Common Problems

Questions on common issues setting up Jenkins X.

We have tried to collate common issues here with work arounds. If your issue isn’t listed here please let us know.

Jenkins X does not startup

If your install fails to start there could be a few different reasons why the Jenkins X pods don’t start.

Your cluster could be out of resources. You can check the spare resources on your cluster via jx status:

jx status

We also have a diagnostic command that looks for common problems jx step verify install:

jx step verify install

A common issue for pods not starting is if your cluster does not have a default storage class setup so that Persistent Volume Claims can be bound to Persistent Volumes as described in the install instructions.

You can check your storage class and persistent volume setup via:

kubectl get pvc

If things are working you should see something like:

$ kubectl get pvc
NAME                        STATUS    VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
jenkins                     Bound     pvc-680b39b5-94f1-11e8-b93d-42010a840238   30Gi       RWO            standard       12h
jenkins-x-chartmuseum       Bound     pvc-6808fb5e-94f1-11e8-b93d-42010a840238   8Gi        RWO            standard       12h
jenkins-x-docker-registry   Bound     pvc-680a415c-94f1-11e8-b93d-42010a840238   100Gi      RWO            standard       12h
jenkins-x-mongodb           Bound     pvc-680d6fd9-94f1-11e8-b93d-42010a840238   8Gi        RWO            standard       12h
jenkins-x-nexus             Bound     pvc-680fc692-94f1-11e8-b93d-42010a840238   8Gi        RWO            standard       12h

If you see status of Pending then this indicates that you have no default storage class setup on your kubnernetes cluster or you have ran out of persistent volume space.

Please try create a default storage class for your cluster or contact your operations team or cloud provider.

If the Persistent Volume Claims are all Bound and things still have not started then try

kubectl get pod

If a pod cannot start try

kubectl describe pod some-pod-name

Maybe that gives you a clue. Is it RBAC related maybe?

If you are still stuck try create an issue

http: server gave HTTP response to HTTPS client

If your pipeline fails with something like this:

The push refers to a repository []
time="2018-07-09T21:18:31Z" level=fatal msg="build step: pushing []: Get http: server gave HTTP response to HTTPS client"

Then this means that you are using the internal docker registry inside Jenkins X for your images but your kubernetes cluster’s docker daemons has not been configured for insecure-registries so that you can use http to talk to the docker registry service jenkins-x-docker-registry in your cluster.

By default docker wants all docker registries to be exposed over https and to use TLS and certificates. This should be done for all public docker registries. However when using Jenkins X with an internal local docker registry this is hard since its not available at a public DNS name and doesn’t have HTTPS or certificates; so we default to requiring insecure-registry be configured on all the docker daemons for your kubernetes worker nodes.

We try to automate this setting when using jx create cluster e.g. on AWS we default this value to the IP range to match most kubernetes service IP addresses.

On EKS we default to using ECR to avoid this issue. Similarly we will soon default to GCR and ACR on GKE and AKS respectively.

So a workaround is to use a real external docker registry or enable insecure-registry on your docker daemons on your compute nodes on your Kubernetes cluster.

Helm fails with Error: UPGRADE FAILED: incompatible versions client[…] server[…]’

Generally speaking this happens when your laptop has a different version of helm to the version used in our build pack docker images and/or the version of tiller thats running in your server.

The simplest fix for this is to just not use tiller at all - which actually helps avoid this problem ever happening and solves a raft of security issues too.

However switching from using Tiller to No Tiller does require a re-install of Jenkins X (though you could try do that in separate set of namespaces then move projects across incrementally?).

The manual workaround is to install the exact same version of helm as used on the server

Or you can try switch tiller to match your client version:

  • run helm init --upgrade

Though as soon as a pipeline runs it’ll switch the tiller version again so you’ll have to keep repeating the above.

error creating jenkins credential jenkins-x-chartmuseum 500 Server Error

This is a pending issue which we will hopefully fix soon.

It basically happens if you have an old API token in ~/.jx/jenkinsAuth.yaml for your jenkins server URL. You can either:

  • remove it from that file by hand

  • run the following command jx delete jenkins token:

    jx delete jenkins token admin

errors with

If you see errors like:

error:failed to add the repository 'jenkins-x' with URL ''


Looks like "" is not a valid chart repository or cannot be reached

then it looks like you have a reference to an old chart museum URL for Jenkins X charts.

The new URL is:

It could be your helm install has an old repository URL installed. You should see…

$ helm repo list

If you see this…

$ helm repo list

then please run…

helm repo remove jenkins-x
helm repo add jenkins-x

and you should be good to go again.

Another possible cause is an old URL in your environment’s git repository may have old references to the URL.

So open your env/requirements.yaml in your staging/production git repositories and modify them to use the URL instead of like this env/requirements file

git errors: POST 401 Bad credentials

This indicates your git API token either was input incorrectly or has been regenerated and is now incorrect.

To recreate it with a new API token value try the following (changing the git server name to match your git provider):

jx delete git token -n github <yourUserName>
jx create git token -n github <yourUserName>

Invalid git token to scan a project

If you get an error in Jenkins when it tries to scan your repositories for branches something like:

hudson.AbortException: Invalid scan credentials *****/****** (API Token for acccessing git service inside pipelines) to connect to, skipping

Then your git API token was probably wrong or has expired.

To recreate it with a new API token value try the following (changing the git server name to match your git provider):

jx delete git token -n GitHub admin
jx create git token -n GitHub admin

What are the credentials to access core services?

Authenticated core services of Jenkins X include Jenkins, Nexus, ChartMuseum. The default username is adminand the password by default is generated and printed out in the terminal after jx create cluster or jx install.

Set Admin Username and Password values for Core Services

You can also set the admin username via the --default-admin-username=username flag.

If you would like to set the default password yourself then you can set the flag --default-admin-password=foo to the two comamnds above.

If you don’t have the terminal console output anymore you can look in the local file ~/.jx/jenkinsAuth.yaml and find the password that matches your Jenkins server URL for the desired cluster.

Persistent Volume Claims do not bind

If you notice that the persistent volume claims created when installing Jenkins X don’t bind with

kubectl get pvc

The you should check that you have a cluster default storage class for dynamic persistent volume provisioning. See here for more details.

I cannot connect to nodes on AWS

If you don’t see any valid nodes returned by kubectl get node or you get errors running jx status something like:

Unable to connect to the server: dial tcp: lookup on no such host

it could be your kube config is stale. Try

aws eks --region <CLUSTER_REGION> update-kubeconfig --name <CLUSTER_NAME>

That should regenerate your local ~/kube/config file and so kubectl get node or jx status should find your nodes

How can I diagnose exposecontroller issues?

When you promote a new version of your application to an environment, such as the Staging Environment a Pull Request is raised on the environment repository.

When the master pipeline runs on an environment a Kubernetes Job is created for exposecontroller which runs a pod until it terminates.

It can be tricky finding the log for temporary jobs since the pod is removed.

One way to diagnose logs in your, say, Staging environment is to download and install kail and add it to your PATH.

Then run this command:

kail -l job-name=expose -n jx-staging

If you then promote to the Staging environment or retrigger the pipeline on the master branch of your Staging git repository (e.g. via jx start pipeline) then you should see the output of the exposecontroller pod.

Why is promotion really slow?

If you find you get lots of warnings in your pipelines like this…

"Failed to query the Pull Request last commit status for ref xyz Could not find a status for repository myorg/environment-mycluster-staging with ref xyz

and promotion takes 30 minutes from a release pipeline on an application starting to the change hitting Staging then its mostly probably due to Webhooks.

When we import projects or create quickstarts we automate the setup of CI/CD pipelines for the git repository. What this does is setup Webhooks on the git repository to trigger Jenkins X to trigger pipelines (either using Prow for serverless Jenkins X Pipelines or the static jenkins server if not).

However sometimes your git provider (e.g. GitHub may not be able to do connect to your Jenkins X installation (e.g. due to networking / firewall issues).

The easiest way to diagnose this is opening the git repository (e.g. for your environment repository).

jx get env


  • click on the generated URL for, say, your Staging git repository
  • click the Settings icon
  • select the Webhooks tab on the left
  • select your Jenkins X webhook URL
  • view the last webhook - did it succeed? Try re-trigger it? That should highlight any network issues etc

If you cannot use public webhooks you could look at something like ultrahook

How do I see the log of exposecontroller?

Usually we run the exposecontroller as a post install Job when we perform promotion to Staging or Production to expose services over Ingress and possibly inject external URLs into applications configuration.

So the Job will trigger a short lived Pod to run in the namespace of your environment, then the pod will be deleted.

If you want to view the logs of the exposecontroller you will need to watch for the logs using a selector then trigger the promotion pipeline to capture it.

One way to do that is via the kail CLI:

kail -l  job-name=expose

This will watch for exposecontroller logs and then dump them to the console. Now trigger a promotion pipeline and you should see the output within a minute or so.

Cannot create TLS certificates during Ingress setup

cert-manager cert-manager is a seperate project from Jenkins X.

Newly created GKE clusters or existing cluster running kubernetes v1.12 or older will encounter the following error when configuring Ingress with site-wide TLS:

Waiting for TLS certificates to be issued...
Timeout reached while waiting for TLS certificates to be ready

This issue is caused by the cert-manager pod not having the disable-validation label set, which is a known cert-manager issue which is documented on their website. The following steps, taken from the cert-manager/troubleshooting-installation webpage, should resolve the issue:

Check if the disable-validation label exists on the cert-manager pod.

kubectl describe namespace cert-manager

If you cannot see the label on your namespace, you should add it with:

kubectl label namespace cert-manager

Confirm the label has been added to the cert-manager pod.

kubectl describe namespace cert-manager

Name:         cert-manager
Annotations:  <none>
Status:       Active

Now rerun jx Ingress setup:

jx upgrade ingress

While the ingress command is running, you can tail the cert-manager logs in another terminal and see what is happening. You will need to find the name of your cert-manager pod using:

kubectl get pods --namespace cert-manager

Then tail the logs of the cert-manager pod.

kubectl logs YOUR_CERT_MNG_POD --namespace cert-manager -f

Your TLS certificates should now be set up and working, otherwise checkout the official cert-manager troubleshooting instructions.

Recreating a cluster with the same name

If you want to destroy a cluster that was created with boot and recreate it with the exact same name, there is some clean that needs to be done first.

Make sure you uninstall jx:

jx uninstall

Delete the cluster either from the web console or terminal by using the Kubernetes provider CLI command:

gcloud container clusters delete <cluster-name> --zone <cluster-zone>

After you have successfully done this, remove the ~/.jx and ~/.kube directories:

rm -rf ~/.jx ~/.kube

Delete any repositories created by jx on your Github organisations account.

Delete the local git jenkins-x-boot-config repository.

That should leave your Kubernetes provider and your local environment in a clean state.

No External IP assigned to the ingress in EKS

If you have never created an Elastic Load Balancer (ELB) in AWS, then the jx boot command will fail to assign an ip to the ingress controller.

In this case, the output from jx boot might look like this:

error: failed to discover the Ingress domain: getting a domain for ingress service kube-system/jxing-nginx-ingress-controller: Timed out after 5m0s, last error: %!s(<nil>)

To verify this is the case, run:

kubectl get svc --all-namespaces

If you see pending for external IP for jxing-nginx-ingress-controller

NAMESPACE     NAME                                  TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
kube-system   jxing-nginx-ingress-controller        LoadBalancer   X.X.X.X         <pending>     80:32632/TCP,443:32036/TCP   4h45m

then create an ELB manually outside of EKS, and tear it down.

After that, run jx boot again.

Other issues

Please let us know and see if we can help? Good luck!

Last modified July 13, 2020: release 0.0.1819 (0981d42)