libvirt/docs/testtck.rst
Peter Krempa 5c1ba38913 docs: Use relative links within the web page
Replace full/external links which point to content within
'https://libvirt.org/' with relative links so that the web page works
fully locally.

This does not change the links in 'docs/manpages' as we want the
installed man page to work from everywhere (even when the local docs are
not installed) and the generated API docs which take links from the C
source.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2024-10-09 16:00:44 +02:00

6.7 KiB

libvirt TCK : Technology Compatibility Kit

The libvirt TCK provides a framework for performing testing of the integration between libvirt drivers, the underlying virt hypervisor technology, related operating system services and system configuration. The idea (and name) is motivated by the Java TCK.

In particular the libvirt TCK is intended to address the following scenarios:

  • Validate that a new libvirt driver is in compliance with the (possibly undocumented!) driver API semantics
  • Validate that an update to an existing driver does not change the API semantics in a non-compliant manner
  • Validate that a new hypervisor release is still providing compatibility with the corresponding libvirt driver usage
  • Validate that an OS distro deployment consisting of a hypervisor and libvirt release is configured correctly

Thus the libvirt TCK will allow developers, administrators and users to determine the level of compatibility of their platform, and evaluate whether it will meet their needs, and get awareness of any regressions that may have occurred since a previous test run.

Libvirt-TCK is maintained using a GIT repository. GitLab is also the place where the whole TCK development workflow (issues, merge requests, comments) happens.

Using TCK

TCK can be used independently of the environment, i.e. both on your local host or in a VM. We strongly recommend using a VM for the tests as TCK might affect your current host setup, see Running TCK.

Installing dependencies

Since TCK is based on libvirt Perl bindings, you'll need to have the proper version of the bindings installed for the version of libvirt you wish to test in order to be able execute the TCK test suite successfully. Additionally, a number of Perl dependencies will need to be installed as well, some will be available through the system package manager and some will likely need to be installed from CPAN (Perl's equivalent of Python's PyPI). Here's where libvirt-ci's lcitool can help with preparing a test environment in a fresh VM, taking care of the dependencies along the way. A simple example of getting a machine from lcitool would be:

$ lcitool install --target fedora-38 tck-fedora38 --wait

would get you a new Fedora 38 VM named tck-fedora38. There are different ways of getting a fresh local VM with lcitool, so please refer to Installing local VMs for further details, especially to utilize vendor cloud images for this purpose.

Once you have a fresh virtual machine, you need to pre-install it with all necessary build dependencies to be able to build libvirt, libvirt Perl bindings and run the TCK test suite inside it. You'd do that by running

$ lcitool update tck-fedora38 libvirt,libvirt-perl,libvirt-tck+runtime

Again, for further details on how to update lcitool virtual machines, please refer to Updating VMs with a given project dependencies

Note that lcitool only installs build dependencies, so as mentioned above you'll need both libvirt and libvirt Perl bindings installed in order to be able to run TCK. You can (depending on use case) either build both inside the VM and install manually or install the corresponding RPMs from GitLab CI build artifacts.

We also recommend executing TCK using the Avocado framework as the test harness engine which means that you'll have to install Avocado in the test environment as well. You can get it either from PyPI (recommended), or if you're on Fedora you can make use of the Avocado module. Using Avocado is not mandatory for the time being and you can skip it, but in the future we plan on making the TCK internal coupling with Avocado tighter.

Running TCK

Once you have all the dependencies installed, you can then proceed with either of the following procedures to execute the test suite as root.

Replicating upstream CI test suite execution locally

Similarly to how local container builds utilize the standalone ci/jobs.sh script containing functions describing GitLab job definitions it can be utilized to run integration test suite as well. In this case, one needs to get a copy of their libvirt repository containing the changes to be tested inside the VM (either by cloning it manually or sharing the repo e.g. via virtiofs). Make sure that the user which is going to execute the following has passwordless "sudo" permissions (lcitool's default "test" user does). Then it's just a matter of running

$ source ci/jobs.sh
$ run_integration

Manual invocation

If you want to have more control over the whole procedure or simply don't want to run the exact same steps as libvirt's upstream CI pipeline does in context of integration tests then start by cloning the TCK repository and run

# avocado --config avocado.config run

from the TCK's git root.

If you don't want to install Avocado you can execute tests using the libvirt-tck binary directly (again, from the git root). You'll need to pass a few options that Avocado takes care of:

# PERL5LIB=./lib perl bin/libvirt-tck -c <path_to_config> --force ./scripts

Running with the --force argument is not necessary and you can safely omit it, but it becomes useful if you need to interrupt a test run for some reason. In such case using --force ensures the first thing TCK does before running any tests is that it will clean up all resources from the previous test run which may have been left behind if you had interrupted the previous TCK's execution.

Note that running with root privileges is necessary since some tests need access to system resources or configs. This, along with the fact that some tests might affect the host system are good reasons to consider using a test VM as described above.

Contributing a test

We'd appreciate if you provided a functional test case whenever you're adding a new feature or fixing a bug in libvirt with the only complication being that in case you're adding a new public API then a Perl binding will have to be introduced first. After that, the best way to start is looking at some existing tests, copy-pasting one that fits your scenario the best and tweak the remaining bits.