Test-Infra

This project deploys the OpenShift Assisted Installer in Minikube and spawns libvirt VMs that represent bare metal hosts.

Table of contents

Test-Infra

Prerequisites

CentOS 8 or RHEL 8 host
File system that supports d_type
Ideally on a bare metal host with at least 64G of RAM.
Run as a user with password-less sudo access or be ready to enter sudo password for prepare phase.
Make sure to unset the KUBECONFIG variable in the same shell where you run make.
Get a valid pull secret (JSON string) from redhat.com if you want to test the installation (not needed for testing only the discovery flow). Export it as:

export PULL_SECRET='<pull secret JSON>'

Installation Guide

Check the Install Guide for installation instructions.

OS parameters used for configuration

Variable	Description
AGENT_DOCKER_IMAGE	agent docker image to use, will update assisted-service config map with given value
ASSISTED_SERVICE_HOST	FQDN or IP address to where assisted-service is deployed. Used when DEPLOY_TARGET="onprem".
BASE_DNS_DOMAINS	base DNS domains that are managed by assisted-service, format: domain_name:domain_id/provider_type.
BASE_DOMAIN	base domain, needed for DNS name, default: redhat.com
CLUSTER_ID	cluster id , used for install_cluster command, default: the last spawned cluster
CLUSTER_NAME	cluster name, used as prefix for virsh resources, default: test-infra-cluster
DEPLOY_MANIFEST_PATH	the location of a manifest file that defines image tags images to be used
DEPLOY_MANIFEST_TAG	the Git tag of a manifest file that defines image tags to be used
DEPLOY_TAG	the tag to be used for all images (assisted-service, assisted-installer, agent, etc) this will override any other os parameters
DEPLOY_TARGET	Specifies where assisted-service will be deployed. Defaults to "minikube". "onprem" will deploy assisted-service in a pod on the localhost.
ENABLE_AUTH	configure assisted-service to authenticate API requests, default: false
HTTPS_PROXY_URL	A proxy URL to use for creating HTTPS connections outside the cluster
HTTP_PROXY_URL	A proxy URL to use for creating HTTP connections outside the cluster
IMAGE_BUILDER	image-builder image to use, will update assisted-service config map with given value
INSTALLER_IMAGE	assisted-installer image to use, will update assisted-service config map with given value
IPv4	Boolean value indicating if IPv4 is enabled. Default is yes
IPv6	Boolean value indicating if IPv6 is enabled. Default is no
ISO	path to ISO to spawn VM with, if set vms will be spawn with this iso without creating cluster. File must have the '.iso' suffix
KUBECONFIG	kubeconfig file path, default: /.kube/config
MASTER_MEMORY	memory for master VM, default: 16984MB
NETWORK_CIDR	network CIDR to use for virsh VM network, default: "192.168.126.0/24"
NETWORK_NAME	virsh network name for VMs creation, default: test-infra-net
NO_PROXY_VALUES	A comma-separated list of destination domain names, domains, IP addresses, or other network CIDRs to exclude proxying
NUM_MASTERS	number of VMs to spawn as masters, default: 3
NUM_WORKERS	number of VMs to spawn as workers, default: 0
OCM_BASE_URL	OCM API URL used to communicate with OCM and AMS, default: https://api-integration.6943.hive-integration.openshiftapps.com
OCM_CLIENT_ID	ID of Service Account used to communicate with OCM and AMS for Agent Auth and Authz
OCM_CLIENT_SECRET	Password of Service Account used to communicate with OCM and AMS for Agent Auth and Authz
OC_MODE	if set, use oc instead of minikube
OC_SCHEME	Scheme for assisted-service url on oc, default: http
OC_SERVER	server for oc login, required if oc-token is provided, default: https://api.ocp.prod.psi.redhat.com:6443
OC_TOKEN	token for oc login (an alternative for oc-user & oc-pass)
OFFLINE_TOKEN	token used to fetch JWT tokens for assisted-service authentication (from https://cloud.redhat.com/openshift/token)
OPENSHIFT_VERSION	OpenShift version to install, default: "4.6"
PROXY	Set HTTP and HTTPS proxy with default proxy targets. The target is the default gateway in the network having the machine network CIDR
PULL_SECRET	pull secret to use for cluster installation command, no option to install cluster without it.
REMOTE_SERVICE_URL	URL to remote assisted-service - run infra on existing deployment
ROUTE53_SECRET	Amazon Route 53 secret to use for DNS domains registration.
SERVICE	assisted-service image to use
SERVICE_BASE_URL	update assisted-service config map SERVICE_BASE_URL parameter with given URL, including port and protocol
SERVICE_BRANCH	assisted-service branch to use, default: master
SERVICE_NAME	assisted-service target service name, default: assisted-service
SERVICE_REPO	assisted-service repository to use, default: https://github.com/openshift/assisted-service
SSH_PUB_KEY	SSH public key to use for image generation, gives option to SSH to VMs, default: ssh_key/key_pub
SSO_URL	URL used to fetch JWT tokens for assisted-service authentication
WORKER_MEMORY	memory for worker VM, default: 8892MB
PUBLIC_CONTAINER_REGISTRIES	comma-separated list of registries that do not require authentication for pulling assisted installer images
CHECK_CLUSTER_VERSION	If "True", the controller will wait for CVO to finish

Instructions

Host preparation

On the bare metal host:

Note: don't do it from /root folder - it will breaks build image mounts and fail to run

dnf install -y git make
cd /home/test
git clone https://github.com/openshift/assisted-test-infra.git

When using this infra for the first time on a host, run:

make create_full_environment

This will install required packages, configure libvirt, pull relevant Docker images, and start Minikube.

Usage

There are different options to use test-infra, which can be found in the makefile.

Full flow cases

The following is a list of stages that will be run:

Start Minikube if not started yet
Deploy services for assisted deployment on Minikube
Create cluster in assisted-service service
Download ISO image
Spawn required number of VMs from downloaded ISO with parameters that can be configured by OS environment (check makefile)
Wait until nodes are up and registered in assisted-service
Set nodes roles in assisted-service by matching VM names (worker/master)
Verify all nodes have required hardware to start installation
Install nodes
Download kubeconfig-noingress to build/kubeconfig
Waiting till nodes are in installed state, while verifying that they don't move to error state
Verifying cluster is in state installed
Download kubeconfig to build/kubeconfig

Note: Please make sure no previous cluster is running before running a new one (it will rewrite its build files).

Run full flow with install

To run the full flow, including installation:

make run_full_flow_with_install

Or to run it together with create_full_environment (requires sudo password):

make all

Run full flow without install

To run the flow without the installation stage:

make run_full_flow

Run full flow with ipv6

To run the flow with default IPv6 settings without install. It is identical to running

make run_full_flow IPv4=no IPv6=yes PROXY=yes VIP_DHCP_ALLOCATION=no

make run_full_flow_with_ipv6

Run only deploy nodes (without pre deploy of all assisted service)

make deploy_nodes or make deploy_nodes_with_install

Redeploy nodes

make redeploy_nodes or make redeploy_nodes_with_install

Redeploy with assisted services

make redeploy_all or make redeploy_all_with_install

Cleaning

Following sections show how to perform cleaning of test-infra environment.

Clean all include minikube

make destroy

Clean nodes only

make destroy_nodes

Delete all virsh resources

Sometimes you may need to delete all libvirt resources

make delete_all_virsh_resources

Install cluster

Install cluster after nodes were deployed. Can take ClusterId as OS environment

make install_cluster

Create cluster and download ISO

make download_iso

Deploy Assisted Service and Monitoring stack

make run
make deploy_monitoring

`deploy_assisted_service` and Create cluster and download ISO

make download_iso_for_remote_use

start_minikube and Deploy UI and open port forwarding on port 6008, allows to connect to it from browser

make deploy_ui

Kill all open port forwarding commands, will be part of destroy target

make kill_all_port_forwardings

Test `assisted-service` image

make redeploy_all SERVICE=<image to test>
or
export PULL_SECRET='<pull secret JSON>'; make redeploy_all_with_install SERVICE=<image to test>

Test agent image

make redeploy_all AGENT_DOCKER_IMAGE=<image to test>
or
make redeploy_all_with_install AGENT_DOCKER_IMAGE=<image to test>

Test installer image or controller image

make redeploy_all INSTALLER_IMAGE=<image to test> CONTROLLER_IMAGE=<image to test>
or
export PULL_SECRET='<pull secret JSON>'; make redeploy_all_with_install INSTALLER_IMAGE=<image to test> CONTROLLER_IMAGE=<image to test>

Test installer, controller, `assisted-service` and agent images in the same flow

make redeploy_all INSTALLER_IMAGE=<image to test> AGENT_DOCKER_IMAGE=<image to test> SERVICE=<image to test>
or
export PULL_SECRET='<pull secret JSON>'; make redeploy_all_with_install INSTALLER_IMAGE=<image to test> CONTROLLER_IMAGE=<image to test> AGENT_DOCKER_IMAGE=<image to test> SERVICE=<image to test>

Test infra image

Assisted-test-infra builds an image including all the prerequisites to handle this repository.

make image_build

In case you would like to build the image with a different `assisted-service` client

make image_build SERVICE=<assisted service image URL>

Test with Authentication

To test with Authentication, the following additional environment variables are required:

export ENABLE_AUTH=true
export OCM_CLIENT_ID=<SSO Service Account Name>
export OCM_CLIENT_SECRET=<SSO Service Account Password>
export OCM_BASE_URL=https://api.openshift.com
export OFFLINE_TOKEN=<User token from https://cloud.redhat.com/openshift/token>

UI is not available when Authentication is enabled.
The PULL_SECRET variable should be taken from the same Red Hat cloud environment as defined in OCM_URL (integration, stage or production).

Single Node - Bootstrap in place with Assisted Service

To test single node bootstrap in place flow with assisted service

export PULL_SECRET='<pull secret JSON>'
export OPENSHIFT_INSTALL_RELEASE_IMAGE=<relevant release image if needed>
export NUM_MASTERS=1
make redeploy_all_with_install or if service is up  make redeploy_nodes_with_install

Single Node - Bootstrap in place with Assisted Service and IPv6

To test single node bootstrap in place flow with assisted service and ipv6

export PULL_SECRET='<pull secret JSON>'
export OPENSHIFT_INSTALL_RELEASE_IMAGE=<relevant release image if needed>
export NUM_MASTERS=1
make run_full_flow IPv6=yes IPv4=no PROXY=yes VIP_DHCP_ALLOCATION=no

On-prem

To test on-prem in the e2e flow, two additonal environment variables need to be set:

export DEPLOY_TARGET=onprem
export ASSISTED_SERVICE_HOST=<fqdn-or-ip>

Setting DEPLOY_TARGET to "onprem" configures assisted-test-infra to deploy the assisted-service using a pod on your local host.

ASSISTED_SERVICE_HOST defines where the assisted-service will be deployed. For "onprem" deployments, set it to the FQDN or IP address of the host.

Optionally, you can also provide OPENSHIFT_INSTALL_RELEASE_IMAGE and PUBLIC_CONTAINER_REGISTRIES:

export OPENSHIFT_INSTALL_RELEASE_IMAGE=quay.io/openshift-release-dev/ocp-release:4.6.8-x86_64
export PUBLIC_CONTAINER_REGISTRIES=quay.io

If you do not export the optional variables, it will run with the default specified in assisted-service/onprem-environment.

Then run the same commands described in the instructions above to execute the test.

To run the full flow:

make all

To cleanup after the full flow:

make destroy

flyemsafe/assisted-test-infra