Where can I see initialization errors for an Azure Function App?

Question

Where can I see initialization errors for an Azure Function App?

Reinhart, Jeff (MNIT) 40

I built a Function App using the following tutorial:

https://learn.microsoft.com/en-us/azure/azure-functions/create-first-function-vs-code-python

Using V2 and Python 3.11.

I deploy the app via the CLI successfully. I open the Function App in Azure Portal and the http_trigger is listed under Functions. I tested the route and it works.

Then I introduce code that will cause an error on initialization. Immediately after:

app = func.FunctionApp(http_auth_level=func.AuthLevel.ANONYMOUS)

I introduce a divide by zero error that will run when the function app initializes:

n = 100/0

I deploy the app via the CLI. It says it is successful. I open the Function App in Azure Portal and the http_trigger is no longer listed under Functions. This is the only indication that I can find that there was an error on initialization.

Where can I see what the error is on initialization?

I have tried Log Analytics and Application Insights. From what I can see, these only report on errors post initialization that are triggered when routes are called.

I have checked Function App | Deployment | Deployment Center | Logs. The deployment shows it was successful.

I have tried downloading the *docker.log file from Kudu, but that does not include any errors / tracebacks from the initialization.

Being able to see the error on init is very important for debugging on more complex projects. I have a project that I can run successfully from local, but when it deploys, I get the above behavior. I have isolated which part of the code causes the error by testing when deployments result in an HTTP trigger and when they do not. I need to know what the actual initialization error is from Python to diagnose the cause of the error.

Reinhart, Jeff (MNIT) 40 Reputation points

2025-03-11T14:05:59.9233333+00:00
The lack of logging on deployment/query/sync was confirmed by Microsoft Support on ticket with Tracking ID #2412120040008195.

When Python code is deployed to an HTTP Trigger Azure Function, the deployment process runs the code in Azure. I assume this is to query and sync the HTTP Triggers. If there are errors in the code and the query/sync process fails, those errors are not logged.

If you run the deployment via the CLI, the deployment process reports that the deployment was successful with “remote build succeeded!”. If you run the deployment via the Azure Tools VS Code extension the process will say it succeeded and you may or may not see “No HTTP triggers found” in the output.

At the root URL, it will say “your function 4.0 app is up and running”. However, when you call the route, you will get a 404 error. You may also see that the HTTP Trigger is not listed as a function when you look at the Azure Function | Overview via Portal (this is intermittent).

Not having the errors logged makes debugging very difficult. If you have something that is operating system dependent, the code will run on your local but then will not run in Azure and you will be forced to remove parts of your code and redeploy until you find the part of your code that is causing the error. Essentially debugging in the dark.

One example would be trying to reference a directory in your local repository that has no files. In Windows, this will work. When the code gets to the Linux environment, it will not because a directory with no files does not exist.

To test this, do the following:

Create an Azure Function following these steps: https://learn.microsoft.com/en-us/azure/azure-functions/create-first-function-vs-code-python#create-an-azure-functions-project

Deploy the following code:

import azure.functions as func import logging app = func.FunctionApp(http_auth_level=func.AuthLevel.ANONYMOUS) @app.route(route="http_trigger") def http_trigger(req: func.HttpRequest) -> func.HttpResponse: logging.info('Python HTTP trigger function processed a request!') return func.HttpResponse( "This HTTP triggered function executed successfully.", status_code=200 )

The code will work at https://<yourfunctionapp>.azurewebsites.net/api/http_trigger. Then deploy this code:

import azure.functions as func import logging app = func.FunctionApp(http_auth_level=func.AuthLevel.ANONYMOUS) n = 1/0 @app.route(route="http_trigger") def http_trigger(req: func.HttpRequest) -> func.HttpResponse: logging.info('Python HTTP trigger function processed a request!') return func.HttpResponse( "This HTTP triggered function executed successfully.", status_code=200 )

Wait a few minutes and try the route again. You will get a 404 error.

The code intentionally introduces a divide by zero error that will occur when the code is deployed. The divide by zero error will not be logged anywhere in Azure.

Please do not comment with a list of things to try unless you have tried the above steps and have found a way to get the divide by zero error to show in the logs and have specific instructions on how to make that happen. Again, this issue was confirmed by Microsoft Support.

This issue was also shared on the Share Your Ideas page by another poster: https://feedback.azure.com/d365community/idea/9f19ba41-30a0-ef11-95f6-000d3a01397d

If you are having the same issue, please consider upvoting and commenting on that post.

3 answers

Your answer

Answer 1

Hi @Reinhart, Jeff (MNIT),

Sorry for the inconvenience happed we know that you are still facing the issue, so we also suggest you please follow the below and please let us know the if that is working as expected.

Set logging level to DEBUG explicitly with logging.basicConfig(level=logging.DEBUG) at the start of your code and ensure Application Insights sampling is disabled or configured correctly to capture all logs.
Check host.json for correct logging configuration (disable sampling, set correct log levels) also verify Docker logging configuration (ensure logs go to stdout/stderr in a custom container setup).
Check Kudu logs in the LogFiles directory for more details and make sure Azure Function settings allow the logging level for debug and info.

Hope this helps you in finding the logs if you have any further concerns, please reach out us.

Thank you, Have a great day ahead.

Answer 2

Pinaki Ghatak 5,700 Microsoft Employee

Hello @Reinhart, Jeff (MNIT)

To find the error that occurred during initialization, you can try checking the Function App's logs in Kudu. Here are the steps to do that:

Go to your Function App in the Azure Portal.
Click on "Platform features" in the left-hand menu.
Under "Development Tools", click on "Advanced tools (Kudu)".
In the Kudu dashboard, click on "Debug console" and select "CMD".
Navigate to the "LogFiles" folder and then to the "Application" folder.
Look for the log file with the most recent timestamp and open it.
Search for any error messages or stack traces that occurred during the initialization of your Function App.

If you still cannot find the error, you can try adding more detailed logging to your code to help diagnose the issue. You can use the logging module in Python to log messages at different levels of severity. For example, you can add a logging.error() statement to log any errors that occur during initialization.

I hope this helps you find the error and diagnose the issue with your Function App. I hope that this response has addressed your query and helped you overcome your challenges. If so, please mark this response as Answered. This will not only acknowledge our efforts, but also assist other community members who may be looking for similar solutions.

Reinhart, Jeff (MNIT) 40

Go to your Function App in the Azure Portal.
Click on "Platform features" in the left-hand menu.
Under "Development Tools", click on "Advanced tools (Kudu)".

There is no "Advanced tools (Kudu)" under "Development Tools":

User's image I assume you mean "click Go".

4. In the Kudu dashboard, click on "Debug console" and select "CMD".

There is not a "Debug console" in the Kudu dashboard: User's image

So I will assume you mean SSH or Bash. I will uses Bash.

Navigate to the "LogFiles" folder and then to the "Application" folder.
Look for the log file with the most recent timestamp and open it.

There are no files in LogFiles/Application:

User's image

Search for any error messages or stack traces that occurred during the initialization of your Function App.

No logs to search.

Other things I have tried:

Modifying host.json logging:

{
   "version": "2.0",
   "logging": {
     "logLevel": {
       "default": "Trace",
       "Function": "Trace",
       "Worker": "Trace"
     },
     "applicationInsights": {
       "samplingSettings": {
         "isEnabled": true,
         "excludedTypes": "Request"
       }
     }
   },
   "extensionBundle": {
     "id": "Microsoft.Azure.Functions.ExtensionBundle",
     "version": "[4.*, 5.0.0)"
   }
 }

This seems to cover configuration categories that might include app initialization.

In my function_app.py code, I have the following:

import azure.functions as func
import logging, sys

# This successfully registers a logger. That is confirmed by logging.info('Python HTTP trigger function processed a request!')
# successfully logging when the http_trigger is called.
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)

# https://learn.microsoft.com/en-us/azure/developer/python/sdk/azure-sdk-logging
# Direct logging output to stdout. Without adding a handler,
# no logging output is visible.
handler = logging.StreamHandler(stream=sys.stdout)
logger.addHandler(handler)

# Initialization level logging. THIS DOES NOT LOG.
logging.debug('This is a debug log.')
logging.info('This is an info log.')

app = func.FunctionApp(http_auth_level=func.AuthLevel.ANONYMOUS)

@app.route(route="http_trigger")
def http_trigger(req: func.HttpRequest) -> func.HttpResponse:
    logging.info('Python HTTP trigger function processed a request!')
    return func.HttpResponse(
        "This HTTP triggered function executed successfully.",
        status_code=200
    )

The logging.info() call that is called from the http_trigger() function shows in Application Insights | Investigate | Transaction search if that route is called. I can search "Python HTTP trigger" and see that trace. However, the logging from this code block:

# Initialization level logging. THIS DOES NOT LOG.
 logging.debug('This is a debug log.')
 logging.info('This is an info log.')

Is never logged.

I have also tried looking in the latest log using Azure App Service | REST API | Current Docker logs | Download as zip. The logging calls that are not called from the http_trigger function also do not log in that log file.

What am I missing?

Anonymous

2024-12-30T06:49:03.15+00:00

Hi @Reinhart, Jeff (MNIT),

Just checking in to see if the information above was helpful. If you have any further updates on this issue, please feel free to post them here.

Thank you.
Anonymous

2024-12-31T00:52:43.43+00:00

Hi @Reinhart, Jeff (MNIT),

Following up to see if the above answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

Thankyou.

Answer 3

Where to Check for Initialization Errors in Azure Function Apps:

Application Insights

If Application Insights is configured for your Function App, it can capture exceptions during initialization.
- Ensure Application Insights is enabled and check the Failures and Search tabs for traces or exceptions during startup.
Steps to check:
```
- Navigate to your Function App in the Azure Portal.

   - Go to **Application Insights > Search**.

      - Filter logs by **Exceptions** or search for specific terms like "divide by zero."
```

Azure Monitor Logs

Use Log Analytics to query the logs from your Function App.

A helpful query might look like this:

     traces
     | where message contains "Exception"
     | sort by timestamp desc

This may capture initialization exceptions if the logs are correctly configured. Kudu Diagnostic Console

- The **Kudu Console** can sometimes provide more detailed logs.

   - Access Kudu via `https://<your-function-app-name>.scm.azurewebsites.net/`.

      - Check the `LogFiles/Application` folder for detailed error logs.

Streaming Logs
- Use Azure's Live Metrics Stream or stream logs from the CLI to capture runtime and initialization logs.
```
     
     az webapp 
```
This can sometimes reveal unfiltered error messages. Function App Configuration
- Ensure that the AzureWebJobsDashboard and AzureWebJobsStorage connection strings are configured correctly. Missing or incorrect values can lead to initialization issues.
Inspect Docker Logs (For Linux Containers or Custom Images)
- If your Function App is running in a container, initialization errors may be in the container logs.
- Download the container logs using:
```
     az webapp log download --name <function-app-name> --resource-group <resource-group-name>
     
     
```
- Look for docker.log and other files in the zip archive.
Enable Detailed Logs
- Modify your host.json file to include detailed logging:
```
     
```

{ "logging": { "logLevel": { "default": "Information", "Function": "Trace" } } } ```

Redeploy your Function App, and recheck the logs.

Why the http_trigger Disappears:

When an initialization error occurs, the Function App fails to fully load. The absence of the http_trigger in the portal indicates that the app's startup process was interrupted. The error, in this case, prevents the Function App from registering the function.

Reinhart, Jeff (MNIT) 40 Reputation points

2024-12-20T21:09:08.7+00:00

Thank you for your post! Were you able to test your suggestions before posting? If there are any more details I can provide so you can test on your end, please let me know. Unfortunately, I am still not able to see the Python error that occurs when the app initializes.

Application Insights

There are no errors under Investigate > Transaction search results that include “Exception” or “ZeroDivisionError” or “division by zero”. I tried multiple versions of these terms.

Azure Monitor Logs

In the Monitoring | Logs blade, I ran the following KQL query:

traces

| where message contains "Exception"

There were no results for this query.

Docker Logs

As I said in my original post, I have tried downloading the *docker.log file from Kudu. I downloaded the current file from logfiles. That does not include any errors / tracebacks from the initialization. There are no files in logfiles/application. Since you did not describe what “logs are correctly configured” means, I am not sure what else to do. If you think it would help to provide more info on configuring logs, I would appreciate if you would share that.

Azure's Live Metrics Stream

From what I have seen, this does not capture initialization errors. It only captures calls to the application. I have watched the stream during deployment and have not seen any errors from initialization.

Container Logs

The app is not running from a container.

Enable Detailed Logs

Your suggested code was missing an “s” on the “Functions” element, but I got the setting { "logging": { "logLevel": { "default": "Information", "Functions": "Trace" } } } added to host.json.

I went through all of the above possibilities again and did not find any Python errors from initialization. All messages state that the app deployed successfully and is “ready to serve requests.”