k3s/tests/e2e
Manuel Buil 487d08ad11 Add a netpol test for podSelector & ingress type
Signed-off-by: Manuel Buil <mbuil@suse.com>
2022-10-26 17:52:28 +02:00
..
amd64_resource_files Add a netpol test for podSelector & ingress type 2022-10-26 17:52:28 +02:00
cis_amd64_resource_files Add scripts to run e2e test using ansible (#5134) 2022-08-05 09:16:10 -07:00
clusterreset Add journalctl logs to E2E tests 2022-10-06 12:27:44 -07:00
docker Add journalctl logs to E2E tests 2022-10-06 12:27:44 -07:00
dualstack Add a netpol test for podSelector & ingress type 2022-10-26 17:52:28 +02:00
scripts Add the gateway parameter in netplan 2022-10-18 13:14:10 +02:00
secretsencryption Make all E2E tests ordered 2022-09-13 14:17:39 -07:00
splitserver Add journalctl logs to E2E tests 2022-10-06 12:27:44 -07:00
upgradecluster Add journalctl logs to E2E tests 2022-10-06 12:27:44 -07:00
validatecluster Add journalctl logs to E2E tests 2022-10-06 12:27:44 -07:00
README.md Add info on libvirt ubuntu 22.04 (#6316) 2022-10-26 08:40:02 -07:00
e2e_test_playbook.yaml Add scripts to run e2e test using ansible (#5134) 2022-08-05 09:16:10 -07:00
testutils.go Replace deprecated ioutil package (#6230) 2022-10-07 17:36:57 -07:00
vagrantdefaults.rb Expand startup integration test (#6030) 2022-08-31 15:40:17 -07:00

README.md

End-to-End (E2E) Tests

E2E tests cover multi-node K3s configuration and administration: bringup, update, teardown etc. across a wide range of operating systems. E2E tests are run nightly as part of K3s quality assurance (QA).

Framework

End-to-end tests utilize Ginkgo and Gomega like the integration tests, but rely on Vagrant to provide the underlying cluster configuration.

Currently tested operating systems are:

Format

All E2E tests should be placed under tests/e2e/<TEST_NAME>.
All E2E test functions should be named: Test_E2E<TEST_NAME>.
A E2E test consists of two parts:

  1. Vagrantfile: a vagrant file which describes and configures the VMs upon which the cluster and test will run
  2. <TEST_NAME>.go: A go test file which calls vagrant up and controls the actual testing of the cluster

See the validate cluster test as an example.

Setup

To run the E2E tests, you must first install the following:

  • Vagrant
  • Libvirt
  • Vagrant plugins

Vagrant

Download the latest version (currently 2.2.19) of Vagrant from the website. Do not use built-in packages, they often old or do not include the required ruby library extensions necessary to get certain plugins working.

Libvirt

Follow the OS specific guides to install libvirt/qemu on your host:

  • openSUSE
  • ubuntu 20.04
  • ubuntu 22.04:
    sudo apt install ruby-libvirt qemu libvirt-daemon-system libvirt-clients ebtables dnsmasq-base libxslt-dev libxml2-dev libvirt-dev zlib1g-dev ruby-dev libguestfs-tools
    
  • debian
  • fedora

Vagrant plugins

Install the necessary vagrant plugins with the following command:

vagrant plugin install vagrant-libvirt vagrant-scp vagrant-k3s vagrant-reload

Running

Generally, E2E tests are run as a nightly Jenkins job for QA. They can still be run locally but additional setup may be required. By default, all E2E tests are designed with libvirt as the underlying VM provider. Instructions for installing libvirt and its associated vagrant plugin, vagrant-libvirt can be found here. VirtualBox is also supported as a backup VM provider.

Once setup is complete, all E2E tests can be run with:

go test -timeout=15m ./tests/e2e/... -run E2E

Tests can be run individually with:

go test -timeout=15m ./tests/e2e/validatecluster/... -run E2E
#or
go test -timeout=15m ./tests/e2e/... -run E2EClusterValidation

Additionally, to generate junit reporting for the tests, the Ginkgo CLI is used. Installation instructions can be found here.

To run the all E2E tests and generate JUnit testing reports:

ginkgo --junit-report=result.xml ./tests/e2e/...

Note: The go test default timeout is 10 minutes, thus the -timeout flag should be used. The ginkgo default timeout is 1 hour, no timeout flag is needed.

Debugging

In the event of a test failure, the cluster and VMs are retained in their broken state. Startup logs are retained in vagrant.log.
To see a list of nodes: vagrant status
To ssh into a node: vagrant ssh <NODE>
Once you are done/ready to restart the test, use vagrant destroy -f to remove the broken cluster.