Question

我使用Gitlab在GKE中创建了一个集群，并使用gitab的界面安装了Helm＆Tiller以及其他一些东西，例如ingress和gitlabRunner。但是，当我尝试使用gcloud的头盔进行安装时，显示“错误：传输正在关闭”。

我做了gcloud container clusters get-credentials ...。

$ kubectl get pods --all-namespaces

NAMESPACE               NAME                                                     READY     STATUS    RESTARTS   AGE
default                 jaeger-deployment-59ffb979c8-lmjk5                       1/1       Running   0          17h
gitlab-managed-apps     certmanager-cert-manager-6c8cd9f9bf-67wnh                1/1       Running   0          17h
gitlab-managed-apps     ingress-nginx-ingress-controller-75c4d99549-x66n4        1/1       Running   0          21h
gitlab-managed-apps     ingress-nginx-ingress-default-backend-6f58fb5f56-pvv2f   1/1       Running   0          21h
gitlab-managed-apps     prometheus-kube-state-metrics-6584885ccf-hr8fw           1/1       Running   0          22h
gitlab-managed-apps     prometheus-prometheus-server-69b9f444df-htxsq            2/2       Running   0          22h
gitlab-managed-apps     runner-gitlab-runner-56798d9d9d-nljqn                    1/1       Running   0          22h
gitlab-managed-apps     tiller-deploy-74f5d65d77-xk6cc                           1/1       Running   0          22h
kube-system             heapster-v1.6.0-beta.1-7bdb4fd8f9-t8bq9                  2/2       Running   0          22h
kube-system             kube-dns-7549f99fcc-bhg9t                                4/4       Running   0          22h
kube-system             kube-dns-autoscaler-67c97c87fb-4vz9t                     1/1       Running   0          22h
kube-system             kube-proxy-gke-cluster2-pool-1-05abcbc6-0s6j             1/1       Running   0          20h
kube-system             kube-proxy-gke-cluster2-pool-2-67e57524-ht5p             1/1       Running   0          22h
kube-system             metrics-server-v0.2.1-fd596d746-289nd                    2/2       Running   0          22h
visual-react-10450736   production-847c7d879c-z4h5t                              1/1       Running   0          22h
visual-react-10450736   production-postgres-64cfcf9464-jr74c                     1/1       Running   0          22h

$ ./helm install stable/wordpress --tiller-namespace gitlab-managed-apps --name wordpress

E0127 10:27:29.790366     418 portforward.go:331] an error occurred forwarding 39113 -> 44134: error forwarding port 44134 to pod 86b33bdc7bc30c08d98fe44c0772517c344dd1bdfefa290b46e82bf84959cb6f, uid : exit status 1: 2019/01/27 04:57:29 socat[11124] E write(5, 0x14ed120, 186): Broken pipe

Error: transport is closing

另一个

$ ./helm install incubator/jaeger --tiller-namespace gitlab-managed-apps --name jaeger --set elasticsearch.rbac.create=true --set provisionDataStore.cassandra=false --set provisionDataStore.elasticsearch=true --set storage.type=elasticsearch

E0127 10:30:24.591751     429 portforward.go:331] an error occurred forwarding 45597 -> 44134: error forwarding port 44134 to pod 86b33bdc7bc30c08d98fe44c0772517c344dd1bdfefa290b46e82bf84959cb6f, uid : exit status 1: 2019/01/27 05:00:24 socat[13937] E write(5, 0x233d120, 8192): Connection reset by peer

Error: transport is closing

我尝试自己转发端口，但它永远不会返回提示，需要花费很长时间。
kubectl port-forward --namespace gitlab-managed-apps tiller-deploy 39113:44134

显然，从Gitab的ui安装任何东西都使用Helm，并且这些都不会失败。但是从shell这样做失败。请帮帮我。预先感谢。

Answer 1

我知道已经很晚了，但是我会分享给大家，以防其他人为这个问题而苦恼。我在gitlab论坛中找到了答案：HERE。

技巧是从分er服务帐户中导出和解码证书，并将其作为参数传递给头盔，如下所示：

helm list --tiller-connection-timeout 30 --tls --tls-ca-cert tiller-ca.crt --tls-cert tiller.crt --tls-key tiller.key ---all --tiller-namespace gitlab-managed-apps

无法在群集上进行“临时安装”。耕iller机由gitab安装

1 个答案: