3727767e85a

export CONSUL_HTTP_TOKEN=b69317df-62be-59a3-9d98-21743514ed9d
# 设置 DNS 默认令牌,需要使用具有读写的 token 值环境变量
consul acl set-agent-token default 2782c566-2efd-9b1b-e9fc-e3727767e85a
# 查看 DNS SRV 记录。
dig @127.0.0.1 -p 8600 video-get.service.consul SRV

# 返回
# 注意,如果 SRV 没有返回 9082 端口和 10.0.0.24 地址,说明注册 service 的配置文件中没有定义 address 和 port 字段
# 例如注册服务时, consul services register ./config/video-get.json  ,在 json 文件中,必须定义 "port": 9082, 和 "address": "10.0.0.12", 字段
;; ANSWER SECTION:
video-get.service.consul. 0     IN      SRV     1 1 9082 0a000018.addr.dc.consul.
video-get.service.consul. 0     IN      SRV     1 1 9082 0a00000c.addr.dc.consul.

;; ADDITIONAL SECTION:
0a000018.addr.dc.consul. 0      IN      A       10.0.0.24
sv-v.node.dc.consul.    0       IN      TXT     "consul-version=1.18.0"
0a00000c.addr.dc.consul. 0      IN      A       10.0.0.12
sv-v.node.dc.consul.    0       IN      TXT     "consul-version=1.18.0"

python 服务发现示例

import random
import re
import socket
import struct
from dns import resolver

# 配置自定义 DNS 服务器
custom_resolver = resolver.Resolver()
custom_resolver.nameservers = ['10.0.0.1', '10.0.0.32', '10.0.0.2']
# 设置端口号(这里假设所有服务器都使用 8600 端口)
for server in custom_resolver.nameservers:
    custom_resolver.port = 8600  # 注意:通常不需要对每个服务器单独设置,除非它们使用不同端口


def extract_ip_prefix(encoded_address:str):
    match = encoded_address.split(".")[0]
    return  match

def decode_address(encoded_address):
    hex_address = extract_ip_prefix(encoded_address)
    binary_address = bytes.fromhex(hex_address)
    return socket.inet_ntoa(binary_address)    

def srv_to_address_random_choice(srv_record_name):
    try:
        answer = custom_resolver.resolve(srv_record_name, 'SRV')
        selected_rdata = random.choice(answer)
        ip_address = decode_address(selected_rdata.target.to_text())
        print(f"Service: {srv_record_name} - {ip_address}:{selected_rdata.port} ")
        return ip_address,selected_rdata.port
    except resolver.NoAnswer as e:
        print(f"No answer found for the query: {srv_record_name}")
    except resolver.NXDOMAIN as e:
        print(f"Domain not found: {srv_record_name}")
    except Exception as e:
        print(f"Error occurred while resolving DNS: {str(e)}")
        
def main():
    # 查询 SRV 记录
    srv_record_name = 'video-get.service.consul'
    # srv_record_name = 'consul.service.consul'
    srv_to_address_random_choice(srv_record_name)

if __name__ == "__main__":
    main()

负载均衡

在官方文档中只有 Nginx 的负载均衡实现:https://developer.hashicorp.com/consul/tutorials/load-balancing/load-balancing-nginx

使用 Caddy 实现 SRV 发现和负载均衡: https://caddyserver.com/docs/caddyfile/directives/reverse_proxy#dynamic-upstreams

  • 29
    点赞
  • 15
    收藏
    觉得还不错? 一键收藏
  • 0
    评论

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值