On-Premises

Using Boot On-Premises

On-premises kubernetes clusters tend to differ wildly so its hard for us to know your Jenkins X installation will totally work on any cluster given we typically rely on things like persistence, ingress, storage.

Here are some recommendations to hopefully get you started. If you hit any issues please join our community we can hopefully help you.

Configuration

Please set your provider to kubernetes via this in your jx-requirements.yml:

clusterConfig:
    provider: kubernetes

Ingress

If you don’t have a real ingress solution for your on-premises cluster you can start off using the nginx-controller and the IP address of your api server.

You can set ingress.domain to be 1.2.3.4.nip.io where 1.2.3.4 is the IP address of your ingress service.

By default boot will try to recreate the ingress.domain by discovering the IP address on the nginx controller service - which is commonly generated dynamically on the public clouds.

If you are on-premises and using a hard coded IP address for ingress you may want to set this on your jx-requirements.yml

clusterConfig:
    provider: kubernetes
ingress:
  domain: 1.2.3.4.nip.io
  ignoreLoadBalancer: true

General advice

We recommend starting with the most simple possible installation and get that working, then gradually try to be more complex. e.g. start off by ignoring these features:

  • vault
  • DNS
  • TLS / certificates
  • cloud storage for artifacts

Then once you have something working, incrementally try enabling each of those in turn.

If a helm step fails some resources may have already been created, running the step again will result in a conflict (resource already exists). To delete all created resource:

  • configure jx to keep the tmp folder with compiled helm files:

    export JX_NO_DELETE_TMP_DIR=true

  • go to the tmp folder visible in the logs and execute

    helm template jenkins-x . --namespace jx | kubectl delete -f -

Last modified June 18, 2021: docs: Fix usage of "on-premises" (91a1e4b3eb)