可以通过向它们所属服务的 dns 发送网络请求来访问 kubernetes 集群中的 Pod。网络请求必须发送到[service].[namespace].svc.cluster.local，并在该服务的所有成员之间实现负载平衡。
这可以让某个 pod 到达另一个 pod 可以正常工作，但如果 pod 尝试通过他所属的服务访问自身，则会失败。它始终会导致超时。
这是 Kubernetes 中的一个错误(在我的例子中是 minikube v0.35.0)还是需要一些额外的配置？
下面是一些调试信息：
让我们从其他 pod 联系该服务。这工作正常：
daemon@auth-796d88df99-twj2t:/opt/docker$ curl -v -X POST -H "Accept: application/json" --data '{}' http://message-service.message.svc.cluster.local:9000/message/get-messages
Note: Unnecessary use of -X or --request, POST is already inferred.
* Trying 10.107.209.9...
* TCP_NODELAY set
* Connected to message-service.message.svc.cluster.local (10.107.209.9) port 9000 (#0)
> POST /message/get-messages HTTP/1.1
> Host: message-service.message.svc.cluster.local:9000
> User-Agent: curl/7.52.1
> Accept: application/json
> Content-Length: 2
> Content-Type: application/x-www-form-urlencoded
>
* upload completely sent off: 2 out of 2 bytes
< HTTP/1.1 401 Unauthorized
< Referrer-Policy: origin-when-cross-origin, strict-origin-when-cross-origin
< X-Frame-Options: DENY
< X-XSS-Protection: 1; mode=block
< X-Content-Type-Options: nosniff
< Content-Security-Policy: default-src 'self'
< X-Permitted-Cross-Domain-Policies: master-only
< Date: Wed, 20 Mar 2019 04:36:51 GMT
< Content-Type: text/plain; charset=UTF-8
< Content-Length: 12
<
* Curl_http_done: called premature == 0
* Connection #0 to host message-service.message.svc.cluster.local left intact
Unauthorized
现在，我们尝试让 Pod 联系他所属的服务：
Note: Unnecessary use of -X or --request, POST is already inferred.
* Trying 10.107.209.9...
* TCP_NODELAY set
* connect to 10.107.209.9 port 9000 failed: Connection timed out
* Failed to connect to message-service.message.svc.cluster.local port 9000: Connection timed out
* Closing connection 0
curl: (7) Failed to connect to message-service.message.svc.cluster.local port 9000: Connection timed out
如果我正确阅读了 curl 调试日志，dns 将解析为 IP 地址 10.107.209.9。可以通过该 ip 从任何其他 Pod 访问 Pod ，但 Pod 不能使用它来访问自身。
以下是尝试访问自身的 Pod 的网络接口：
daemon@message-58466bbc45-lch9j:/opt/docker$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
2: sit0@NONE: <NOARP> mtu 1480 qdisc noop state DOWN group default qlen 1000
link/sit 0.0.0.0 brd 0.0.0.0
296: eth0@if297: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
link/ether 02:42:ac:11:00:09 brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 172.17.0.9/16 brd 172.17.255.255 scope global eth0
valid_lft forever preferred_lft forever
以下是部署到minikube的kubernetes文件：
apiVersion: v1
kind: Namespace
metadata:
name: message
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
labels:
app: message
name: message
namespace: message
spec:
replicas: 1
selector:
matchLabels:
app: message
strategy:
rollingUpdate:
maxSurge: 1
maxUnavailable: 0
type: RollingUpdate
template:
metadata:
labels:
app: message
spec:
containers:
- name: message
image: message-impl:0.1.0-SNAPSHOT
imagePullPolicy: Never
ports:
- name: http
containerPort: 9000
protocol: TCP
env:
- name: NAMESPACE
valueFrom:
fieldRef:
fieldPath: metadata.namespace
- name: KAFKA_KUBERNETES_NAMESPACE
value: kafka
- name: KAFKA_KUBERNETES_SERVICE
value: kafka-svc
- name: CASSANDRA_KUBERNETES_NAMESPACE
value: cassandra
- name: CASSANDRA_KUBERNETES_SERVICE
value: cassandra
- name: CASSANDRA_KEYSPACE
value: service_message
---
# Service for discovery
apiVersion: v1
kind: Service
metadata:
name: message-service
namespace: message
spec:
ports:
- port: 9000
protocol: TCP
selector:
app: message
---
# Expose this service to the api gateway
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
name: message
namespace: message
annotations:
kubernetes.io/ingress.class: traefik
spec:
rules:
- host: api.fload.cf
http:
paths:
- path: /message
backend:
serviceName: message-service
servicePort: 9000
这是一个已知的minikube问题。讨论包含以下解决方法：
1)尝试：
minikube ssh
sudo ip link set docker0 promisc on
2) 使用无头服务：clusterIP: None