由灵雀云开源的网络组件 kube-ovn,将已被 openstack 社区采用的成熟网络虚拟化技术 ovs/ovn 引入 kubernetes 平台;为 kubernetes 网络打开了新的大门,令人耳目一新;强烈推荐大家试用该网络组件,反馈建议以帮助项目早日走向成熟。
- 介绍 https://blog.csdn.net/alauda_andy/article/details/88886128
- 项目地址 https://github.com/alauda/kube-ovn
kube-ovn 提供了针对企业应用场景下容器网络实用功能,并为实现更高级的网络管理控制提供了可能性;现有主要功能:
- 1.Namespace 和子网的绑定,以及子网间的访问控制;
- 2.静态IP分配;
- 3.动态QoS;
- 4.分布式和集中式网关;
- 5.内嵌 LoadBalancer;
- 6.Pod IP对外直接暴露
- 7.流量镜像
- 8.IPv6
kube-ovn 的安装十分简单,详见项目的安装文档;基于 kubeasz,以下两步将安装一个集成了 kube-ovn 网络的 k8s 集群;
- 在 ansible hosts 中设置变量
CLUSTER_NETWORK="kube-ovn"
- 执行安装
ansible-playbook 90.setup.yml
或者easzctl setup
kubeasz 项目为kube-ovn
网络生成的 ansible role 如下:
roles/kube-ovn
├── defaults
│ └── main.yml # kube-ovn 相关配置文件
├── tasks
│ └── main.yml # 安装执行文件
└── templates
├── crd.yaml.j2 # crd 模板
├── kube-ovn.yaml.j2 # kube-ovn yaml 模板
└── ovn.yaml.j2 # ovn yaml 模板
安装成功后,可以验证所有 k8s 集群功能正常,查看集群的 pod 网络如下:
$ kubectl get pod --all-namespaces -o wide
NAMESPACE NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
kube-ovn kube-ovn-cni-5php2 1/1 Running 2 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn kube-ovn-cni-7dwmx 1/1 Running 2 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-ovn kube-ovn-cni-lhlvl 1/1 Running 2 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn kube-ovn-controller-57955db7b4-6x6hd 1/1 Running 0 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn kube-ovn-controller-57955db7b4-chvz4 1/1 Running 0 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-ovn ovn-central-bb8747d77-tr5nz 1/1 Running 0 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn ovs-ovn-2qhhr 1/1 Running 0 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn ovs-ovn-np8rn 1/1 Running 0 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn ovs-ovn-pkjw4 1/1 Running 0 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-system coredns-55f46dd959-76qb5 1/1 Running 0 35h 10.16.0.12 192.168.1.42 <none> <none>
kube-system coredns-55f46dd959-wn8kw 1/1 Running 0 35h 10.16.0.11 192.168.1.43 <none> <none>
kube-system heapster-fdb7596d6-xmmrx 1/1 Running 0 35h 10.16.0.15 192.168.1.42 <none> <none>
kube-system kubernetes-dashboard-68ddcc97fc-dwzbf 1/1 Running 0 35h 10.16.0.14 192.168.1.42 <none> <none>
kube-system metrics-server-6c898b5b8b-zvct2 1/1 Running 0 35h 10.16.0.13 192.168.1.43 <none> <none>
直观上 kube-ovn 与传统 k8s 网络(flannel/calico等)比较最大的不同是 pod 子网的分配:
- 传统网络插件下,集群中 pod 一般是不同 node 节点分配不同的子网;然后通过 overlay 等技术打通不同 node 节点的 pod 子网;
- kube-ovn 中 pod 网络根据其所在的 namespace 而定; namespace 在创建时可以根据 annotation 来配置它的子网/网关等参数;默认使用 10.16.0.0/16 的子网;
新建一个 subnet 并绑定 namespace 测试分配一个新的 pod 子网
# 创建一个 namespace: test-ns
$ cat > test-ns.yaml << EOF
apiVersion: v1
kind: Namespace
metadata:
annotations:
name: test-ns
EOF
$ kubectl apply -f test-ns.yaml
# 创建一个 subnet: test-subnet 并绑定 namespace test-ns
$ cat > test-subnet.yaml << EOF
apiVersion: kubeovn.io/v1
kind: Subnet
metadata:
name: test-subnet
spec:
protocol: IPv4
default: false
namespaces:
- test-ns
cidrBlock: 10.17.0.0/24
gateway: 10.17.0.1
excludeIps:
- 10.17.0.1..10.17.0.10
EOF
$ kubectl apply -f test-subnet.yaml
# 在 test-ns 中创建 nginx 部署
$ kubectl run -n test-ns nginx --image=nginx --replicas=2 --port=80 --expose
# 在 default 中创建 busy 客户端
$ kubectl run busy --image=busybox sleep 360000
创建成功后,查看 pod 地址的分配,可以看到确实 test-ns 中 pod 使用新的子网,而 default 中 pod 使用了默认子网,并验证 pod 之间的联通性(默认可通)
$ kubectl get pod --all-namespaces -o wide
NAMESPACE NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
default busy-6c55ccddc5-qrm5j 1/1 Running 0 31h 10.16.0.16 192.168.1.43 <none> <none>
kube-ovn kube-ovn-cni-5php2 1/1 Running 2 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn kube-ovn-cni-7dwmx 1/1 Running 2 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-ovn kube-ovn-cni-lhlvl 1/1 Running 2 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn kube-ovn-controller-57955db7b4-6x6hd 1/1 Running 0 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn kube-ovn-controller-57955db7b4-chvz4 1/1 Running 0 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-ovn ovn-central-bb8747d77-tr5nz 1/1 Running 0 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn ovs-ovn-2qhhr 1/1 Running 0 35h 192.168.1.41 192.168.1.41 <none> <none>
kube-ovn ovs-ovn-np8rn 1/1 Running 0 35h 192.168.1.43 192.168.1.43 <none> <none>
kube-ovn ovs-ovn-pkjw4 1/1 Running 0 35h 192.168.1.42 192.168.1.42 <none> <none>
kube-system coredns-55f46dd959-76qb5 1/1 Running 0 35h 10.16.0.12 192.168.1.42 <none> <none>
kube-system coredns-55f46dd959-wn8kw 1/1 Running 0 35h 10.16.0.11 192.168.1.43 <none> <none>
kube-system heapster-fdb7596d6-xmmrx 1/1 Running 0 35h 10.16.0.15 192.168.1.42 <none> <none>
kube-system kubernetes-dashboard-68ddcc97fc-dwzbf 1/1 Running 0 35h 10.16.0.14 192.168.1.42 <none> <none>
kube-system metrics-server-6c898b5b8b-zvct2 1/1 Running 0 35h 10.16.0.13 192.168.1.43 <none> <none>
test-ns nginx-755464dd6c-s6flj 1/1 Running 0 31h 10.17.0.12 192.168.1.42 <none> <none>
test-ns nginx-755464dd6c-zct56 1/1 Running 0 31h 10.17.0.11 192.168.1.43 <none> <none>
- 更多的测试(pod网络QOS限速,namespace网络隔离等)请参考 kube-ovn 项目说明文档