Application Insights logs are missing or incorrect for Azure Functions apps

Článek
08/24/2023

You can closely monitor the function app through the integration between Azure Functions and Application Insights. And you can use Application Insights without any custom configuration.

If the Application Insights logs are missing, or if the data appears to be partial or inaccurate, use the following steps to resolve the issue.

Check configuration of function app

Navigate to your function app in the Azure portal.
Select Diagnose and solve problems to open Azure Functions diagnostics.
In the Search bar, type Function Configuration Checks and open it.
You see a diagnostic report of all function app configuration checks. In particular for Application Insights, the following checks are performed:
- Only one of the following connection settings exists:
  - APPINSIGHTS_INSTRUMENTATIONKEY Application Insights Instrumentation key
  - APPLICATIONINSIGHTS_CONNECTION_STRING connection
    
    We recommend that you use the APPLICATIONINSIGHTS_CONNECTION_STRING for more stable behavior. The ability to use APPINSIGHTS_INSTRUMENTATIONKEY will be deprecated by 2025.
- The AzureWebJobsDashboard built-in logging is disabled, as recommended.
- Sampling is enabled for the Azure Functions telemetry (enabled by default).

Recommendation: The function app should be on version 4 and the runtime version should be at least 4.15.2xx. This is because, from this version onwards, you can track the log flows from Azure Functions to the Application Insights service. By monitoring the log flows, you can check for missing logs.

Logs are missing or partial

Application Insights collects log, performance, and error data. Sampling configuration is used to reduce the volume of telemetry. The Sampling feature is enabled by default with the settings shown in the following host.json example. Excluded types aren't sampled.

{
  "logging": {
    "applicationInsights": {
      "samplingSettings": {
        "isEnabled": true,
        "maxTelemetryItemsPerSecond" : 20,
        "excludedTypes": "Request;Exception"
      }
    }
  }
}

If you notice any partially missing logs, this might occur because of sampling. To determine the actual sampling rate, use an Analytics query that uses the required time interval that's shown in the following code snippet. If you observe that the TelemetrySavedPercentage for any sampling type is less than 100, then that type of telemetry is being sampled.

union requests,dependencies,pageViews,browserTimings,exceptions,traces
| where timestamp > todatetime("mm/dd/yyyy hh:mm:ss") and timestamp < todatetime("mm/dd/yyyy hh:mm:ss")
| summarize TelemetrySavedPercentage = 100/avg(itemCount), TelemetryDroppedPercentage = 100-100/avg(itemCount) by bin(timestamp, 1d), itemType
| sort by timestamp asc

For more information, see Data collection, retention, and storage in Application Insights.

Control volume and verbosity of logs

You can increase or suppress the logs that are written. To do this, you can use a combination of log level and categories as configured in host.json.

The Azure Functions logger includes a category for every log. The category indicates which part of the runtime code or your function code generated the log. For example:

The Host.Results and Function.<YOUR_FUNCTION_NAME> are some of the available categories.
A log level is assigned to every log. The value indicates relative importance, such as Warning or Information.

For more information, see the other categories and log levels available.

You can configure how your application should write the logs by following the sample code snippet:

{
  "version": "2.0",
  "logging": {
    "logLevel": {
      "default": "Information", // catch all default, with modifications below for individual categories.
      "Function": "Warning", // Warning level from all Functions (except the ones configured below).
      "Host.Aggregator": "Trace", // Log all traces in the 'customMetrics' table of (and shown on Metrics/Alerts blade in AI) - use either this or Host.Results
      "Host.Results": "Error", // Error and Critical requests are only logged in the 'requests' table of the AI (and shown on Monitor Functions blade in Functions App) - use either this or Host.Aggregator
      "Function.Function1": "Information", //Information level logs from Function 1, logged in 'traces', 'dependencies' and 'customMetrics' tables of AI
      "Function.Function2.User": "Information" //user code logs from Function2, logged in 'traces' table of AI
    },
    "applicationInsights": {
      "samplingSettings": {
        "isEnabled": true,
        "maxTelemetryItemsPerSecond": 1,
        "excludedTypes": "Exception"
      }
    }
  }
}

To configure these values at the App settings level (to avoid redeployment on host.json changes), override specific host.json values by creating an equivalent value as an application setting. For more information, see Override host.json values.

For more examples about how to suppress logs, see functions-log-suppression.

The virtual network integrated function app doesn't generate any logs

If a function app is integrated with a virtual network, you must open port 443 for outgoing traffic in your server firewall to allow the Application Insights SDK or Application Insights Agent to send data to the portal for the following URLs:

dc.applicationinsights.azure.com
dc.applicationinsights.microsoft.com
dc.services.visualstudio.com
*.in.applicationinsights.azure.com

For more information, see IP addresses used by Azure Monitor.

Contact us for help

If you have questions or need help, create a support request, or ask Azure community support. You can also submit product feedback to Azure feedback community.

Sdílet prostřednictvím