(1). K8S创建Pod时,是否会把机器名称(hostname)和IP地址保存在ETCD?
通过对ETCD的数据(Key)穷举,你会发现:
呵呵呵,机器名称(hostname)和IP地址的映射在ETCD里压根就没有,那也就是说:K8S将这些数据保存在内存里.
# 1. 查看pod
[root@master ~]# kubectl get pods
NAME READY STATUS RESTARTS AGE
busybox-deployment-74c549f577-rwl2w 1/1 Running 0 46m
nginx-6799fc88d8-mx7t6 1/1 Running 0 13m
nginx-6799fc88d8-n4t5j 1/1 Running 0 33m
nginx-6799fc88d8-shrn8 1/1 Running 0 13m
# 2. 查看key,只有一个是我们创建Flannel时创建的.
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem ls /
/coreos.com
# 2.1 查看:/coreos.com/network
# /coreos.com/network/config 是用来配置Flannel的网段.
$ /coreos.com/network/subnets 用来记录每个宿主机的分配的网段信息.
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem ls /coreos.com/network
/coreos.com/network/config
/coreos.com/network/subnets
# 2.2 我有两台宿主机,所以就有两个网段.
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem ls /coreos.com/network/subnets
/coreos.com/network/subnets/172.20.28.0-24
/coreos.com/network/subnets/172.20.97.0-24
# 2.3 查看:/coreos.com/network/subnets/172.20.28.0-24
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem get /coreos.com/network/subnets/172.20.28.0-24
{"PublicIP":"10.211.55.102","BackendType":"vxlan","BackendData":{"VtepMAC":"b6:83:de:ad:68:6d"}}
# 2.4 查看:/coreos.com/network/subnets/172.20.97.0-24
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem get /coreos.com/network/subnets/172.20.97.0-24
{"PublicIP":"10.211.55.101","BackendType":"vxlan","BackendData":{"VtepMAC":"fa:7f:96:a9:e8:bc"}}
# 2.5 查看:/coreos.com/network/config
[root@master ~]# /opt/etcd/bin/etcdctl --ca-file=/opt/etcd/ssl/ca.pem --cert-file=/opt/etcd/ssl/server.pem --key-file=/opt/etcd/ssl/server-key.pem get /coreos.com/network/config
{
"Network": "172.20.0.0/16",
"SubnetMin": "172.20.10.0",
"SubnetMax": "172.20.254.0",
"Backend": {
"Type":"vxlan"
}
}
(2). 域名解析(Service底层实现)
在我搭建DNS解析时,我一直的理解是:我可以通过Pod的名称(hostname),来访问(而非IP)Pod.
所以,造成在搭建DNS上花费了一个多小时后,到最后才理解到:
在K8S创建Service的时候,实际也是建立DNS信息的时刻,换句话说:K8S的Service是建立在DNS之上的,它是对DNS解析的一种封装.从上面的分析我们也看得出来,DNS信息是保存在内存中的,而非ETCD中.
也能理解K8S的这设计,因为:Pod的名称是动态的,求证过程如下:
# 1. 查看pod信息
[root@master ~]# kubectl get pods,svc
NAME READY STATUS RESTARTS AGE
pod/busybox-deployment-74c549f577-rwl2w 0/1 Error 0 107m
pod/nginx-6799fc88d8-mx7t6 0/1 Completed 0 73m
pod/nginx-6799fc88d8-n4t5j 0/1 Completed 0 93m
pod/nginx-6799fc88d8-shrn8 0/1 Completed 0 73m
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/kubernetes ClusterIP 10.0.0.1 <none> 443/TCP 22h
service/nginx NodePort 10.0.0.142 <none> 80:36826/TCP 92m
# 2. 进入容器内部(busybox-deployment-74c549f577-rwl2w)
[root@master ~]# kubectl exec -it busybox-deployment-74c549f577-rwl2w -- sh
# 2.1 尝试ping nginx pod的名称
/ # ping nginx-6799fc88d8-mx7t6
ping: bad address 'nginx-6799fc88d8-mx7t6'
# 2.1 尝试ping nginx pod的名称
/ # ping nginx-6799fc88d8-n4t5j
ping: bad address 'nginx-6799fc88d8-n4t5j'
# 2.1 尝试ping nginx pod的名称
/ # ping nginx-6799fc88d8-shrn8
ping: bad address 'nginx-6799fc88d8-shrn8'
# **************************************************************
# 2.2 尝试ping service的名称
# **************************************************************
/ # ping nginx
PING nginx (10.0.0.142): 56 data bytes
64 bytes from 10.0.0.142: seq=0 ttl=64 time=0.125 ms
# **************************************************************
# 2.3 获取得信息
# **************************************************************
/ # nslookup nginx
Server: 10.0.0.2
Address: 10.0.0.2:53
Non-authoritative answer:
Name: nginx.default.svc.cluster.local
Address: 10.0.0.142
