Notice Deduplication

Overview

When your application encounters an error loop (e.g., a bug in a frequently-called endpoint), it can generate thousands of identical error reports in seconds. Notice deduplication protects your Checkend instance by:

• Detecting identical errors within a configurable time window (default: 60 seconds)
• Incrementing an occurrence counter instead of creating duplicate notice records
• Still tracking total occurrences for accurate error frequency metrics
• Skipping redundant notifications for duplicate errors

How It Works

Deduplication uses a composite key to identify identical errors:

• App ID — Errors are deduplicated per app
• Problem Fingerprint — Based on error class and location
• Backtrace Fingerprint — Distinguishes different call paths to the same error

Configuration

Deduplication is enabled by default. You can configure it using environment variables:

export NOTICE_DEDUP_WINDOW_SECONDS=300

API Response

The errors endpoint response includes deduplication information:

First Occurrence (201 Created)

{
  "id": 12345,
  "problem_id": 789,
  "deduplicated": false,
  "occurrence_count": 1
}

Duplicate Within Window (200 OK)

Note: No id field since no notice record was created.

{
  "problem_id": 789,
  "deduplicated": true,
  "occurrence_count": 5
}

Tracking Occurrences

Even when notices are deduplicated, Checkend tracks the total number of error occurrences:

• notices_count — Number of full notice records created
• deduplicated_count — Number of deduplicated occurrences (no notice created)
• total_occurrences — Sum of both counts

The last_noticed_at timestamp is updated for both regular and deduplicated notices, ensuring accurate "last seen" information.

Behavior Details