Acceptance Tests for the cloudscale.ch IaaS Platform

To ensure that our cloud platform continues to meet our quality standards over time, we use a set of acceptance tests to validate various aspects of our offering:

• Features work as documented.
• Response times meet our expectations.
• Regressions are avoided.

These tests are run regularly against our public infrastructure as well as our internal test environment where upgrades are staged prior to rollout.

Implemented Tests

Category	Test Name	Images
API	test_duplicate_headers	-
	test_invalid_duplicate_headers	-
	test_cors_headers	-
Custom Image	test_custom_image_with_slug	custom
	test_custom_image_with_uuid	custom
	test_custom_image_with_uefi	custom
Floating IP	test_floating_ip_connectivity	default
	test_multiple_floating_ips	default
	test_floating_ip_stability	default
	test_floating_ip_failover	default
	test_floating_ip_mass_failover	default
	test_floating_network	default
Load Balancer	test_simple_tcp_load_balancer	default
	test_load_balancer_end_to_end	default
	test_multiple_listeners	default
	test_multiple_listeners_multiple_pools	default
	test_balancing_algorithm_round_robin	default
	test_balancing_algorithm_source_ip	default
	test_balancing_algorithm_least_connections	default
	test_backend_health_monitors	default
	test_pool_member_change	default
	test_private_load_balancer_frontend	default
	test_floating_ip	default
	test_floating_ip_reassign	default
	test_frontend_allowed_cidr	default
	test_proxy_protocol	default
	test_ping	default
Private Network	test_private_ip_address_on_all_images	all
	test_private_network_connectivity_on_all_images	all
	test_multiple_private_network_interfaces	default
	test_no_private_network_port_security	default
	test_private_network_without_dhcp	default
	test_private_network_mtu	default
	test_private_network_only_on_all_images	all
	test_private_network_attach_later	default
	test_private_network_dhcp_dns_replies	default
Public Network	test_public_ip_address_on_all_images	all
	test_public_network_connectivity_on_all_images	all
	test_public_network_mtu	default
	test_public_network_port_security	default
	test_public_network_ipv4_only_on_all_images	all
	test_reverse_ptr_record_of_server	default
	test_reverse_ptr_record_of_floating_ip	default
Server	test_change_flavor_from_flex_to_flex	default
	test_change_flavor_from_flex_to_plus	default
	test_change_flavor_from_plus_to_flex	default
	test_change_flavor_from_plus_to_plus	default
	test_hostname	default
	test_rename_server	default
	test_reboot_server	default
	test_stop_and_start_server	default
	test_rename_server_group	default
	test_no_cpu_steal_on_plus_flavor	default
	test_random_number_generator	default
	test_metadata_on_all_images	all
Volume	test_attach_and_detach_volume_on_all_images	all
	test_expand_volume_online_on_all_images	all
	test_expand_filesystem_online_on_common_images	common
	test_expand_filesystem_on_boot_on_common_images	common
	test_maximum_number_of_volumes	default

Warning

⚠️ Running these tests yourself may incur unexpected costs and may result in data loss if run against a production account with live systems. Therefore, we strongly advise you to use a separate account for these tests.

Installation

ℹ︎ Note that you need at least Python 3.6.

To install the tests, you have to clone the repository:

git clone git@github.com:cloudscale-ch/acceptance-tests.git

Now, every time you want to run the tests in a new shell, use the following command first:

source acceptance-tests/pre-flight

You will be automatically switched to the acceptance-tests directory, ready to run the tests as outlined below.

Running Tests

To run all tests, run py.test as follows:

py.test .

Running Individual Tests

To only run a specific test, run py.test as follows:

py.test . -k <test-name>

Running Tests Against a Specific Image

By default, all tests are run against the default image, most tests are run against a set of common images, and some tests are run against all images provided by cloudscale.ch.

To run all tests against a specific image, use this image as the default:

py.test --default-image ubuntu-20.04 --default-image-only

Note that our default image is Debian 10. If you pick a different default image your results may differ.

Running Tests Against a Custom Image

Custom images can be used as the default image by specifying their slug, along with a username that can be used to connect via SSH. Note that custom images are less likely to pass all tests without prior modification, as the acceptance tests mainly focus on our common images.

py.test --default-image custom:alpine --default-image-only --username alpine

Running Tests Against a Specific Zone

By default, tests are run against a randomly selected zone.

Alternatively, you can specify the zone to run the tests against:

py.test --zone rma1
py.test --zone lpg1

Connect to Test Hosts

During test development, it can be useful to manually connect to hosts created by the tests. In this case it is necessary to explicitly specify your own SSH key, since tests connect to hosts using temporary SSH keys only:

py.test --ssh-key ~/.ssh/id_rsa.pub

Running a Test Multiple Times

Sometimes it is useful to run a specific test multiple times in a row:

py.test --count=10 test_floating_ip.py

Events Log

During execution, the acceptance tests generate a detailed log in the events directory (one file per test-run). Each line in such an event log is a structured JSON object.

Using a custom command, you can create a human-readable output of this log:

invoke pretty-print --file events/<file>

You can include filters as well:

invoke pretty-print --file events/<file> --regex outcome=failed

Or, during test execution, you can follow the log in a separate terminal window while it is being written. This will tail all the event logs that are currently being written. No need to specify a single file.

invoke tail

Cleanup

During normal operation, all resources created by the acceptance tests are automatically cleaned up. However, if the process receives a SIGKILL signal, or if it crashes, there may be resources left afterwards.

If you want to be sure, you can clean up all resources created by any acceptance test using the cleanup command:

invoke cleanup

All resources created by acceptance tests receive a unique tag, based on a securely hashed version of the API token, so using this command should be reasonably safe. However, we still strongly advise you to use a separate account for these tests as a precaution.

Developing New Tests

Create a New Branch

In order to review tests and to be able to develop multiple tests in parallel, they should be developed in a separate Git branch:

git branch <your_branch_name>

Writing Tests to be Run Against Specific Images

If you write a test with the image fixture, it will be called with the default image. This default image can be changed using the --default-image command line parameter.

If you want to ensure that a test runs against all common images, use the image fixture and include all_images in the name of your test:

def test_all_images_have_a_hosts_file(create_server, image):
    server = create_server(image=image)

If you use common_images in the name of your test, only common images will be tested:

def test_common_images_have_a_hosts_file(create_server, image):
    server = create_server(image=image)

Commit Your Test

git add <new_or_changed_files>
git commit

Push Your Branch and Create a Pull Request

git push origin <your_branch_name>

To create a pull request follow the link that will be displayed upon pushing a branch.