Failed test_integrity_sync System test suit on 4.8.0 Beta 2

[Deblintrake09]

Description

During the testing in Release 4.8.0 - Beta 2 - System tests, an error has been encountered in the agentless_cluster environment, specifically in this test: test_cluster/test_integrity_sync/test_integrity_sync.py.

The test was executed in a Ubuntu 22.04.2 LTS T3.Large 30GB SSD instance, provided by the devops team.

Current behavior

All test cases failed with different error types

Evidence Report: agentless_cluster.zip

Expected results

All the tests should have passed

[rauldpm]

>                   assert worker_stats['mode'] == '0660', f"{file} permissions were expected to be '660' in {host}, but " \
                                                           f"they are {worker_stats['mode']}."
E                                                          KeyError: 'mode'

>               assert result == file_test_content_master, f"File {file} inside {host} should contain " \
                                                           f"{file_test_content_master}, but it has: {result}"
E               AssertionError: File /var/ossec/etc/lists/test_file inside wazuh-worker1 should contain test_content_from_master, but it has: 
E               assert '' == 'test_content_from_master'
E                 - test_content_from_master

>               assert result == '', f"File {file} was expected to be removed from {host}, but it still exists."
E               AssertionError: File /var/ossec/etc/lists/test_file was expected to be removed from wazuh-worker1, but it still exists.
E               assert '/var/ossec/e...sts/test_file' == ''
E                 + /var/ossec/etc/lists/test_file

E                   TimeoutError: Did not found the expected callback in wazuh-master: .*File too large to be synced: /var/ossec/etc/rules/test_file_too_big

[juliamagan]

Based on the errors reported, and the report itself, it has been concluded that the errors are due to the tests being launched separately and not in the expected order. It is already known that system tests right now are very dependent, so they have to be launched in a certain way so that they do not fail. We can check that all tests pass as expected in
agentless_cluster.zip.

As a proposal, I will add a small comment in the documentation of the system tests indicating how they should be launched.

[rafabailon]

LGTM

[davidjiglesias]

LGTM