test(hosting): fix Outerloop tests asserting null proxyless EffectivePort

[Ankit Jain (radical)]

The Outerloop Tests workflow fails every scheduled night (7/7 this week, flagged in #18223). Two [OuterloopTest] + Docker-required Hosting tests fail deterministically on the Hosting (8-core-ubuntu-latest) job and pass on windows-latest:

• Aspire.Hosting.Tests.DistributedApplicationTests.ProxylessContainerCanBeReferenced
• Aspire.Hosting.Tests.DistributedApplicationTests.WithEndpointProxySupportDisablesProxies

Both fail the same way:

Assert.IsType() Failure: Value is null
Expected: typeof(int)
Actual:   null
  at Aspire.Hosting.Tests.DistributedApplicationTests.AssertAllocatedProxylessPort(Service service)
  at Aspire.Hosting.Tests.DistributedApplicationTests.ProxylessContainerCanBeReferenced() line 1835

Root cause

Both tests assert on the redisNoPort proxyless service's Status.EffectivePort. PR #17924 ("Add proxyless endpoint port allocator") made Aspire pre-assign the proxyless host port synchronously (Service.Spec.Port) and intentionally excluded proxyless services from the startup address-wait in DcpExecutor — connection strings are built from the Aspire-assigned Spec.Port, which is available immediately, so startup need not block on DCP echoing the bound port.

DCP still reports that port back asynchronously via Status.EffectivePort. After StartAsync() and the app responding, DCP may not yet have populated EffectivePort for the proxyless service. The fast 8-core Linux runner reads the service list before that update lands → null; Windows lands the update first → pass. The unit-test fake sets EffectivePort = Spec.Port synchronously, so unit tests never hit the race.

This is a test bug, not a product regression — the orchestrator behavior is correct by design.

The fix

Test-side only. Both tests now wait for the proxyless service to report a non-null Status.EffectivePort (using KubernetesHelper.GetResourceByNameAsync, the existing watch-based waiter already used elsewhere in this file) before running AssertAllocatedProxylessPort, instead of asserting against a once-fetched service list. This preserves the end-to-end check that DCP actually bound the Aspire-assigned port. No product code changes.

Call-outs

• WithEndpointProxySupportDisablesProxies is not yet tracked by an issue; it shares the exact same root cause and is fixed by the same change.
• Validated on Linux in a Docker container (the environment that fails in CI): both tests fail with the verbatim null error before the change and pass (2/2, ~41s) after it.
• No product code touched; DcpExecutor's intentional proxyless exclusion is unchanged.

Fixes #8773

[github-actions]

🚀 Dogfood this PR with:

⚠️ WARNING: Do not do this without first carefully reviewing the code of this PR to satisfy yourself it is safe.

curl -fsSL https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.sh | bash -s -- 18249

Or

• Run remotely in PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.ps1) } 18249"

[Copilot]

Pull request overview

This PR fixes two deterministically failing [OuterloopTest] tests on Linux CI (ProxylessContainerCanBeReferenced and WithEndpointProxySupportDisablesProxies) that fail because DCP hasn't yet populated Service.Status.EffectivePort when the test assertion runs. The root cause is that PR #17924 intentionally excluded proxyless services from the DCP startup address-wait (since connection strings use the synchronously-assigned Spec.Port), but the tests asserted against EffectivePort from a one-shot list fetch, creating a race condition on fast Linux CI agents.

Changes:

• Replaced the synchronous GetEndpointService(serviceList, ...) helper with a new WaitForAllocatedProxylessServiceAsync that uses the existing watch-based KubernetesHelper.GetResourceByNameAsync to wait until DCP populates EffectivePort.
• Removed the now-unnecessary serviceList fetch from ProxylessContainerCanBeReferenced (the second test retains it for an unrelated Assert.All assertion).

[github-actions]

Retrying the failed CI jobs for this pull request from the CI run attempt. The rerun is being tracked in the rerun attempt.

[Ankit Jain (radical)]

Consolidating the unapproved Outerloop test fixes into a single PR: #18250. Closing in favor of that one (same change, preserved authorship).