View and troubleshoot problems detected by Amazon CloudWatch Application Insights
The topics in this section provide detailed information about the detected problems and insights displayed by Application Insights. It also provides suggested resolutions for detected issues with your account or your configuration.
Troubleshooting topics
CloudWatch console overview
An overview of problems that impact your monitored applications can be found under
the CloudWatch Application Insights pane in the overview page of the CloudWatch console
The CloudWatch Application Insights overview pane displays the following:
-
The severity of the problems detected: High/Medium/Low
-
A short summary of the problem
-
The problem source
-
The time the problem started
-
The resolution status of the problem
-
The affected resource group
To view the details of a specific problem, under Problem Summary, select the description of the problem. A detailed dashboard displays insights into the problem and related metric anomalies and snippets of log errors. You can provide feedback on the relevance of the insight by selecting whether it is useful.
If a new resource is detected that is not configured, the problem summary description takes you to the Edit configuration wizard to configure your new resource. You can view or edit your resource group configuration by choosing View/edit configuration in the upper right‐hand corner of the detailed dashboard.
To return to the overview, choose Back to overview, which is next to the CloudWatch Application Insights detailed dashboard header.
Application Insights problem summary page
Application Insights problem summary page
CloudWatch Application Insights provides the following information about detected problems on the problem summary page:
-
A short summary of the problem
-
The start time and date of the problem
-
The problem severity: High/Medium/Low
-
The status of the detected problem: In‐progress/Resolved
-
Insights: Automatically generated insights on the detected problem and possible root cause
-
Feedback on insights: Feedback you have provided about the usefulness of the insights generated by CloudWatch Application Insights
-
Related observations: A detailed view of the metric anomalies and error snippets of relevant logs related to the problem across various application components
CloudWatch agent merge conflict failures
CloudWatch Application Insights installs and configures the CloudWatch agent on customer instances. This includes creation of a CloudWatch agent configuration file with configurations for metrics or logs. A merge conflict can occur if a customer's instance already has a CloudWatch agent configuration file with different configurations defined for the same metrics or logs. To resolve the merge conflict, use the following steps:
-
Identify the CloudWatch agent configuration files on your system. For more information about the file locations, see CloudWatch agent files and locations.
-
Remove the resource configurations that you want to use in Application Insights from the existing CloudWatch agent configuration file. If you want to only use Application Insights configurations, delete the existing CloudWatch agent configuration files.
Alarms are not created
For some metrics, Application Insights predicts the alarm threshold based on previous data points for the metric. To enable this prediction, the following criteria must be met.
-
Recent data points – There must be a minimum of 100 data points from the last 24 hours. The data points don't need to be continuous and can be scattered throughout the 24 hour time frame.
-
Historical data – There must be a minimum of 100 data points spanning the time frame from 15 days before the current date to 1 day before the current date. The data points don't need to be continuous and can be scattered throughout the 15 day time frame.
Note
For some metrics, Application Insights delays the creation of alarms until the preceding conditions are met. In this case, you get a configuration history event that the metric lacks sufficient data points to establish the alarm threshold.
Feedback
Feedback
You can provide feedback on the automatically generated insights on detected problems by designating them useful or not useful. Your feedback on the insights, along with your application diagnostics (metric anomalies and log exceptions), are used to improve the future detection of similar problems.
Configuration errors
CloudWatch Application Insights uses your configuration to create monitoring telemetries for the components. When Application Insights detects an issue with your account or your configuration, information is provided in the Remarks field of the Application summary about how to resolve the configuration issue for your application.
The following table shows suggested resolutions for specific remarks.
| Remarks | Suggested resolution | Additional notes |
|---|---|---|
|
The quota for CloudFormation has already been reached. |
Application Insights creates one CloudFormation stack for each application to manage CloudWatch agent installation and configuration for all application components. By default, each AWS account can have 2000 stacks. See AWS CloudFormation Limits. To resolve this, raise the limit for CloudFormation stacks. |
n/a |
|
No SSM instance role on the following instances. |
For Application Insights to be able to install and configure CloudWatch agent on application instances, AmazonSSMManagedInstanceCore and CloudWatchAgentServerPolicy policies must be attached to the instance role. |
Application Insights calls the SSM DescribeInstanceInformation API to get the list of instances with SSM permission. After the role is attached to the instance, it takes time for SSM to include the instance in the DescribeInstanceInformation result. Until SSM includes the instance in the result, NO_SSM_INSTANCE_ROLE error remains present for the application. |
|
New components may need configuration. |
Application Insights detects that there are new components in the application Resource Group. To resolve this, configure the new components accordingly. |
n/a |
