README ¶
Rook Test Framework
The Rook Test Framework is used to run end to end and integration tests on Rook. The framework depends on a running instance of Kubernetes.
The framework also provides scripts for starting Kubernetes using kubeadm
or minikube
so users can
quickly spin up a Kubernetes cluster.The Test framework is designed to install Rook, run tests, and uninstall Rook.
Requirements
- Docker version => 1.2 && < 17.0
- Ubuntu 16 (the framework has only been tested on this version)
- Kubernetes with kubectl configured
- Rook
Instructions
Setup
Install Kubernetes
You can choose any Kubernetes flavor of your choice. The test framework only depends on kubectl being configured. The framework also provides scripts to install Kubernetes. There are two scripts to start the cluster:
- Minikube (recommended for MacOS): Run minikube.sh to setup a single-node Minikube Kubernetes.
- Minikube v0.21.0 and higher is supported. Older minikube versions do not have cephfs or rbd tools installed.
- If using Kubernetes v1.8 or higher, then Minikube v0.23.0 or higher is required.
- Kubeadm (recommended for Ubuntu): run kubeadm.sh to setup a single-node K8s cluster using kubeadm
Minikube (recommended for MacOS)
Starting the cluster on Minikube is as simple as running:
tests/scripts/minikube.sh up
To copy Rook images generated from your local build into the Minikube VM, run the following commands after minikube.sh up
succeeded:
tests/scripts/minikube.sh update
tests/scripts/minikube.sh helm
Stopping the cluster and destroying the Minikube VM can be done with:
tests/scripts/minikube.sh clean
Kubeadm (recommended for Ubuntu)
Starting the cluster using kubeadm
is as simple as running:
tests/scripts/kubeadm.sh up
kubeadm.sh
starts Kubernetes on the host, so all the images generated by your local build should be available to the cluster without any additional commands.
Stopping the cluster can be done with:
tests/scripts/kubeadm.sh clean
Alternate Kubernetes Versions
These two scripts can install any version of Kubernetes you wish based on the KUBE_VERSION
environment variable.
To use an alternate version, simply set this variable before running the relevant up
command from above.
For example, if you wanted to use v1.9.6
, you would run export KUBE_VERSION=v1.9.6
first before running up
.
Install Helm
Use helm.sh to install Helm and set up Rook charts defined under _output/charts
(generated by build):
- To install and set up Helm charts for Rook run
tests/scripts/helm.sh up
. - To clean up
tests/scripts/helm.sh clean
.
NOTE: *kubeadm.sh
, minikube.sh
and helm.sh
scripts depend on some artifacts under the _output/
directory generated during build time,
these scripts should be run from project root. e.g., tests/script/kubeadm.sh up
.
NOTE: If Helm is not available in your PATH
, Helm will be downloaded to a temporary directory (/tmp/rook-tests-scripts-helm
) and used from that directory.
The temporary directory the Helm binary is in needs to be given to the integration
binary calls. The flag for that is --helm=HELM_BINARY
, e.g., --helm=/tmp/rook-tests-scripts-helm/linux-amd64/helm
.
Run Tests
From the project root do the following:
1. Build rook:
Run make build
2. Start Kubernetes
Using one of the following:
- Using Kubeadm
tests/scripts/kubeadm.sh up
tests/scripts/helm.sh up
- Using minikube
tests/scripts/minikube.sh up
tests/scripts/minikube.sh helm
tests/scripts/helm.sh up
3. Run integration tests:
Integration tests can be run using tests binary _output/tests/${platform}/integration
that is generated during build time e.g.:
~/integration -test.v
Test parameters
In addition to standard go tests parameters, the following custom parameters are available while running tests:
Parameter | Description | Possible values | Default |
---|---|---|---|
rook_platform | platform Rook needs to be installed on | kubernetes | kubernetes |
k8s_version | version of Kubernetes to be installed | v1.8+ | v1.8.5 |
rook_image | Rook image name to be installed | valid image name | rook/ceph |
skip_install_rook | skips installing Rook (if already installed) | true or false | false |
Running Tests with parameters.
To run all integration tests run
go test -v -timeout 1800s github.com/rook/rook/tests/integration
To run a specific suite (uses regex)
go test -v -timeout 1800s -run SmokeSuite github.com/rook/rook/tests/integration
To run specific tests inside a suite:
go test -v -timeout 1800s -run SmokeSuite github.com/rook/rook/tests/integration -testify.m TestRookClusterInstallation_SmokeTest
go test -v -timeout 1800s -run SmokeSuite github.com/rook/rook/tests/integration --skip_install_rook
If the skip_install_rook
flag is set to true, then Rook is not uninstalled either.
Run Longhaul Tests
Longhaul tests are integration tests that run for extended period of time. A load profile can be configured using the following load test flags
Parameter | Description | Possible values | Default |
---|---|---|---|
load_parallel_runs | performs concurrent operations | any number | 20 |
load_volumes | number of volumes | >1 | 1 |
load_time | number of seconds to run | >1 | 1800 |
load_size | size of load profile (3M, 10M, or 50M per thread) | small, medium, or large | medium |
enable_chaos | kill random pods in Rook cluster | true or false | false |
e.g.
go test -run TestObjectLongHaul github.com/rook/rook/tests/longhaul --load_parallel_runs=20 --load_time 1800 --load_size small --load_volumes 3
The longhaul test just like other test is going to install Rook if it's not already installed, but it is not going to clean up test data or uninstall Rook after the run. Longhaul test is designed to run multiple times on the same setup and installation of Rook to tests its stability. Test data and Rook should be cleaned up manually after the test.
You can measure memory, CPU, IOPS, throughput, and other settings on a cluster using Prometheus. The metrics collected during load test can be visualize using Grafana. Here a couple of helpful links to get prometheus and grafana started and collect metrics: kube-prometheus and cluster-deploy-script
Prerequisites:
- Go installed and GO_PATH set
- Dep installed
- When running tests locally, make sure
kubectl
is accessible globally in yourPATH
as the test framework useskubectl