我在 Ubuntu 18.04 上使用 Kubespray 构建了一个 Kubernetes 集群,并且面临 DNS 问题,因此基本上容器无法通过它们的主机名进行通信。
有效的事情:
- 通过 IP 地址进行容器通信
- 互联网正在从容器中工作
- 能够解决
kubernetes.default
Kubernetes 主控:
root@k8s-1:~# cat /etc/resolv.conf | grep -v ^\\#
nameserver 127.0.0.53
search home
root@k8s-1:~#
荚:
root@k8s-1:~# kubectl exec dnsutils cat /etc/resolv.conf
nameserver 169.254.25.10
search default.svc.cluster.local svc.cluster.local cluster.local home
options ndots:5
root@k8s-1:~#
CoreDNS pod 是健康的:
root@k8s-1:~# kubectl get pods --namespace=kube-system -l k8s-app=kube-dns
NAME READY STATUS RESTARTS AGE
coredns-58687784f9-8rmlw 1/1 Running 0 35m
coredns-58687784f9-hp8hp 1/1 Running 0 35m
root@k8s-1:~#
CoreDNS pod 的日志:
root@k8s-1:~# kubectl describe pods --namespace=kube-system -l k8s-app=kube-dns | tail -n 2
Normal Started 35m kubelet, k8s-2 Started container coredns
Warning DNSConfigForming 12s (x33 over 35m) kubelet, k8s-2 Nameserver limits were exceeded, some nameservers have been omitted, the applied nameserver line is: 4.2.2.1 4.2.2.2 208.67.220.220
root@k8s-1:~# kubectl logs --namespace=kube-system coredns-58687784f9-8rmlw
.:53
2020-02-09T22:56:14.390Z [INFO] plugin/reload: Running configuration MD5 = b9d55fc86b311e1d1a0507440727efd2
2020-02-09T22:56:14.391Z [INFO] CoreDNS-1.6.0
2020-02-09T22:56:14.391Z [INFO] linux/amd64, go1.12.7, 0a218d3
CoreDNS-1.6.0
linux/amd64, go1.12.7, 0a218d3
root@k8s-1:~#
root@k8s-1:~# kubectl logs --namespace=kube-system coredns-58687784f9-hp8hp
.:53
2020-02-09T22:56:20.388Z [INFO] plugin/reload: Running configuration MD5 = b9d55fc86b311e1d1a0507440727efd2
2020-02-09T22:56:20.388Z [INFO] CoreDNS-1.6.0
2020-02-09T22:56:20.388Z [INFO] linux/amd64, go1.12.7, 0a218d3
CoreDNS-1.6.0
linux/amd64, go1.12.7, 0a218d3
root@k8s-1:~#
CoreDNS 似乎暴露了:
root@k8s-1:~# kubectl get svc --namespace=kube-system | grep coredns
coredns ClusterIP 10.233.0.3 <none> 53/UDP,53/TCP,9153/TCP 37m
root@k8s-1:~#
root@k8s-1:~# kubectl get ep coredns --namespace=kube-system
NAME ENDPOINTS AGE
coredns 10.233.64.2:53,10.233.65.3:53,10.233.64.2:53 + 3 more... 37m
root@k8s-1:~#
这些是我有问题的 pod - 所有集群都因为这个问题而受到影响:
root@k8s-1:~# kubectl get pods -o wide -n default
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
busybox 1/1 Running 0 17m 10.233.66.7 k8s-3 <none> <none>
dnsutils 1/1 Running 0 50m 10.233.66.5 k8s-3 <none> <none>
nginx-86c57db685-p8zhc 1/1 Running 0 43m 10.233.64.3 k8s-1 <none> <none>
nginx-86c57db685-st7rw 1/1 Running 0 47m 10.233.66.6 k8s-3 <none> <none>
root@k8s-1:~#
能够通过 IP 地址使用 DNS 和容器访问互联网:
root@k8s-1:~# kubectl exec -it nginx-86c57db685-st7rw -- sh -c "ping 10.233.64.3"
PING 10.233.64.3 (10.233.64.3) 56(84) bytes of data.
64 bytes from 10.233.64.3: icmp_seq=1 ttl=62 time=0.481 ms
64 bytes from 10.233.64.3: icmp_seq=2 ttl=62 time=0.551 ms
...
root@k8s-1:~# kubectl exec -it nginx-86c57db685-st7rw -- sh -c "ping google.com"
PING google.com (172.217.21.174) 56(84) bytes of data.
64 bytes from fra07s64-in-f174.1e100.net (172.217.21.174): icmp_seq=1 ttl=61 time=77.9 ms
...
root@k8s-1:~# kubectl exec -it nginx-86c57db685-st7rw -- sh -c "ping kubernetes.default"
PING kubernetes.default.svc.cluster.local (10.233.0.1) 56(84) bytes of data.
64 bytes from kubernetes.default.svc.cluster.local (10.233.0.1): icmp_seq=1 ttl=64 time=0.030 ms
64 bytes from kubernetes.default.svc.cluster.local (10.233.0.1): icmp_seq=2 ttl=64 time=0.069 ms
...
实际问题:
root@k8s-1:~# kubectl exec -it nginx-86c57db685-st7rw -- sh -c "ping nginx-86c57db685-p8zhc"
ping: nginx-86c57db685-p8zhc: Name or service not known
command terminated with exit code 2
root@k8s-1:~#
root@k8s-1:~# kubectl exec -it nginx-86c57db685-st7rw -- sh -c "ping dnsutils"
ping: dnsutils: Name or service not known
command terminated with exit code 2
root@k8s-1:~#
oot@k8s-1:~# kubectl exec -ti busybox -- nslookup nginx-86c57db685-p8zhc
Server: 169.254.25.10
Address: 169.254.25.10:53
** server can't find nginx-86c57db685-p8zhc.default.svc.cluster.local: NXDOMAIN
*** Can't find nginx-86c57db685-p8zhc.svc.cluster.local: No answer
*** Can't find nginx-86c57db685-p8zhc.cluster.local: No answer
*** Can't find nginx-86c57db685-p8zhc.home: No answer
*** Can't find nginx-86c57db685-p8zhc.default.svc.cluster.local: No answer
*** Can't find nginx-86c57db685-p8zhc.svc.cluster.local: No answer
*** Can't find nginx-86c57db685-p8zhc.cluster.local: No answer
*** Can't find nginx-86c57db685-p8zhc.home: No answer
command terminated with exit code 1
root@k8s-1:~#
我是否遗漏了什么或如何使用主机名修复容器之间的通信?
非常感谢
更新
更多检查:
root@k8s-1:~# kubectl exec -ti dnsutils -- nslookup kubernetes.default
Server: 169.254.25.10
Address: 169.254.25.10#53
Name: kubernetes.default.svc.cluster.local
Address: 10.233.0.1
我创建了 StatefulSet:
kubectl apply -f https://raw.githubusercontent.com/kubernetes/website/master/content/en/examples/application/web/web.yaml
我可以 ping 服务“nginx”:
root@k8s-1:~/kplay# k exec dnsutils -it nslookup nginx
Server: 169.254.25.10
Address: 169.254.25.10#53
Name: nginx.default.svc.cluster.local
Address: 10.233.66.8
Name: nginx.default.svc.cluster.local
Address: 10.233.64.3
Name: nginx.default.svc.cluster.local
Address: 10.233.65.5
Name: nginx.default.svc.cluster.local
Address: 10.233.66.6
使用 FQDN 时还可以联系 statefulset 成员
root@k8s-1:~/kplay# k exec dnsutils -it nslookup web-0.nginx.default.svc.cluster.local
Server: 169.254.25.10
Address: 169.254.25.10#53
Name: web-0.nginx.default.svc.cluster.local
Address: 10.233.65.5
root@k8s-1:~/kplay# k exec dnsutils -it nslookup web-1.nginx.default.svc.cluster.local
Server: 169.254.25.10
Address: 169.254.25.10#53
Name: web-1.nginx.default.svc.cluster.local
Address: 10.233.66.8
但不只使用主机名:
root@k8s-1:~/kplay# k exec dnsutils -it nslookup web-0
Server: 169.254.25.10
Address: 169.254.25.10#53
** server can't find web-0: NXDOMAIN
command terminated with exit code 1
root@k8s-1:~/kplay# k exec dnsutils -it nslookup web-1
Server: 169.254.25.10
Address: 169.254.25.10#53
** server can't find web-1: NXDOMAIN
command terminated with exit code 1
root@k8s-1:~/kplay#
它们都生活在同一个命名空间中:
root@k8s-1:~/kplay# k get pods -n default
NAME READY STATUS RESTARTS AGE
busybox 1/1 Running 22 22h
dnsutils 1/1 Running 22 22h
nginx-86c57db685-p8zhc 1/1 Running 0 22h
nginx-86c57db685-st7rw 1/1 Running 0 22h
web-0 1/1 Running 0 11m
web-1 1/1 Running 0 10m
另一个确认我能够 ping 服务的测试:
kubectl create deployment --image nginx some-nginx
kubectl scale deployment --replicas 2 some-nginx
kubectl expose deployment some-nginx --port=12345 --type=NodePort
root@k8s-1:~/kplay# k exec dnsutils -it nslookup some-nginx
Server: 169.254.25.10
Address: 169.254.25.10#53
Name: some-nginx.default.svc.cluster.local
Address: 10.233.63.137
最后的想法
有趣的事实,但也许这就是 Kubernetes 应该如何工作的?如果想单独访问某个 pod,我可以访问服务主机名和 statefulset 成员。至少在我的 k8s 使用中(可能适用于每个人),如果不是 statefulset,则到达单个 pod 似乎不是很重要。
我建议您遵循这一点,以便我们可以隔离您的 CoreDNS 中可能存在的问题,并且您可以看到它工作正常。
可以使用 DNS 记录访问 pod,但正如您所说,这在常规 K8s 实现中并不是很重要。
例子
是的,您的 CoreDNS 正在按预期工作,并且您描述的一切都是预期的。