[Elastic Agent] Add e2e test for Agent to be installed on top of Filebeat & Metricbeat docker containers

[EricDavisX]

This comes from Jason / Ruflin who reported some problems seen somewhere and we want to have a small automated test in place to make sure it always works, at least on a basic level.

I'll add this ticket to the main e2e testing issue for tracking.

Output of ticket:
elastic/e2e-testing repo has an Ingest Manager / Agent oriented test that proves a Filebeat (and Metricbeat) container can have Fleet Agent installed on it and both sets of processes will stream data to ES.

[elasticmachine]

Pinging @elastic/ingest-management (Team:Ingest Management)

[mdelapenya]

I think it's a matter of abstracting one of the existing scenarios, converting it into:

Original:

Scenario: Deploying an agent
  When an agent is deployed to Fleet
  Then the agent is listed in Fleet as online
    And system package dashboards are listed in Fleet

New:

Scenario Outline: Deploying an agent
  When a <baseImage> agent is deployed to Fleet
  Then the agent is listed in Fleet as online
    And system package dashboards are listed in Fleet
Examples:
| baseImage |
| centos |
| metricbeat |
| filebeat |

Then the implementation code would read that variable to spin up the proper container. Things to consider: what versions of the beats do we want to use?

[EricDavisX]

That may be a good way to start and proceed. I think we'll need to write explicit test validation steps however, to confirm that the Beats are functioning when env comes up (show me the Beats data in ES) then install Agent, and then confirm Agent as well as Beats data flow (again).

[EricDavisX]

If we could have the prior GA version of Beats that might be best - as it would closely mimic an upgrade scenario for a customer going from Beats to Agent usage. But maybe we'd want to run the simple scenario twice...with current-minus-1 GA version and current snapshot version. we'll have to chat the way we'll coordinate / manage running some portions of tests multiple times. very exciting.

[mostlyjason]

For reference, here is the issue that kicked off this thread elastic/beats#19369

[EricDavisX]

interesting enough, we have the start of some validation for this in the Ingest Manager / Demo test server, it has beats installed on a given host and then we install the Agent on top and run the Agent Beats. We do not have any data validation or 'usage' on top of it, however - if the installs succeed that is about all we know. It would still be great to have this coverage formalized or validated further in either scenario.

[cachedout]

This issue will be resolve once this PR is merged: #1269

[EricDavisX]

@cachedout - hi - I know the 2 tasks and am not sure if this is solved by 1269 as noted, did you mean to link to another?

@mdelapenya I am fine to close this out and open a new one to host specifics of what we want to validate. Their side by side existence is a good start. We'd ideally want to confirm the separate data being indexed.

I've logged that ticket as:
#1292