When application errors spike, you need to quickly identify what failed, how often, and why. Error analysis in Application Real-Time Monitoring Service (ARMS) surfaces HTTP errors and uncaught exceptions across your application instances, with time series charts to spot trends and direct links to distributed traces and business logs for root cause investigation.
What counts as an error
ARMS classifies an API call as an error when the response indicates a failure visible to the caller:
| Error type | Criteria |
|---|---|
| HTTP errors | Any HTTP response with a status code greater than 400 |
| Uncaught exceptions | Explicit error information returned when a try...catch statement fails to catch an exception |
Exclude HTTP status codes from error analysis
Not all HTTP status codes represent actionable errors. For example, you may want to exclude 404 responses from error analysis if they are expected in your application.
To exclude specific HTTP status codes, configure filter conditions on the Custom Configuration tab of the Application Settings page. ARMS skips excluded status codes during analysis.
Prerequisites
Before you begin, ensure that you have:
An application monitored by Application Monitoring. For setup instructions, see Overview
Open the error analysis page
Log on to the ARMS console. In the left-side navigation pane, choose Application Monitoring > Applications.
On the Applications page, select a region in the top navigation bar and click the name of the application.
NoteIf the
icon appears in the Language column, the application is connected to Application Monitoring. If a hyphen (-) appears, the application is connected to Managed Service for OpenTelemetry.In the left-side navigation pane, click Application Details.
Select an application instance, set the time period, and then click the Error Analysis tab.
Monitor error counts over time
The Errors section displays a time series curve of error counts over the selected time period. Use this chart to spot error spikes and correlate them with deployments or configuration changes.
| Action | Description |
|---|---|
| Hover over the chart | View error count at a specific point in time |
| Click and drag | Zoom into a narrower time range |
Click the  icon | View error statistics for a specific time period, or compare error counts across different dates within the same time window. Use this to verify a fix or detect regressions after a deployment. |
Click the  icon | View the API details of the metric |
Break down errors by HTTP status code
The HTTP - Status Code section displays the time series curves of HTTP status code statistics for the application over the selected time period. Use this chart to identify which status codes (such as 404, 500, or 502) contribute to error spikes.
| Action | Description |
|---|---|
| Hover over the chart | View detailed statistics at a specific point in time |
| Click and drag | Zoom into a narrower time range |
| Click a legend entry | Show or hide data for a specific status code |
| Click the icon | View statistics for a specific time period, or compare the statistics in the same time period on different dates |
| Click the icon | View the API details of the metric |
Investigate individual errors
The error list displays all individual errors that occurred during the selected time period. From this list, you can drill into a specific error's distributed trace or view its associated business logs to identify the root cause.
Trace an error through the call chain
Click a trace ID in the TraceId column to open the distributed trace for that error. The trace shows the full call chain across services, helping you pinpoint where the failure occurred. For more information, see Trace query.
View business logs for an error
Click View Logs in the Actions column to view the business logs associated with the error.
To use this feature, you must associate trace IDs with the business logs of your application beforehand. Without this association, ARMS cannot link logs to specific errors. For setup instructions, see Associate trace IDs with business logs.