did you specify the right host or port? error on Kubernetes

Question

I have followed the helloword tutorial on http://kubernetes.io/docs/hellonode/.

When I run:

kubectl run hello-node --image=gcr.io/PROJECT_ID/hello-node:v1 --port=8080

I get:

The connection to the server localhost:8080 was refused - did you specify the right host or port?

Why does the command line try to connect to the localhost?

score 119 · Accepted Answer · edited Jan 25 '17 at 11:15

119

The issue is that your kubeconfig is not right. To auto-generate it run:

gcloud container clusters get-credentials "CLUSTER NAME"

This worked for me.

edited Jan 25 '17 at 11:15

Chillar Anand

27,936
9
119
136

answered May 12 '16 at 13:45

Alan Wagner

2,230
1
16
13

I was also getting this error after updating the configuration of a cluster. Running this get-credentials command worked for me. – Edo Oct 18 '17 at 13:14
12

thanks, for me it asks region name too, `gcloud container clusters get-credentials "CLUSTER NAME" --region=us-central1-a` – Jeyan Apr 12 '20 at 23:42
1

@Alanwagner please update your answer to include region like Jeyan comment. or Jeyan please can you post yours as answer, and it will be the accepted answer – Kris Swat Jul 13 '21 at 17:58

score 44 · Answer 2 · edited May 28 '21 at 02:16

Make sure your config is set to the project -

gcloud config set project [PROJECT_ID]

Run a checklist of the Clusters in the account:

gcloud container clusters list

Check the output :

NAME           LOCATION       MASTER_VERSION  MASTER_IP      MACHINE_TYPE  NODE_VE.      NUM_NODES  STATUS
alpha-cluster  asia-south1-a  1.9.7-gke.6     35.200.254.78  f1-micro      1.9.7-gke.6   3          RUNNING

Run the following cmd to fetch credentials for your running cluster:

gcloud container clusters get-credentials your-cluster-name --zone your-zone       --project your-project

The following output follows:

Fetching cluster endpoint and auth data.
kubeconfig entry generated for alpha-cluster.

Try checking details of the node running kubectl such as below to list all pods in the current namespace, with more details:

kubectl get nodes -o wide

Should be good to go.

`$ gcloud container clusters get-credentials --zone your-zone --project ` did the trick for me. Thanks! — hunj, Apr 30 '20 at 20:01

score 37 · Answer 3 · answered Aug 22 '20 at 04:18

I had same error, this worked for me. Run

minikube status

if the response is

type: Control Plane
host: Stopped
kubelet: Stopped
apiserver: Stopped
kubeconfig: Stopped

run minikube start

type: Control Plane
host: Running
kubelet: Running
apiserver: Running
kubeconfig: Configured

You can proceed

score 12 · Answer 4 · answered Dec 21 '18 at 06:35

12

After running "kubeinit" command, kubernetes asks you to run following as regular user

mkdir -p $HOME/.kube

sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config

sudo chown $(id -u):$(id -g) $HOME/.kube/config

But if you run this as a regular user, you will get "The connection to the server localhost:8080 was refused - did you specify the right host or port?" when trying to access as a root user and vice versa. So try accessing "kubectl" as the user who executed the above commands.

answered Dec 21 '18 at 06:35

Farhan Haider

1,244
1
13
22

If doing this from a worker node - make sure the kubeconfig is present https://stackoverflow.com/questions/36078305/can-we-run-kubectl-form-worker-minion-node – Alex Punnen Jul 24 '19 at 06:09
1

Despite installation, the `kubernetes` folder does not exist on my Machine at `/etc/...` so no question of editing `admin.conf`. Any pointers on how to resolve this? – Apurva Kunkulol Mar 04 '20 at 14:20

score 8 · Answer 5 · edited Mar 28 '21 at 12:50

8

I was trying to connect with local-host and end up with same problem, then I found, I need to start a proxy to the Kubernetes API server.

kubectl proxy --port=8080

https://kubernetes.io/docs/tasks/extend-kubernetes/http-proxy-access-api/

edited Mar 28 '21 at 12:50

Syscall

19,327
10
37
52

answered Mar 28 '21 at 12:34

Rafiq

8,987
4
35
35

score 6 · Answer 6 · answered May 14 '17 at 04:13

Reproduce the same error when doing a tutorial from Udacity called Scalable Microservices with Kubernetes https://classroom.udacity.com/courses/ud615, at the point of Using Kubernetes, Part 3 of Lesson.

Launch a Single Instance:

kubectl run nginx --image=nginx:1.10.0

Error:

Unable to connect to the server: dial tcp [::1]:8080: connectex: No connection could be made because the target machine actively refused it.

How I resolved the Error:

Login to Google Cloud Platform

Navigate to Container Engine Google Cloud Platform, Container Engine

Click CONNECT on Cluster

Use login Credentials to access Cluster [NAME] in your Teminal

Proceeded With Work!!!

score 5 · Answer 7 · answered Mar 09 '19 at 06:33

5

I had the same issue after a reboot, I followed the guide described here

So try the following:

$ sudo -i
# swapoff -a
# exit
$ strace -eopenat kubectl version

After that it works fine.

answered Mar 09 '19 at 06:33

Ganesh Kathiresan

2,068
2
22
33

1

That is the most bizarre solution imaginable… how on earth does using swap matter? The only thing less likely to matter feels like the next line, `strace`ing, it which just logs syscalls right? – ijoseph Jun 03 '22 at 19:50
2

Well k8s requires swap to be turned off. As for `strace`, it's just a check, it's not part of the soultion – Ganesh Kathiresan Jun 12 '22 at 18:07
1

Hm, looking into it more, it makes sense why k8s wants to not use swap. However, it still doesn't make sense why the client machine (using `kubectl`) needs to have swap turned off; seems like just server machines (nodes in the cluster hosting pods) would require that. – ijoseph Jun 12 '22 at 21:01
1

Ok your point got me thinking. I'm going to take a wild guess here: In all the scenarios including the questions and answers here, they are running `kubectl` in the master node itself, basically where `kubeadm` is running. So in essence I guess we are getting `kubeadm` working and hence the `kubectl` client starts to work. You can verify that as the errors are all pointing to `localhost` – Ganesh Kathiresan Jun 13 '22 at 14:17

score 4 · Answer 8 · answered Nov 13 '20 at 07:29

I was getting an error when running

sudo kubectl get pods
The connection to the server localhost:8080 was refused - did you specify the right host or port?

Finally for my environment this command parameter works

sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get pods

when executing kubectl as non root.

score 4 · Answer 9 · edited Jan 19 '21 at 20:29

4

The correct answer, from all above, is to run the commands below:

sudo cp /etc/kubernetes/admin.conf $HOME/

sudo chown $(id -u):$(id -g) $HOME/admin.conf

export KUBECONFIG=$HOME/admin.conf

edited Jan 19 '21 at 20:29

Greenonline

1,330
8
23
31

answered Jan 18 '21 at 22:18

Mohamed Elhassaany

49
5

score 4 · Answer 10 · answered Jun 24 '21 at 18:42

Happening because your kubectl is not able to connect to kubernetes server. Run your cluster.

minikube start

If you want to access service w.r.t your kube config file, you can access it via

 kubectl --kubeconfig ~/.kube/config  get jobs

~/.kube/config : Path of config file, modify w.r.t your file path

score 3 · Answer 11 · answered Apr 17 '16 at 06:51

3

This errors means that kubectl is attempting to connect to a Kubernetes apiserver running on your local machine, which is the default if you haven't configured it to talk to a remote apiserver.

answered Apr 17 '16 at 06:51

Robert Bailey

17,866
3
50
58

I'm trying to work on local machine and still get this error.. how can solve it (I cannot create a pod) – ItayB May 17 '16 at 14:06
1

I find the solution.. I had to run 'docker run -d...' from: http://kubernetes.io/docs/getting-started-guides/docker/ – ItayB May 17 '16 at 14:29

score 3 · Answer 12 · answered May 11 '18 at 12:30

Reinitialising gcloud with proper account and project worked for me.

gcloud init

After this retrying the below command was successful and kubeconfig entry was generated.

gcloud container clusters get-credentials "cluster_name"

check the cluster info with

kubectl cluster-info

score 3 · Answer 13 · answered Jul 18 '19 at 05:00

Regardless of your environment (gcloud or not ) , you need to point your kubectl to kubeconfig. By default, kubectl expects the path as $HOME/.kube/config or point your custom path as env variable (for scripting etc ) export KUBECONFIG=/your_kubeconfig_path

Please refer :: https://kubernetes.io/docs/concepts/configuration/organize-cluster-access-kubeconfig/

If you don't have a kubeconfig file for your cluster, create one by referring :: https://kubernetes.io/docs/tasks/access-application-cluster/configure-access-multiple-clusters/

It is required to find cluster's ca.crt , apiserver-kubelet-client key and cert.

score 3 · Answer 14 · answered Aug 06 '20 at 05:18

3

I got the same trouble since nearly release, seem must use KUBECONFIG explicit

sudo cp /etc/kubernetes/admin.conf $HOME/

sudo chown $(id -u):$(id -g) $HOME/admin.conf

export KUBECONFIG=$HOME/admin.conf

answered Aug 06 '20 at 05:18

Osama Zahid

47
1

1

In my case `current-context` inside `KUBECONFIG` file was changed so it's good idea to try original file `sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get nodes`. – laimison Nov 18 '20 at 19:59
I would say that this is one of the correct answers. Just add admin.conf to the home directory of the regular user. – Kostia Shiian Jul 25 '21 at 05:26

score 3 · Answer 15 · answered Jun 23 '21 at 10:30

as an improvement to Omokahfe:

minikube status

if the response is

E0623 09:12:24.603405   21127 status.go:396] kubeconfig endpoint: extract IP: "minikube" does not appear in /home/<user>/.kube/config
minikube
type: Control Plane
host: Running
kubelet: Running
apiserver: Running
kubeconfig: Misconfigured
timeToStop: Nonexistent


WARNING: Your kubectl is pointing to stale minikube-vm.
To fix the kubectl context, run `**minikube update-context**`

run

minikube update-context

then it will show

* "minikube" context has been updated to point to 10.254.183.66:8443
* Current context is "minikube"

and then

minikube status

will show

type: Control Plane
host: Running
kubelet: Running
apiserver: Running
kubeconfig: Configured
timeToStop: Nonexistent

score 3 · Answer 16 · answered Apr 20 '22 at 07:32

3

activate Docker in your system
run minikube start command in terminal

answered Apr 20 '22 at 07:32

Himanshi Jain

31
1

This advice is already in another answer. – Stephen Ostermiller Apr 26 '22 at 18:42

score 2 · Answer 17 · answered Jun 15 '16 at 09:55

I had this problem using a local docker. The thing to do is check the logs of the containers its spins up to figure out what went wrong. For me it transpired that etcd had fallen over

   $ docker logs <etcdContainerId>
   <snip>
   2016-06-15 09:02:32.868569 C | etcdmain: listen tcp 127.0.0.1:7001: bind: address already in use

Aha! I'd been playing with Cassandra in a docker container and I'd forwarded all the ports since I wasn't sure which it needed exposed and 7001 is one of its ports. Stopping Cassandra, cleaning up the mess and restarting it fixed things.

score 2 · Answer 18 · answered Mar 13 '18 at 19:19

If you created a cluster on AWS using kops, then kops creates ~/.kube/config for you, which is nice. But if someone else needs to connect to that cluster, then they also need to install kops so that it can create the kubeconfig for you:

export AWS_ACCESS_KEY_ID=$(aws configure get aws_access_key_id)
export AWS_SECRET_ACCESS_KEY=$(aws configure get aws_secret_access_key)
export CLUSTER_ALIAS=kubernetes-cluster

kubectl config set-context ${CLUSTER_ALIAS} \
    --cluster=${CLUSTER_FULL_NAME} \
    --user=${CLUSTER_FULL_NAME}

kubectl config use-context ${CLUSTER_ALIAS}

kops export cluster --name ${CLUSTER_FULL_NAME} \
  --region=${CLUSTER_REGION} \
  --state=${KOPS_STATE_STORE}

score 2 · Answer 19 · answered Mar 16 '18 at 22:49

2

try run with sudo permission mode
example sudo kubectl....

answered Mar 16 '18 at 22:49

Italo José

1,558
1
17
50

score 1 · Answer 20 · answered Mar 04 '21 at 14:00

In case someone, like myself, came across this thread because of the underlying error in their Cloud Build step while switching from gcr.io/cloud-builders/kubectl to gcr.io/google.com/cloudsdktool/cloud-sdk, you would need to explicitly call get-credentials for kubectl to work. My pipeline:

steps:
  - name: gcr.io/google.com/cloudsdktool/cloud-sdk
    entrypoint: 'sh'
    args:
      - '-c'
      - |
        gcloud container clusters get-credentials --zone "$$CLOUDSDK_COMPUTE_ZONE" "$$CLOUDSDK_CONTAINER_CLUSTER"
        kubectl call-what-you-need-here
options:
  env:
    - 'CLOUDSDK_COMPUTE_ZONE=europe-west3-a'
    - 'CLOUDSDK_CONTAINER_CLUSTER=my-cluster'

score 0 · Answer 21 · answered Jun 16 '18 at 09:56

I was also getting same below error:

Unable to connect to the server: dial tcp [::1]:8080: connectex: No connection could be made because the target machine actively refused it.

Then I just execute below command and found everything working fine.

PS C:> .\minikube.exe start

Starting local Kubernetes v1.10.0 cluster... Starting VM... Downloading Minikube ISO 150.53 MB / 150.53 MB [============================================] 100.00% 0s Getting VM IP address... Moving files into cluster... Downloading kubeadm v1.10.0 Downloading kubelet v1.10.0 Finished Downloading kubelet v1.10.0 Finished Downloading kubeadm v1.10.0 Setting up certs... Connecting to cluster... Setting up kubeconfig... Starting cluster components... Kubectl is now configured to use the cluster. Loading cached images from config file. PS C:> .\minikube.exe start Starting local Kubernetes v1.10.0 cluster... Starting VM... Getting VM IP address... Moving files into cluster... Setting up certs... Connecting to cluster... Setting up kubeconfig... Starting cluster components... Kubectl is now configured to use the cluster.

score 0 · Answer 22 · answered Aug 20 '20 at 14:01

I got this issue when using " Bash on Windows " with azure kubernetes

az aks get-credentials -n <myCluster>-g <myResourceGroup>

The config file is autogenerated and placed in '~/.kube/config' file as per OS (which is windows in my case)

To solve this - Run from Bash commandline cp <yourWindowsPathToConfigPrintedFromAbobeCommand> ~/.kube/config

score 0 · Answer 23 · answered Jan 05 '22 at 01:00

0

I resolved this issue by removing the incorrect environment variable for the kube config with the command

unset KUBECONFIG

answered Jan 05 '22 at 01:00

operationIntrudeN313

191
1
4

score 0 · Answer 24 · answered Mar 17 '22 at 15:14

Fix the Error – The connection to the server localhost:8080 was refused

Check if the kubeconfig environment variable is exported if not exported

export KUBECONFIG=/etc/kubernetes/admin.conf or $HOME/.kube/config

Check your .kube or config in the home directory file. If you did not found it, then you need to move that to the home directory. using the following command

cp /etc/kubernetes/admin.conf $HOME/

chown $(id -u):$(id -g) $HOME/admin.conf

export KUBECONFIG=$HOME/admin.conf

Whenever you are starting Master Node you may require to set the environment variable. Hence it’s a repetitive task for you. It can be set permanently using the following command.

echo 'export KUBECONFIG=$HOME/admin.conf' >> $HOME/.bashrc

score 0 · Answer 25 · answered Oct 21 '22 at 17:16

when i log in as a root user and run the command kubectl get pods showing error

W1021 22:14:58.411448   41786 loader.go:223] Config not found: /etc/kubernetes/admin.conf The connection to the server localhost:8080 was refused - did you specify the right host or port?

and without the root user showing pods no error.

Christina Chen · Answer 26 · 2022-11-09T11:15:49.493

0

In my case, I started my docker desktop and enabled kubernetes and apply the kubectl command on my terminal, then it worked

edited Nov 09 '22 at 11:15

answered Nov 09 '22 at 11:15

Christina Chen

1
1

score 0 · Answer 27 · answered Dec 24 '22 at 04:07

0

in case minikube is used, the following commands might be used to switch to proper profile:

minikube profile # get current profile
minikube profile <minikube profile> # set profile

answered Dec 24 '22 at 04:07

bambrikii

23
5

score -1 · Answer 28 · answered Aug 03 '21 at 18:22

-1

If you are using docker desktop, make sure you have Kubernetes enabled:

Go to Preferences > Kubernetes and make sure 'Enable Kubernetes' is checked.

answered Aug 03 '21 at 18:22

Jeff Hogg

771
6
4

score -1 · Answer 29 · answered Apr 08 '22 at 11:56

-1

For me, it was simply a full disk, kubectl also gives this error if your disk is full. Sadly, clearing space does not immediately help, a reboot was necessary for me.

answered Apr 08 '22 at 11:56

bjrne

355
5
14

score -3 · Answer 30 · edited Oct 04 '19 at 13:34

-3

Solution is this:

minikube delete
minikube start --vm-driver none

edited Oct 04 '19 at 13:34

ahwayakchih

2,101
19
24

answered Oct 04 '19 at 12:39

Jack

17

did you specify the right host or port? error on Kubernetes

30 Answers30

Linked

Related