Correlating Applicaton Insights logs for cloud_RoleInstance with the instance name in Metrics for an App Service

Justin Wright 0

I'm looking at some Azure Metrics and can see that there's a specific instance in our App Service that seems to have consistently higher response times. It has a name like: ln1ldlwk0000LE

To example what's actually going wrong with that instance, I'd like to find that instance in Application Insights logs to see what's going on that specific instance during the time period I see the spike on the metrics view. However, in AppInsights the requests table (among others) uses cloud_RoleInstance. These values look completely different; example: d53628ee0c3e407eaafc19bcf26a8c4d

Has anyone tried to correlate the former to the latter? I'd love ideas.

PS: What I mean by the instance on the Metrics view is when you split a metric by instance:
User's image

Justin Wright 0 Reputation points

2024-04-25T15:50:31.6466667+00:00

More context if it helps: The App Service is running on an App Service Plan for Linux
SnehaAgrawal-MSFT 20,856 Reputation points

2024-04-26T09:51:46.9066667+00:00

@Justin Wright Thanks for reaching here! When you split a metric by instance in Azure Metrics, you can view the performance metrics for each individual instance of the App Service, In Azure App Service, the "instance" on the Metrics view refers to the individual instances of the app service that are running in the App Service Plan.

This can be useful for identifying performance issues or resource constraints on specific instances of the app service.

For details see- https://learn.microsoft.com/en-us/azure/azure-monitor/essentials/analyze-metrics#apply-metric-splitting

Note that- Application Insights, is feature of Azure Monitor, is an extensible Application Performance Management service and the logs in Application Insights are generated by application code. In Application Insights Application Map uses the cloud role name property to identify the components on the map. The Application Insights SDK or Agent automatically adds the cloud role name property to the telemetry emitted by components in an Azure App Service environment.

Cloud role instance is the Name of the instance where the application is running. Computer name for on-premisis, instance name for Azure.

For details suggest you refer here- https://learn.microsoft.com/en-us/azure/azure-monitor/app/app-map?tabs=net#set-or-override-cloud-role-name

I hope this helps! Let me know if you have any further questions.
Justin Wright 0 Reputation points

2024-04-30T15:17:54+00:00

I appreciate the background information, but I'm looking to correlate the two.

Specifically, how can I use the "instance" name that is provided in Azure Metrics, to find logs in Application Insights when Application Insights is using a different identifier (cloud_RoleInstance)?
Justin Wright 0 Reputation points

2024-04-30T22:17:00.1733333+00:00

For example, where does the "instance" name come from in Azure Metrics. If I'm able to customize the value for cloud_RoleInstance as the second link you provided suggests, what can I use to have it match the "instance" name when splitting in Azure Metrics?
SnehaAgrawal-MSFT 20,856 Reputation points

2024-05-02T12:02:58.34+00:00

@Justin Wright Thanks for reply! Having internal discussions on this to gather more insights on this.
SnehaAgrawal-MSFT 20,856 Reputation points

2024-05-07T06:07:02.71+00:00

@Justin Wright Thanks for reply!

It seems you are starting journey from AppService Resource Metrics blade. And then wants to find related logs in Application Insights RP Logs.

As per my knowledge this is NOT possible,

"Metrics" that you see under "Monitoring" of AppService is different from what is stored in the Application Insights. The "Metrics" shows the PAAS level metric, but the AppInsights stores what "your application sends".

Suggest you to start troubleshooting journey from App Insights Resource Metrics Blade and then go to logs in App Insights RP.

App Service Resource has platform logs and metrics, which are different from signals in App Insights that are from your code.

Let us know if further query or issue remains.
Justin Wright 0 Reputation points

2024-05-07T15:14:57.08+00:00

Sorry, but that isn't helpful at all. Overriding the cloud_RoleName, or adding a custom dimension in an ITelemetryInitializer seems like a possible solution, but the primary issue is that I don't know from what the "instance" name comes from in the Azure Metrics views. This is what I need.
SnehaAgrawal-MSFT 20,856 Reputation points

2024-05-08T12:07:33.7433333+00:00

@Justin Wright We received your feedback that the answer provided on the thread was not helpful. Thank you for taking time to share your feedback.

The App Service Resource encompasses platform logs and metrics, providing insights into the performance and behavior of the underlying infrastructure. These logs and metrics are distinct from the signals captured by App Insights, which specifically pertain to the functioning of your code. While both offer valuable data for analysis, they serve different purposes.

They are fundamentally separate entities and should not be merged.

Therefore, it's advisable to treat them as separate entities to ensure accurate interpretation and effective troubleshooting.

Let us know.
Justin Wright 0 Reputation points

2024-05-10T16:09:02.3+00:00

If the per instance metrics cannot be correlated to any information that one could analyze in order to take action, what value do they provide?

If I set up an alert that a specific instance is exceeding a threshold, but I'm unable to investigate why, what value does that alert have?
SnehaAgrawal-MSFT 20,856 Reputation points

2024-05-14T08:04:09.95+00:00

@Justin Wright Please be aware that a "metric instance" should not be confused with the data sent to the LA workspace by the application.

It's recommended to begin the troubleshooting journey from App Insights Resource Metrics Blade and then go to logs in App Insights RP.

One method to facilitate this correlation is by dynamically setting the AppRoleInstance name, using something like "Environment.MachineName". This approach enables easier alignment of information between the Metrics Explorer and the log data.

Hope this helps.

Share via

Correlating Applicaton Insights logs for cloud_RoleInstance with the instance name in Metrics for an App Service