When deploying a follower to openshift 4.5 via start.sh, the follower deployment is created referring to the seedfetcher image by it's external url. This causes the follower pod to fail to start as it cannot pull the image for it's first init container.
Detail
The following logic determines if an internal or external url is generated:
if ! [ -z ${TEST_PLATFORM+x} ] && [[ $TEST_PLATFORM =~ ^openshift4 ]] && [[ "$internal" == "true" ]]; then
echo "image-registry.openshift-image-registry.svc:5000/$CONJUR_NAMESPACE_NAME/$1:$CONJUR_NAMESPACE_NAME"
else
echo "$DOCKER_REGISTRY_PATH/$CONJUR_NAMESPACE_NAME/$1:$CONJUR_NAMESPACE_NAME"
fi
TEST_PLAFORM is only set in test.sh and test.sh is only called from the Jenkinsfile. So when a user's entrypoint is start.sh, an internal url will not be generated.
Workaround
Add the following to bootstrap.env when using openshift 4x
Summary
When deploying a follower to openshift 4.5 via start.sh, the follower deployment is created referring to the seedfetcher image by it's external url. This causes the follower pod to fail to start as it cannot pull the image for it's first init container.
Detail
The following logic determines if an internal or external url is generated:
From: https://github.com/cyberark/kubernetes-conjur-deploy/blob/master/utils.sh#L48
TEST_PLAFORM is only set in test.sh and test.sh is only called from the Jenkinsfile. So when a user's entrypoint is start.sh, an internal url will not be generated.
Workaround
Add the following to bootstrap.env when using openshift 4x