Error Center¶

Guance Error Center is a full-stack solution designed to help teams systematically track and manage application errors. It automatically aggregates error information from performance monitoring, user access, and logs, categorizing massive amounts of discrete error reports into manageable tasks, and provides a complete analysis and handling workflow.

Core Concepts¶

Error¶

Refers to an abnormal event unit identified from APM, RUM, and Logs, containing three essential elements: error type, error message, and error stack.

Error Issue¶

The system uses intelligent fingerprinting algorithms to aggregate multiple errors (Errors) with the same root cause into a unique Error Issue (problem unit). Each Issue has an independent lifecycle status, following a standard flow from "Triage" to "Resolved", and is associated with complete analysis context (including performance traces, user sessions, source code mapping, and log metrics).

Error Delivery Logic¶

The Error Center employs an Intelligent Aggregation Mechanism for unified management of error data. The system only processes error data that can generate a unique identifier (issue_id), which is generated by combining the error fingerprint and service context, ensuring continuous tracking of identical errors.

Error Aggregation Mechanism¶

The system achieves intelligent error aggregation through the following steps:

1. Error Fingerprint Generation

a. The system generates a unique fingerprint identifier (fingerprint) for each error:

b. Data Integrity Requirements:

The generation of an error fingerprint is based on data integrity. APM, RUM, and Logs data must contain fields at the above paths to generate a fingerprint; if fields are missing, that piece of data is skipped.

c. Standardization Processing:

• Before calculating the fingerprint, the system performs standardization processing, automatically identifying and filtering variable content (such as timestamps, user IDs, UUIDs, etc.) in the error message (error_message), ensuring consistent fingerprints for identical logical errors;

• Stack information (error_stack) undergoes optimization processing, retaining only key business call lines, filtering out framework internal calls and third-party library calls;

• The system strictly validates field integrity; data missing necessary fields will be excluded.

2. Error Issue Identifier (Issue ID Generation)

a. Based on the error fingerprint, the system combines the context in which the error occurred to generate a unique issue identifier (issue_id), used for tracking and managing the issue within the Error Center. The specific generation logic is as follows:

• APM Errors: service + fingerprint (Service name + fingerprint)
• RUM Errors: app_name + fingerprint (Application name + fingerprint)
• Log Errors: service + fingerprint (Service name + fingerprint)

b. Lifecycle Initiation:

When the system first generates a new issue_id for a particular error combination, it automatically records the first occurrence time of this issue in the workspace and begins full lifecycle status tracking and aggregated analysis for it.

3. Error Data Storage

Typical Use Cases¶

Use Case 1: Backend Service Exception Aggregation¶

An e-commerce platform's order service generates tens of thousands of NullPointerException during a promotion period. Traditionally, these are scattered into alerts due to different order IDs.

Solution:

1. Configure APM error rules to monitor order-service;
2. The Error Center automatically extracts fingerprints, aggregating exceptions from the same code location into a single Issue;
3. The development team only needs to handle one root cause issue; after fixing it, all related alerts disappear.

👆🏻 From "alert storms" to "precise localization", processing efficiency is greatly improved.

Use Case 2: Frontend Error Tracing¶

Users report occasional white screens on the checkout page, but it cannot be reproduced. Traditional logs lack client-side context.

Solution:

1. Configure RUM error rules to collect Web application errors;
2. Filter the error list by RUM and application type, then enter the details page;
3. View "Session Replay" to restore the user's operation path: click coupon → parsing failure → white screen;
4. Check the error stack; SourceMap automatically maps to line 42 of the source code, the JSON parsing logic.

👆🏻 From "cannot reproduce" to "second-level localization", the user experience issue resolution cycle is reduced from days to hours.

Use Case 3: Log Exception Correlation Analysis¶

Application logs frequently show "Connection Timeout", but it's unclear whether it's a network issue or a downstream service problem.

Solution:

1. Configure log error rules to collect logs containing error_type:DBError;
2. In the error details page, view "Correlated Traces". The trace error details show the timeout occurred during the payment-service call;
3. Check "Infrastructure Dependencies" to confirm that the database connection pool monitoring shows connection exhaustion;
4. Root cause identified: Insufficient thread pool configuration in the downstream payment service, not a network issue.

👆🏻 From "surface phenomenon" to "root cause localization", avoiding misjudgment leading to ineffective optimization.

Getting Started¶

Prerequisites¶

Before starting, please confirm your data is already connected to Guance:

• APM: Services are connected and reporting trace data

• RUM: Web and other applications have integrated the SDK

• Logs: Logs are indexed and contain error fields

Configuration¶

The Error Center operates in a configuration-driven manner.

1. Before use, you need to Create error delivery rules to explicitly specify the data scope to monitor (specific log indexes, APM services, or RUM applications).
2. After configuration, the system will only perform intelligent analysis and aggregation processing on errors generated from these selected data sources. You can go to the Error List to view and manage aggregated error Issues, filtering by status, source, etc.;
3. You can further analyze Error Details, delving into a single error, correlating traces, logs, and sessions for root cause analysis.