No error message on proxy failure (e.g. because of Windows excluded port ranges)

[tomasherceg]

Is there an existing issue for this?

• I have searched the existing issues

Describe the bug

I have the same situation as #9634
The issue was closed with a claim that the error message is logged somewhere, but I cannot see any error:

• The Dashboard shows that the service is running, even though the proxy wasn't started.
• The AppHost logs doesn't show any warning or error - there are just the typical info messages and everything seems OK.
• I cannot find any other place where I could see any information that the proxy cannot start because the port is occupied or is in excluded port ranges.

Expected Behavior

I would expect to see some warning either in AppHost logs or in the Aspire dashboard.

Steps To Reproduce

1. Use any template with a single web app.
2. Run netsh interface ipv4 show excludedportrange protocol=tcp to list the excluded port ranges.
3. Pick any port from that range and use it in launchSettings.json:

   {
     "profiles": {
       "https": {
         "commandName": "Project",
         "launchBrowser": true,
         "environmentVariables": {
           "ASPNETCORE_ENVIRONMENT": "Development"
         },
         "applicationUrl": "https://localhost:50500"
       }
     }
   }
   

Exceptions (if any)

No response

Aspire doctor output

Anything else?

• .NET 10.0
• Aspire 13.5.3
• Windows 11 (latest updates installed)

[MihuBot]

I'm a bot. Here is a possible related and/or duplicate issue (I may be wrong):

• Unhealthy Azure Storage Emulator on some devices #9000

[joperezr]

🤖 Aspire triage pass — 2026-09-11 12:00 UTC

Area: area-orchestrator
Kind: bug
Repro: yes-minimal — see lines 19–35 of the issue body
Affected version(s): 13.5.3 (latest: 13.5.3)
Severity: S3 — proxy startup failure is hidden and resource state is misleading, but selecting an available port is a workaround

Urgency: ⚪ backlog — no milestone
Reason: reproducible Windows port-allocation edge case with a reasonable workaround; not important enough for 13.6

Suggested owner(s): David Negstad (@danegsta) (score 9)
Evidence:

• danegsta merged Avoid ephemeral executable target port collisions #19822 in area-orchestrator (last 30d), directly modifying proxy endpoint allocation
• danegsta closed Dashboard project can collide with its randomized OTLP port under aspire start --isolated #19691, aspire stop leaves session-lifetime containers running #19250, EnsureKubernetesAsync times out on GitHub-hosted CI runners in 13.4.x — KubernetesConfigReadRetryCount config removed, no bypass available #18041 as completed in area-orchestrator (last 30d)

Confidence: high
Needs human?: no

[karolz-ms]

David Negstad (@danegsta) silent failures are bad, please investigate