Kubernetes Jenkins插件-奴隶始终离线 [英] Kubernetes Jenkins plugin - slaves always offline

问题描述

我正在尝试使用Jenkins/Kubernetes插件与Jenkins协调Docker从属服务器.

I'm trying to use the Jenkins/Kubernetes plugin to orchestrate docker slaves with Jenkins.

我正在使用此插件: https://github.com/jenkinsci/kubernetes-plugin

我的问题是所有从站都处于脱机状态，因此作业无法执行:

My problem is that all the slaves are offline so the job can't execute:

我在使用minikube的本地机器上以及在由ops小组托管的K8集群上尝试了此操作.我已经尝试了Jenkins 1.9和Jenkins2.我总是得到相同的结果.屏幕截图来自Jenkins 1.642.4，K8 v1.2.0

I have tried this on my local box using minikube, and on a K8 Cluster hosted by our ops group. I've tried both Jenkins 1.9 and Jenkins 2. I always get the same result. The screenshots are from Jenkins 1.642.4, K8 v1.2.0

这是我的配置...请注意，当我单击测试连接"时，我获得了成功.另外请注意，我不需要任何凭据(这是我看到的与记录的示例唯一的区别).

Here is my configuration... note that when I click 'test connection' I get a success. Also note I didn't need any credentials (this is the only difference I can see vs the documented example).

Jenkins日志一遍又一遍显示以下内容:

The Jenkins log shows the following over and over:

    Waiting for slave to connect (11/100): docker-6b55f1b7fafce
Jul 20, 2016 5:01:06 PM INFO org.csanchez.jenkins.plugins.kubernetes.KubernetesCloud$ProvisioningCallback call
Waiting for slave to connect (12/100): docker-6b55f1b7fafce
Jul 20, 2016 5:01:07 PM INFO org.csanchez.jenkins.plugins.kubernetes.KubernetesCloud$ProvisioningCallback call
Waiting for slave to connect (13/100): docker-6b55f1b7fafce
Jul 20, 2016 5:01:08 PM INFO org.csanchez.jenkins.plugins.kubernetes.KubernetesCloud$ProvisioningCallback call

当我运行kubectl get events时，我看到了:

When I run kubectl get events I see this:

24s         24s        1         docker-6b3c2ff27dad3   Pod                                Normal    Scheduled           {default-scheduler }      Successfully assigned docker-6b3c2ff27dad3 to 96.xxx.xx.159
24s         23s        2         docker-6b3c2ff27dad3   Pod                                Warning   MissingClusterDNS   {kubelet 96.xxx.xx.159}   kubelet does not have ClusterDNS IP configured and cannot create Pod using "ClusterFirst" policy. Falling back to DNSDefault policy.
23s         23s        1         docker-6b3c2ff27dad3   Pod       spec.containers{slave}   Normal    Pulled              {kubelet 96.xxx.xx.159}   Container image "jenkinsci/jnlp-slave" already present on machine
23s         23s        1         docker-6b3c2ff27dad3   Pod       spec.containers{slave}   Normal    Created             {kubelet 96.xxx.xx.159}   Created container with docker id 82fcf1bd0328
23s         23s        1         docker-6b3c2ff27dad3   Pod       spec.containers{slave}   Normal    Started             {kubelet 96.xxx.xx.159}   Started container with docker id 82fcf1bd0328

有什么想法吗?

更新:更多csanchez建议的日志信息

UPDATE: more log info as suggested by csanchez

 ➜  docker git:(master) ✗ kubectl get pods --namespace default -o wide
NAME                   READY     STATUS    RESTARTS   AGE       NODE
docker-6bb647254a2a4   1/1       Running   0          1m        96.x.x.159

➜  docker git:(master) ✗ kubectl log docker-6bafbac10b392

    Jul 20, 2016 6:45:10 PM hudson.remoting.jnlp.Main$CuiListener status
INFO: Connecting to 96.x.x.159:50000 (retrying:10)
java.net.ConnectException: Connection refused
    at java.net.PlainSocketImpl.socketConnect(Native Method)
    at java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:350)
    at java.net.AbstractPlainSocketImpl.connectToAddress(AbstractPlainSocketImpl.java:206)
    at java.net.AbstractPlainSocketImpl.connect(AbstractPlainSocketImpl.java:188)
    at java.net.SocksSocketImpl.connect(SocksSocketImpl.java:392)

我将不得不查看该端口50000的用途?

I'll have to look at what this port 50000 is used for??

推荐答案

您需要公开端口8080和50000，如插件示例config

You need to expose both port 8080 and 50000 as described in the plugin example config https://github.com/jenkinsci/kubernetes-plugin/blob/master/src/main/kubernetes/jenkins.yml

这篇关于Kubernetes Jenkins插件-奴隶始终离线的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！