[Fleet] Compatibility checks

[mostlyjason]

Summary of the problem

Fleet needs a way to check for the compatibility of agents and integrations when user are adding or updating them. This prevents users from making unintended mistakes or violating system constraints.

User stories

• As an administrator using Ingest Manager, I would like to see a warning when Agents do not meet the condition on Agent version specified in each integration.
• As an administrator using Ingest Manager, I should be prevented from adding integrations to an agent configuration that do not meet condition specified on the integration, such as only one endpoint security integration per agent configuration, or integrations that are not compatible with the hosted agent.
• As an administrator using Ingest Manager, I would like to be able to specify conditions on when the stream is enabled based on the value of a local system property, such as the operating system type. I would also like to get default paths based on my operating system.

Other

• I may also have a prompt to upgrade my Agent version if its not compatible and there is a newer version [Ingest Management] Add ability to upgrade agent binary in UI #68310
• Related document on input templates which define the type of variables that conditions can be based on.
• Requirements for hosted agents are here https://github.com/elastic/integrations-dev/issues/99
• Other compatibility checks [Ingest Manager] Hosted Agents cannot be upgraded, unenrolled or reassigned to a different config #76843 [Elastic Agent] Disable upgrade capability of Agent beats#21001 [Fleet] Indicate that Agent upgrades cannot be managed in Fleet #76842

[ph]

@hbharding @mostlyjason I am looking at this, do we have some existing design that could cover the user stories?

[mostlyjason]

I think a lot of these use cases could be covered by status reporting on the agent list and agent details page. The user will see an unhealthy status if the agent version is not compatible with an integration. We discussed doing the same for hosted agents here #76841. We haven't defined status reporting for input conditions yet, but I imagine that would also be on the agent details page. This will allow the user to identify and fix any problems.

It'd be a better UX to front-load communicating this information where possible rather than waiting for an unhealthy agent. It will help the user prevent making incompatible changes and reduce the time to fix it. That's an enhancement we could implement later though.

One that's a little different is the restriction that only one endpoint security integration can be in an agent policy. We've already implemented that by hiding it in the add integration list.

[mostlyjason]

CLosing this issue since its a mishmash of separate features. Lets open separate issues when we are ready to prioritize them.