netcat relay services in test suite slow and complicated

[JPvRiel]

The current test suite relies on two containers running netcat and using a shared docker volume where the behave container searches for the logs in the volume

Due to disc IO caching writes, etc, a time delay between sending and event and searching for it is needed. And this might be unreliable. A more reliable solution would be to have behave itself listen on ports and receive network events directly. Unfortunately, only v1.2.6 has support for async patterns such as opening a network socket in the background, sending a network event, and then waiting on the reply.

When 1.2.6 becomes latest/stable, this improvement can be made.

[JPvRiel]

To clarify a test case can fail when it should not. E.g. when run once, this scenario fails, but running a 2nd time succeeds. Added 'over "5" seconds' to tests to allow more time for IO to sync in the background as an ugly workaround.

Scenario: Messages are forwarded to another server in syslog format               # behave/features/syslog_forwarding.feature:19
    Given a valid rsyslog configuration                                             # behave/features/steps/step_background.py:10 0.000s
    And a server "test_syslog_server"                                               # behave/features/steps/step_background.py:30 0.000s
    And an environment variable file "test_syslog_server.env"                       # behave/features/steps/step_background.py:47 0.000s
    Given "rsyslog_omfwd_syslog_enabled" environment variable is "on"               # behave/features/steps/step_syslog_forwarding.py:13 0.000s
    And "rsyslog_omfwd_syslog_host" environment variable is set                     # behave/features/steps/step_syslog_forwarding.py:20 0.000s
    And "rsyslog_omfwd_syslog_port" environment variable is set                     # behave/features/steps/step_syslog_forwarding.py:20 0.000s
    And a file "/tmp/syslog_relay/nc.out"                                           # behave/features/steps/step_syslog_messages.py:10 0.000s
    When sending the syslog message "Testing syslog forwarding" in "RFC3164" format # behave/features/steps/step_syslog_forwarding.py:26 0.001s
    And waiting "1" seconds                                                         # behave/features/steps/step_syslog_forwarding.py:79 1.001s
    And searching lines for the pattern "Testing syslog forwarding"                 # behave/features/steps/step_syslog_messages.py:45 0.001s
    Then the pattern should be found                                                # behave/features/steps/step_syslog_messages.py:81 0.000s

[JPvRiel]

Attempting work around with FIFO between containers, but the blocking read() on the FIFO in python is cumbersome to handle - so using a FIFO is probably as might be almost as much effort as doing a workaround for async socket input.

[JPvRiel]

To confirm the "quick fix" of having two netcat containers act as relay recipients and write to a shared docker volume has timing issues, and even waiting up to 15 seconds can sometimes fail (falsely) . Would be better to have the sut behave test suite start a TCP server and act as the relay recipient itself, but Python asyncio looks quite complex to implement.