[Bug] Power Apps Code Apps intermittently fail to load with `UnexpectedError` even for a brand-new blank app created from the official Vite template with no connections configured.

[admcfirebase-ai]

Describe the bug

Power Apps Code Apps intermittently fail to load with UnexpectedError even for a brand-new blank app created from the official Vite template with no connections configured.

The Power Apps shell loads, but the inner iframe sometimes fails before app code runs. When this happens, the runtime package request under powerapps/appruntime/.../storageproxy/.../index.html returns 503.

Steps to Reproduce

1. Follow the official npm quickstart for Power Apps Code Apps.
2. Create a brand-new blank Vite app.
3. Initialize it with npx power-apps init.
4. Build and publish it with npx power-apps push.
5. Open the generated play URL and refresh it multiple times.
6. The app may work briefly, then fail with UnexpectedError.

Expected behavior

A newly published blank Code App should load reliably, especially when no connections are configured.

Actual behavior

The outer Power Apps shell loads, but the inner iframe intermittently fails with:

{"code":"UnexpectedError","message":"An unexpected error occurred."}

At the same time, the runtime package request returns 503.

Screenshots or Error Messages

Observed iframe error:

{"code":"UnexpectedError","message":"An unexpected error occurred."}

Observed network behavior:

• storageproxy/.../index.html returns 503
• Power Apps shell still loads
• App launch metadata can still report the app as Ready / Published

Environment information

• Framework, build tool or relevant package used: React + Vite + Power Apps Code Apps npm quickstart
• Any connection/components: None for the blank repro app
• SDK version used in repro: @microsoft/power-apps 1.1.1
• Vite plugin version used in repro: @microsoft/power-apps-vite 1.0.2
• CLI version used in repro: power-apps 0.10.0
• Browsers tested: Chrome, Edge, Safari

Additional context

• This reproduces not only with an existing app, but also with a brand-new blank app created from the official guide.
• Because the blank app has no connections, this does not appear to be caused by connector configuration or app code.
• The issue appears intermittent: the same app can work on one refresh and fail on the next.
• I can provide HAR traces and request IDs privately if needed.

[eschavez]

Thanks for the detailed report, including the HAR trace offer and the clean blank-app repro — that's exactly the kind of signal that helps us narrow things down quickly.

We're tracking this. The pattern you're describing — the outer Power Apps shell loading while the inner iframe fails with UnexpectedError and the storageproxy request returns 503 — points to a transient failure in the backend storage proxy that serves the app bundle, rather than anything in the app's own code or connections. The fact that a brand-new blank app reproduces it rules out connector configuration as a factor.

We'll investigate on our end and update this issue as we learn more. If you're able to capture request IDs or a HAR trace from a failing session, feel free to share them here or privately — that would help us pinpoint the affected backend path.

In the meantime, a page refresh typically recovers the session once the transient failure clears on the backend.

[admcfirebase-ai]

Update on Intermittent 503s:
Even my exisiting Code Apps stops working due to this issue.

After extensive edge testing, I have verified this is an intermittent infrastructure routing issue on the hosted runtime, not a CLI build or code packaging issue. (npm run dev works locally 100% of the time).

Reproduction / Behavior Pattern:

The deployed app will work perfectly on initial load.
After several minutes of inactivity, refreshing the tab suddenly results in a [503 Unexpected Error] on the storageproxy/.../index.html request.
This happens randomly across completely different client networks (LAN, WiFi, Mobile Data) and devices.

Crucially: Even a completely blank, newly initialized template Code App exhibits the exact same 503 behavior.
This strongly suggests an issue with load balancer routing and session affinity. Once the session drops due to inactivity, a refresh assigns a new storageproxy node, randomly hitting one in the regional pool that is failing to retrieve files from the underlying blob storage.

Date: Sat, 09 May 2026 07:44:35 GMT
x-ms-correlation-id: 597a3152-a9c0-4af1-970c-5aafc778572e
x-ms-service-request-id: 93b0b357-32b5-4c92-91dd-e292aa074359
x-ms-gateway-clusters: prdil101cae
x-ms-igw-tracking-id: 93b0b357-32b5-4c92-91dd-e292aa07435920260509074435_prdil101caegb0_2
x-ms-islandgateway: _prdil101caegb0_2
x-ms-ppapigateway: _prdil101caegb0_1
Cache-Control: no-cache, no-store

Here are the exact trace IDs from a 503 failure to help them identify the failing proxy nodes (looks like prdil101 cluster):

[aymangamalaly]

We are experiencing a very similar issue with Code Apps hosted in a UAE environment.
Observed behavior:

• Same app, same browser, same machine
• Works correctly on one ISP/network
• Fails consistently on another ISP/network
• Works immediately when using an external VPN
• Code Apps hosted in non-UAE environments work normally without any issue

The app works locally and deploys successfully, and the Power Apps shell loads correctly, but the inner app runtime fails to load.
The failing requests return the following error:

{"code":"UnexpectedError","message":"An unexpected error occurred."}

This appears to be related to routing/connectivity/network path issues toward the environment.api.powerplatform.com endpoints rather than application code.

[admcfirebase-ai]

We are experiencing a very similar issue with Code Apps hosted in a UAE environment. Observed behavior:

• Same app, same browser, same machine • Works correctly on one ISP/network • Fails consistently on another ISP/network • Works immediately when using an external VPN • Code Apps hosted in non-UAE environments work normally without any issue

The app works locally and deploys successfully, and the Power Apps shell loads correctly, but the inner app runtime fails to load. The failing requests return the following error:

{"code":"UnexpectedError","message":"An unexpected error occurred."}

This appears to be related to routing/connectivity/network path issues toward the environment.api.powerplatform.com endpoints rather than application code.

OMG... I was going mad. Thankfully, I am not the only one facing this issue. Yes, we are facing the same isue with our UAE North Environment. Even my old codeapps are dead now due to this issue. I hope this gets sorted out soon.
Thanks

[jzh24516]

I got the same issue, but I am using vibe.powerapps.com, I spent one week to build my apps, it used to work 2 hours ago, then I published (so luckily I did that), so my published app did work, but my vibe code apps in maker preview didn't work at all. Multiple unknow issue cause app start with blank or {"error":"Proxy Error","message":"Vite development server is not running.","details":{"path":"/","method":"GET","errorCode":"ECONNREFUSED"}}. I am very frustrated. Clear off cache, change browser didn't work at all. So instable, let everybody down. Please fix it if this is a known bug.

[aymangamalaly]

I would like to share an update regarding this issue.

For more than one day now, the Code Apps in the same UAE environment have started working normally without any changes from my side.

No changes were made to the application, environment, browser, device, network settings, or firewall configuration.

This further suggests that the issue may have been related to routing/connectivity or the Microsoft UAE Power Platform endpoint rather than the application itself.