Tests and images for testing Rocky with openQA
Go to file
Adam Williamson 33ac181955 Use mirrorlist instead of baseurl for updates tests
The reason we have all this horrible code to use the commented-
out baseurl lines in the repo files instead of the metalinks
that are usually used is a timing issue with the metalink
system. As a protection against stale mirrors, the metalink
system sends the package manager a list of mirrors *and a list
of recent checksums for the repo metadata*. The package manager
goes out and gets the metadata from the first mirror on the
list, then checksums it; if the checksum isn't on the list of
checksums it got from mirrormanager, it assumes that means the
mirror is stale, and tries the next on the list instead.

The problem is that MM's list of checksums is currently only
updated once an hour (by a cron job). So we kept running into
a problem where, when a test ran just after one of the repos
had been regenerated, the infra mirror it's supposed to use
would be rejected because the checksum wasn't on the list - but
not because the mirror was stale, but because it was too fresh,
it had got the new packages and metadata but mirrormanager's
list of checksums hadn't been updated to include the checksum
for the latest metadata.

All this baseurl munging code was getting ridiculous, though,
what with the tests getting more complicated and errors showing
up in the actual repo files and stuff. It occurred to me that
instead of using the baseurl we can just use the 'mirrorlist'
system instead of 'metalink'. mirrorlist is the dumber, older
system which just provides the package manager a list of mirrors
and nothing else - the whole stale-mirror-detection-checksum
thing does not happen with mirrorlists, the package manager just
tries all the mirrors in order and uses the first that works.
And happily, it's very easy to convert the metalink URLs into
mirrorlist URLs, and it saves all that faffing around trying to
fix up baseurls.

Also, adjust upgrade_boot to do the s/metalink/mirrorlist/
substitution, so upgrade tests don't run into the timing issue
in the steps before the main repo_setup run is done by
upgrade_run, and adjust repo_setup_compose to sub this line out
later.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2018-05-18 16:41:13 -07:00
lib Use mirrorlist instead of baseurl for updates tests 2018-05-18 16:41:13 -07:00
needles New needles encrypted tests, blue background bypass brc#1551662 2018-05-16 12:55:29 +02:00
tests Use mirrorlist instead of baseurl for updates tests 2018-05-18 16:41:13 -07:00
.arcconfig Update Phab project name 2017-02-14 15:38:06 -08:00
.gitignore .gitignore 2015-02-03 14:00:53 +01:00
COPYING Decoupled tools from tests 2015-01-26 14:43:01 +01:00
main.pm More tweaking for this damn no root password spoke situation 2018-03-08 20:31:14 -08:00
README.md One more Phab -> Pagure change (dangit) 2017-09-04 11:46:51 -07:00
templates Use virtio-rng for all tests 2018-05-03 13:12:34 -07:00
templates-updates Add Workstation upgrade test to update tests 2018-04-27 17:10:53 -07:00
VARIABLES.md More tweaking for this damn no root password spoke situation 2018-03-08 20:31:14 -08:00

openQA tests for the Fedora distribution

This repository contains tests and images for testing Fedora with openQA. The fedora_openqa library and CLI are used for scheduling tests, and createhdds is used for creating base disk images for the test. For openQA installation instructions, see the Fedora openQA wiki page.

Issues

Issues and pull requests are tracked in os-autoinst-distri-fedora Pagure. Pagure uses a Github-like pull request workflow, so if you're familiar with that, you can easily submit Pagure pull requests. If not, you can read up in the Pagure documentation.

Note that this repository does not use the 'gitflow' system, so the main development branch is master: please branch from master and submit diffs against it. This is not a Python repository and has no tests or linting.

Test development

See official documentation on:

See this example repo on how tests should be structured.

main.pm modular architecture

Since openQA uses only one entrypoint for all tests (main.pm), we have decided to utilize this feature and make tests modular. It means that basic passing through main.pm (without any variables set) results in most basic installation test executed. Developer can customize it with additional variables (for example by setting PACKAGE_SET=minimal to do installation only with minimal package set).

Make your test modular, so that it utilizes _boot_to_anaconda.pm, _software_selection.pm and _do_install_and_reboot.pm tests (that are loaded automatically). Break your test into smaller parts, each dealing with one specific feature (e. g. partitioning, user creation...) and add their loading into main.pm based on reasonable variable setting (so they can be used in other tests also).

Fedora installation (and consequently main.pm) consists of several parts:

Booting into Anaconda or booting live image and starting Anaconda

Since there isn't much variation between tests in this step, we have developed universal _boot_to_anaconda.pm test that is loaded automatically each time except when ENTRYPOINT or UPGRADE is set (see VARIABLES.md).

To customize this step, you can set following variables:

  • GRUB is appended to kernel line before boot. You can set for example inst.updates here.
  • If KICKSTART is set, this part of installation ends here (program doesn't wait for Anaconda to appear). Note that you should set inst.ks yourself by setting GRUB variable.
  • If LIVE is set, program waits for desktop to appear and then clicks on "Install to Hard Drive" button.

Customizing installation by interacting with Anaconda spokes

Most of the differences between tests take place in this part. If you want to add another installation test, you will probably put your variable checking and test loading here. All tests in this part should start on Anaconda's main hub and after they done its part, they should go back to Anaconda's main hub so that next test could be executed. In this phase, universal _software_selection.pm test is loaded that handles selecting what software to install.

To customize this step, you can set following variables:

  • Set PACKAGE_SET to install required package set on "Software selection spoke" - you have to provide correct needles with the name of anaconda_${PACKAGE_SET}_highlighted and anaconda_${PACKAGE_SET}_selected.
  • Set ENCRYPT_PASSWORD to encrypt disk, value of this variable is used as an actual password.

Installing Fedora and waiting for Fedora to reboot

After all customizations are finished, _do_install_and_reboot.pm test is automatically loaded. It starts installation, creates user and sets root password when required, waits for installation to finish and reboots into installed system. Only variables that control flow in this part are these:

  • ROOT_PASSWORD to set root password to this value.
  • When set, USER_LOGIN and USER_PASSWORD are used to create user in Anaconda.

Post-install phase

After installation is finished and installed system is fully booted, you can run additional tests as checks that installed system has correct attributes - that correct file system is used, that RAID is used etc.

Test inheritance

Your test can inherit from basetest, installedtest or anacondatest. Each provides relevant methods that are documented in-line, so read the files (lib/anacondatest.pm, lib/installedtest.pm) for information on these.

  • basetest: A base class provided by os-autoinst - it has empty post_fail_hook() and doesn't set any flags.
  • anacondatest: should be used in tests where Anaconda is running. It uploads Anaconda logs (for example anaconda.log or packaging.log) in post_fail_hook().
  • installedtest: should be used in tests that are running on installed system (either in postinstall phase or in upgrade tests).

There are also several modules that export utility functions, currently utils, anaconda, freeipa, packagetest and tapnet. Your test can use any of these modules and then directly call the functions they export. Again, the functions are documented in-line.

New test development workflow

  1. Put each part of your test as a separate file into tests/ directory, reimplementing run() method and test_flags() method, inheriting from one of the classes mentioned above.
  2. Set correct variables (so that all test parts you have made are executed) in WebUI -> Test suites.
  3. Link your newly created Test suite to medium type in WebUI -> Job groups.
  4. Run test (see openqa_fedora_tools repository).
  5. Create needles (images) by using interactive mode and needles editor in WebUI.
  6. Add new Job template and Test suite into templates file (and templates-updates, if the test is applicable to the update testing workflow)
  7. Add new Test suite and Test case into conf_test_suites.py file in fedora_openqa repository.
  8. Open pull request for the os-autoinst-distri-fedora changes in Pagure. Pagure uses a Github-style workflow (summary: fork the project via the web interface, push your changes to a branch on your fork, then use the web interface to submit a pull request). See the Pagure documentation for more details.
  9. Open a pull request in fedora_openqa Pagure for any necessary fedora_openqa changes.

Language handling

Tests can run in different languages. To set the language which will be used for a test, set the LANGUAGE variable for the test suite. The results of this will be:

  1. The value set will be typed into the language search box in anaconda.
  2. Any needle with at least one tag that starts with LANGUAGE will be unregistered unless it has the tag LANGUAGE-(LANGUAGE) (where (LANGUAGE) is the value set, forced to upper-case).
  3. As a consequence, the chosen language will be selected at the anaconda Welcome screen.

It is very important, therefore, that needles have the correct tags. Any needle which is expected to match for tests run in any language must have no LANGUAGE tags. Other needles must have the appropriate tag(s) for the languages they are expected to match. The safest option if you are unsure is to set no LANGUAGE tag(s). The only danger of this is that missing translations may not be caught.

Note that tags of the form ENV-INSTLANG-(anything) are useless artefacts and should be removed.