Improve the experience when docker isn't running

[davidfowl]

Description

• When the container runtime isn't running, log a critical error but continue running and sending requests to dcp. DCP will wait for the container runtime to be healthy (started) and will run the containers.

Fixes #5533

Checklist

• Is this feature complete?
  • Yes. Ready to ship.
  • No. Follow-up changes expected.
• Are you including unit tests for the changes and scenario tests if relevant?
  • Yes
  • No
• Did you add public API?
  • Yes
    • If yes, did you have an API Review for it?
      • Yes
      • No
    • Did you add <remarks /> and <code /> elements on your triple slash comments?
      • Yes
      • No
  • No
• Does the change make any security assumptions or guarantees?
  • Yes
    • If yes, have you done a threat model and had a security review?
      • Yes
      • No
  • No
• Does the change require an update in our Aspire docs?
  • Yes
    • Link to aspire-docs issue:
  • No

Microsoft Reviewers: Open in CodeFlow

[davidfowl]

When you launch the app host, this is what it says:

This is what the dashboard looks like when the container runtime isn't running.

@danegsta I think dcp should set the state of these containers to something that reflects that. I can do it in the apphost but it feels like the source of truth for what the container state is should be dcp. We could use a new state "Error" or "Container runtime unhealthy" etc.

We show logs in each of the containers that show the error message from DCP:

Here's what it looks like when you run docker:

start-docker-2.mp4

[danegsta]

First thought is that we could add something like MissingRequirement as the new state and a status property for the specific related error to make it easier to surface in the Dashboard.

[davidfowl]

First thought is that we could add something like MissingRequirement as the new state and a status property for the specific related error to make it easier to surface in the Dashboard.

I think we should do something like that. But we should make it specific to the actual state instead of a generic MissingRequirement state (unless there's some other way to show the specific reason).

[mitchdenny]

I think that @danegsta is right about having a MissingRequirements state. The reason is that this PR will need to include an update to KnownResourceStates.TerminalStates to account for getting into this state. This is used by WaitForCompletion(...) (ultimately).

Spoke to @davidfowl behind the scenes on a bit more of the intent here. We don't want DCP to set a terminal state because we want folks to be able to start their Docker instance behind the scenes and have it suddenly come to life without having to restart the apphost.

Given this I'm happy with this PR.

[JamesNK]

This is what I see in the UI:

And resources page:

And this is in the console logs:

Overall I like it.

Suggestions for improvement:

1. There is duplication between error messages printed to app host console and resource console. Maybe share code? I like that the app host console includes the name of the runtime, and I like that the console logs includes the error message from DCP. One set of code that does both?

2. Putting myself in the shoes of an Aspire beginner, I don't know what unhealthy means. Do I need to reinstall Docker? We could look at the error message from DCP and if there is matching content, log in the UI that the container runtime isn't running.

   Get "http://%2F%2F.%2Fpipe%2FdockerDesktopLinuxEngine/v1.47/containers/json?limit=1";: open //./pipe/dockerDesktopLinuxEngine: The system cannot find the file specified.

   Look for "open //./pipe/dockerDesktopLinuxEngine"? It might be better than "The system cannot find the file specified." which could be localized.
   Would need to do something similar for podman which no doubt has a different address.

   If this is best done with DCP changes, a simple improvement for 9.0 could to always suggest checking the container runtime is running if it's unhealthy.

3. Unknown should be a more specific error state and have an error icon. I think that needs DCP changes to work smoothy so it could be a 9.x improvement.

[davidfowl]

I moved the code to a shared location, and I improved the error message as much as I could (using chatgpt 😄 )

[maddymontaquila]

yessssss 🔥

[davidfowl]

/backport to release/9.0-rc1

[github-actions]

Started backporting to release/9.0-rc1: https://github.com/dotnet/aspire/actions/runs/11222760501