环境构建
基础环境构建
[root@master01 cksstudy]# vi studyns01.yaml
apiVersion: v1
kind: Namespace
metadata:
name: studyns01
labels:
role: studyns01
[root@master01 cksstudy]# vi studyns02.yaml
apiVersion: v1
kind: Namespace
metadata:
name: studyns02
labels:
role: studyns02
[root@master01 cksstudy]# vi studyns03.yaml
apiVersion: v1
kind: Namespace
metadata:
name: studyns03
labels:
role: studyns03
[root@master01 cksstudy]# vi studyns04.yaml
apiVersion: v1
kind: Namespace
metadata:
name: studyns04
labels:
role: studyns04
[root@master01 cksstudy]# vi studypod01.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod01
namespace: studyns01
labels:
role: studypod01
spec:
containers:
- name: studypod01
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# vi studypod02.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod02
namespace: studyns01
labels:
role: studypod02
spec:
containers:
- name: studypod02
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# vi studypod03.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod03
namespace: studyns01
labels:
role: studypod03
spec:
containers:
- name: studypod03
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# vi studypod04.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod04
namespace: studyns02
labels:
role: studypod04
spec:
containers:
- name: studypod04
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# vi studypod05.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod05
namespace: studyns03
labels:
role: studypod05
spec:
containers:
- name: studypod05
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# vi studypod06.yaml
apiVersion: v1
kind: Pod
metadata:
name: studypod06
namespace: studyns04
labels:
role: studypod06
spec:
containers:
- name: studypod06
image: training/webapp
command:
- python
- app.py
imagePullPolicy: IfNotPresent
restartPolicy: Always
[root@master01 cksstudy]# kubectl apply -f .
[root@master01 cksstudy]# kubectl -n studyns01 get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
studypod01 1/1 Running 0 14s 10.10.30.108 worker02 <none> <none>
studypod02 1/1 Running 0 14s 10.10.5.46 worker01 <none> <none>
studypod03 1/1 Running 0 14s 10.10.5.47 worker01 <none> <none>
[root@master01 cksstudy]# kubectl -n studyns02 get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
studypod04 1/1 Running 0 16s 10.10.30.109 worker02 <none> <none>
[root@master01 cksstudy]# kubectl -n studyns03 get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
studypod05 1/1 Running 0 19s 10.10.5.48 worker01 <none> <none>
[root@master01 cksstudy]# kubectl -n studyns04 get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
studypod06 1/1 Running 0 21s 10.10.30.110 worker02 <none> <none>
网络测试
[root@master01 cksstudy]# kubectl -n studyns01 exec -ti studypod01 -- /bin/sh
# ping -c 1 10.10.5.46
PING 10.10.5.46 (10.10.5.46) 56(84) bytes of data.
64 bytes from 10.10.5.46: icmp_seq=1 ttl=62 time=0.374 ms
--- 10.10.5.46 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.374/0.374/0.374/0.000 ms
# ping -c 1 10.10.5.47
PING 10.10.5.47 (10.10.5.47) 56(84) bytes of data.
64 bytes from 10.10.5.47: icmp_seq=1 ttl=62 time=0.522 ms
--- 10.10.5.47 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.522/0.522/0.522/0.000 ms
# ping -c 1 10.10.30.109
PING 10.10.30.109 (10.10.30.109) 56(84) bytes of data.
64 bytes from 10.10.30.109: icmp_seq=1 ttl=63 time=0.109 ms
--- 10.10.30.109 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.109/0.109/0.109/0.000 ms
# ping -c 1 10.10.5.48
PING 10.10.5.48 (10.10.5.48) 56(84) bytes of data.
64 bytes from 10.10.5.48: icmp_seq=1 ttl=62 time=0.408 ms
--- 10.10.5.48 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.408/0.408/0.408/0.000 ms
# ping -c 1 10.10.30.110
PING 10.10.30.110 (10.10.30.110) 56(84) bytes of data.
64 bytes from 10.10.30.110: icmp_seq=1 ttl=63 time=0.073 ms
--- 10.10.30.110 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.073/0.073/0.073/0.000 ms
安全策略
策略配置
[root@master01 cksstudy]# vi studynp01.yaml
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: studynp01
namespace: studyns01
spec:
podSelector:
matchLabels:
role: studypod01
policyTypes:
- Ingress
- Egress
ingress:
- from:
- ipBlock:
cidr: 10.10.0.0/16
except:
- 10.10.30.0/24
- namespaceSelector:
matchLabels:
role: studyns03
- podSelector:
matchLabels:
role: studypod02
ports:
- protocol: TCP
port: 5000
egress:
- to:
- ipBlock:
cidr: 10.10.30.0/24
ports:
- protocol: TCP
port: 5000
[root@master01 cksstudy]# kubectl apply -f studynp01.yaml
释义说明:
- 必须字段:类似于Kubernetes其他安全策略,NetworkPolicy 资源需要 apiVersion、 kind 和 metadata 字段。
- pec:NetworkPolicy spec中包含了在一个命名空间中定义特定网络策略所需的所有信息。
- podSelector:每个 NetworkPolicy 都包括一个 podSelector,它对该策略所适用的一组 Pod 进行选择。空的 podSelector 表示匹配命名空间下的所有 Pod。
- policyTypes: 每个 NetworkPolicy 都包含一个 policyTypes 列表,其中包含 Ingress 或 Egress 或两者兼具。policyTypes 字段表示该策略是应用于 进入所选 Pod 的入站流量还是来自所选 Pod 的出站流量,或两者都进行了配置。 如果 NetworkPolicy 未指定 policyTypes 则默认情况下始终设置 Ingress; 如果 NetworkPolicy 有需要配置任何出口规则的话则必须配置 Egress。
- ingress: 每个 NetworkPolicy 可包含一个 ingress 规则的白名单列表。 每个规则都允许同时匹配 from 和 ports 部分的流量。
- egress: 每个 NetworkPolicy 可包含一个 egress 规则的白名单列表。 每个规则都允许匹配 to 和 port 部分的流量。
如上网络策略示例表示:
- 隔离 "studyns01" 命名空间下 "role:=studypod01" 的 Pod ,即隔离studypod01;
- Ingress 规则允许以下 Pod 连接到 "studyns01" 命名空间下的带有 "role:=studypod01" 标签的所有 Pod 的 5000 TCP 端口:
- 允许"studyns01" 命名空间下带有 "role:=studypod01" 标签的所有 Pod入;
- 允许带有 "role=studyns03" 标签的命名空间中的所有 Pod入;
- 允许IP 地址范围为除了 10.10.30.0/24 之外的所有 10.10.0.0/16入。
- Egress 规则允许从"studyns01" 命名空间中带有 "role:=studypod01" 标签的任何 Pod 到 CIDR 10.0.0.0/16 下 5000 TCP 端口的连接。
策略测试
ingress方向测试
- 拥有10.10.30.0/24 ip的Pod(如studypod04)访问studypod01(10.10.5.46):
[root@master01 ~]# kubectl -n studyns02 exec -ti studypod04 -- /bin/bash
root@studypod04:/opt/webapp# nc -v 10.10.30.108 5000 #根据策略应该不通
- 拥有10.10.0.0/16 ip的Pod(如studypod03)访问studypod01(10.10.5.46):
[root@master01 ~]# kubectl -n studyns01 exec -ti studypod03 -- /bin/bash
root@studypod03:/opt/webapp# nc -v 10.10.30.108 5000 #根据策略应该能通
Connection to 10.10.30.108 5000 port [tcp/*] succeeded!
- 所有studyns03命名空间下的Pod(如studypod05)访问studypod01(10.10.5.46):
[root@master01 ~]# kubectl -n studyns03 exec -ti studypod05 -- /bin/bash
root@studypod05:/opt/webapp# nc -v 10.10.30.108 5000 #根据策略应该能通
Connection to 10.10.30.108 5000 port [tcp/*] succeeded!
- 所有带role=studypod02标签的Pod(如studypod02)访问studypod01(10.10.5.46):
[root@master01 ~]# kubectl -n studyns01 exec -ti studypod02 -- /bin/bash
root@studypod02:/opt/webapp# nc -v 10.10.30.108 5000 #根据策略应该能通
Connection to 10.10.30.108 5000 port [tcp/*] succeeded!
egress方向测试
- studypod01访问拥有10.10.30.0/24 ip的Pod(如studypod06):
[root@master01 ~]# kubectl -n studyns01 exec -ti studypod01 -- /bin/bash
root@studypod01:/opt/webapp# nc -v 10.10.30.109 5000 #根据策略应该能通
Connection to 10.10.30.109 5000 port [tcp/*] succeeded!
^C
root@studypod01:/opt/webapp# nc -v 10.10.30.110 5000 #根据策略应该能通
Connection to 10.10.30.110 5000 port [tcp/*] succeeded!
^C
root@studypod01:/opt/webapp# nc -v 10.10.5.46 5000 #根据策略应该不通
^C
to和from行为
在 ingress 的 from 部分或 egress 的 to 部分中指定四种选择器:
- podSelector: 此选择器将在与 NetworkPolicy 相同的命名空间中选择特定的 Pod,ingress 的 from 则表示允许该选择器选定的Pod作为入站流量来源,egress 的 to则表示允许去往该选择器选定的Pod的出站流量(即该podSelector选定的Pod为目的地)。
- namespaceSelector:此选择器将选择特定的命名空间,应将所有 Pod 用作其入站流量来源或出站流量目的地。
- namespaceSelector 和 podSelector: 一个指定 namespaceSelector 和 podSelector 的 to/from 条目选择特定命名空间中的特定 Pod。
示例001:只允许来自带有user=xhyns标签的命名空间下,且带有role=xhyuser的特定Pod的入请求连接。
...
ingress:
- from:
- namespaceSelector:
matchLabels:
user: alice
podSelector:
matchLabels:
role: client
...
示例002:允许来自带有user=xhyns标签的命名空间下的任何Pod,或来自该策略所在命名空间的带有role=xhyuser的Pod的入请求连接。
...
ingress:
- from:
- namespaceSelector:
matchLabels:
user: xhyns
- podSelector:
matchLabels:
role: xhyuser
...
- ipBlock: 此选择器将选择特定的 IP CIDR 范围以用作入站流量来源或出站流量目的地。通常为集群外部IP,Pod IP通常为随机产生,使用Pod IP段无实际意义。
默认策略
默认情况下,如果命名空间中不存在任何策略,则所有进出该命名空间中 Pod 的流量都被允许。 可通过如下方式修改命名空间中的默认行为。
- 默认拒绝所有入站流量
禁止进入studyns01命名空间下的所有Pod的入流量,即默认拒绝所有进入studyns01命名空间任何Pod的流量。
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: default-deny-ingress
namespace: studyns01
spec:
podSelector: {}
policyTypes:
- Ingress
- 默认允许所有入站流量
允许进入studyns01命名空间下的所有Pod的入流量,即默认允许所有进入studyns01命名空间任何Pod的流量。
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: allow-all-ingress
namespace: studyns01
spec:
podSelector: {}
policyTypes:
- Ingress
ingress:
- {}
- 默认拒绝所有出站流量
禁止studyns01命名空间下的所有Pod的出流量,即默认禁止studyns01命名空间任何Pod的出流量。
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: default-deny-egress
namespace: studyns01
spec:
podSelector: {}
policyTypes:
- Egress
- 默认允许所有出站流量
允许studyns01命名空间下的所有Pod的出流量,即默认允许studyns01命名空间任何Pod的出流量。
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: allow-all-egress
namespace: studyns01
spec:
podSelector: {}
egress:
- {}
policyTypes:
- Egress
- 默认拒绝所有入口和所有出站流量
禁止studyns01命名空间下的所有出入流量。
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: default-deny-all
spec:
podSelector: {}
policyTypes:
- Ingress
- Egress