# CI Images Test on Operator

## Pick up the build namespace

* Go to your PR and open the raw build-log. totxt pick-up where the namespace for your PR has been created. You should look like for a log message like this :

2019/09/20 08:51:52 Creating namespace 

now you know that `ci-op-z1jqr9id` is the repository, you shoudd have access to that namespace on https://api.ci.openshift.org/console/ as long as you are the owner of that PR (it creates a `Rolebinding` for this).

## Generate yaml file

* Checkout openshift/pipeline and switch to the branch `openshift/release-next` (I assume your downstream remote is called `openshift`)
* execute the following from **bash** shell :

source ./openshift/resolve-yamls.sh
resolve_resources config /tmp/tekton-pipeline.yaml "nothing"  registry.svc.ci.openshift.org/${NS}/stable

${NS} is the variable where you have stored your namespace.

## Setup a pullSecret to your default SA

You need to have access to that registry, from https://api.ci.openshift.org/console get your login command for login with the CLI.

When logged in, execute this :

oc registry login --to=/tmp/ciconfig.json

you should have a dockerconfig with your token in `/tmp/ciconfig.json`

Create a secret out of it :

kubectl create secret generic regcred \
    --from-file=.dockerconfigjson=/tmp/ciconfig.json \
    --type=kubernetes.io/dockerconfigjson

Get your default SA :

oc get sa -o yaml > /tmp/sa.yaml

Edit the /tmp/sa.yaml and add to the imagePullSecrets section the `-name: regcred` and remove the extra data not needed, this would look like this for me :

apiVersion: v1
imagePullSecrets:
  - name: default-dockercfg-wqsss
  - name: regcred
kind: ServiceAccount
metadata:
  name: default
  namespace: tekton-pipelines
secrets:
  - name: default-dockercfg-wqsss
  - name: default-token-d6dls

now execute :

oc create -f /tmp/sa.yaml

## Apply templte

You can now apply the template with :

oc create -f /tmp/tekton-pipeline.yaml 

It may have a race between the project creation and the deployments, so you may have to apply it twice if it can't find the `tektoncd-pipeline` project

It should be up and have pulled your image from the CI :

% oc describe -n tektoncd-pipeline pod tekton-pipelines-controller-77b5fd554f-wmf5d|grep Image:
    Image:         registry.svc.ci.openshift.org/ci-op-cqghg72m/stable:tektoncd-pipeline-controller

## Caveats

* The images get trashed after a while so you will need to rerun a CI run grab the new namespace and redo that whole shbang :heart_decoration: