k3s/tests/e2e

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:

• Ubuntu 20.04
• Leap 15.3 (stand-in for SLE-Server)
• MicroOS (stand-in for SLE-Micro)

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.

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.