Help for owners
My website on Azure App Service is down or very slow
Short answer
Azure App Service has a built-in tool, Diagnose and solve problems, that shows availability, CPU, memory and recent changes for the last 24 hours with no set-up. Most outages trace to a recent change, an app that runs out of CPU or memory, or ports and connections running out.
Does this sound like your situation?
- The site shows a 500, 502 or 503 page, or never finishes loading.
- It was fine until a recent update or a configuration change.
- It gets slower during the day and recovers after a restart.
- Azure shows the app as running but visitors cannot reach it.
- A container-based app keeps failing to start.
Why does this happen?
A recent change
A release, setting or dependency change is a common start of an outage. App Service diagnostics includes an Application Changes view and change analysis to compare what changed with when the problem began.
Confirmed by learn.microsoft.com
The app is out of CPU or memory
The Availability and Performance category lists CPU Usage, Memory Usage, Web App Down and Web App Slow reports, which show whether the plan is too small or one process is eating resources.
Confirmed by learn.microsoft.com
Network connections or ports run out
The same category lists SNAT Port Exhaustion and TCP Connections reports. An app that opens many outbound connections without reusing them can run out, and requests then fail or hang.
Confirmed by learn.microsoft.com
The container does not start properly
For apps in containers, the app has to answer on the port Azure expects. The technical page on the HTTP ping failure explains the message.
Technical detail: Azure App Service on Linux restarts the container and the site never starts
A mitigation is hiding the real problem
Auto-healing can restart an app on rules such as request count, slow requests, memory or status codes. Microsoft describes it as a temporary measure until the root cause is found.
Confirmed by learn.microsoft.com
What can I do about it today?
- Open Diagnose and solve problemsIn the Azure portal open your web app and choose Diagnose and solve problems in the left menu. Start with Availability and Performance, then Web App Down.
- Save what the report showsTake screenshots of the graphs and of any detected problem. Include the time range.
- Ask what changed in the last dayCheck with whoever deploys: was there a release, a setting change, a new plan size or a certificate or domain change?
- Check the Azure service statusLook at Azure's own status page before assuming the fault is yours.
- Avoid scaling or deleting in a panicChanging the plan size or deleting the app can add cost or lose settings. A restart is the only quick step that is safe, and only once the reports are saved.
When should I bring someone in?
- The site is down for customers and the reports do not point to a clear cause.
- The problem comes back after each restart.
- No one on your side has deploy access to the app.
How do I handle it?
- Read the diagnostics firstI go through availability, CPU, memory, connections and change analysis for the exact time window.
- Check logs and recent releasesI compare the failure time with application logs and the release history.
- Reproduce on a copy where I canA staging copy lets me test a fix without risking the live site.
- Fix the cause, then the guard railsI change the code or settings, then add health checks and alerts so you hear about it before customers do.
- Keep a rollbackEvery release has a tested way back.
The price depends on what I find, so I quote in writing after a short discovery call. Tell me what is happening and I reply within 24 hours.
What do owners ask about this?
Does the diagnostics tool cost extra?
Microsoft describes it as an experience that needs no configuration. Application Insights for code-level exceptions is enabled separately.
Will a restart fix it?
It may clear a symptom for a while. If the cause is a change, a leak or exhausted connections, the problem returns.
Is the outage Azure's fault?
Sometimes. Check Azure's status page, but most App Service outages trace to the app or its configuration.
Sources
How was this page made?
I drafted this page with AI assistance from the official documents listed under Sources. Found a mistake? Tell me.