Update repository tests with supported checksums

[vsedmik]

Problem Statement

SHA1 is no longer supported by Pulp (see https://projects.theforeman.org/issues/37522) which renders even more tests failing in validation for sha1 in stream builds:

WARNING  nailgun.client:client.py:126 Received HTTP 422 response: {"displayMessage":"Validation failed: Checksum type is not included in the list","errors":{"checksum_type":["is not included in the list"]}}

Solution

Define the supported checksums in constants and use them.

PRT test Cases example

trigger: test-robottelo
pytest: tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum

[vsedmik]

trigger: test-robottelo
pytest: tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum

[Satellite-QE]

PRT Result

Build Number: 7598
Build Status: UNSTABLE
PRT Comment: pytest tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum --external-logging
Test Result : ==== 5 failed, 18 passed, 281 deselected, 374 warnings in 750.33s (0:12:30) ====

[dosas]

@vsedmik If there are any of the deb parametrizations failing, this PR should fix that.

[dosas]

Thanks for fixing this, I came across this problem some time ago but did not know what the 'correct' checksums are.

[vsedmik]

Hey @dosas, it is probably not deb-related, but thank you for the info and nailgun PR!

It seems to fail here with HTTP 500, but locally this passed against stream snap 64 and 63

tests/foreman/api/test_repository.py .............          [ 56%]
tests/foreman/cli/test_repository.py ..........             [100%]

=== 23 passed, 281 deselected, 28 warnings in 154.86s (0:02:34) ===

In WebUI the default bahaviour is checksum reset to Default when download policy is changed to on_demand, there is even notification about that, so I supposed same behaviour was added for API since there were some (flaky?!) failures like Failed: DID NOT RAISE <class 'requests.exceptions.HTTPError'> for that original pytest.raises

[vsedmik]

trigger: test-robottelo
pytest: tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum

[Satellite-QE]

PRT Result

Build Number: 7604
Build Status: SUCCESS
PRT Comment: pytest tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum --external-logging
Test Result : ========= 23 passed, 281 deselected, 412 warnings in 787.26s (0:13:07) =========

[vsedmik]

So the explanation for the flaky failures in CI is that after repo creation another task to update the publication is spawned asynchronously (metadata generate) and we need to let it complete before we try to update, otherwise we get 500

2024-07-02 10:07:17 - nailgun.client - WARNING - Received HTTP 500 response: {"displayMessage":"Task 6f2da32e-3668-441d-9489-52b9e2bfb940: RuntimeError: The repository's publication is missing. Please run a 'complete sync' on mPZsjMJQV.","errors":["Task 6f2da32e-3668-441d-9489-52b9e2bfb940: RuntimeError: The repository's publication is missing. Please run a 'complete sync' on mPZsjMJQV."]}

Adding wait_for to the repo create fixture seems to fix the issue.

[vsedmik]

Resolved conflict and rebased.

[vsedmik]

trigger: test-robottelo
pytest: tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum

[Satellite-QE]

PRT Result

Build Number: 7669
Build Status: SUCCESS
PRT Comment: pytest tests/foreman/api/test_repository.py tests/foreman/cli/test_repository.py -k checksum --external-logging
Test Result : ========= 23 passed, 283 deselected, 414 warnings in 750.94s (0:12:30) =========