> ## Documentation Index
> Fetch the complete documentation index at: https://docs.lighton.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Automated retention of logs and response data

> The log retention system ensures that audit logs are automatically anonymized after the configured retention period, helping maintain GDPR compliance while preserving essential historical records.

## **Who Can Manage It**

Only syst. admins can configure and adjust the data retention period through configuration keys (`/admin/settings/configkeyvalue/` URL of your instance) . Regular admins or end-users have no access to this setting.

## **What It Applies To**

The configuration key `LOG_RETENTION_PERIOD_DAYS` defines how long **audit logs (LogEntry objects)** are kept in their original, identifiable form before being anonymized.

✔️ **Impacted data:**

* **Audit logs** that may contain personal information (e.g. user identifiers, metadata linked to actions in the system).

❌ **Not impacted:**

* ML logs (requests, completions, responses)
* Pipeline events
* Task results
* User chat history or conversations

Those elements are **not deleted or anonymized** by this configuration.

## **How It Works**

1. **Default Retention**:
   * By default, `LOG_RETENTION_PERIOD_DAYS` is set to **180 days**.
   * After this period, audit log entries are anonymized.
2. **Anonymization Process**:
   * A scheduled daily job (`schedule_all_companies_anonymization`) runs across all companies.
   * It triggers the `anonymize_company_logs_task()`.
   * This calls the `CompanyDataLogAnonymizationService`, which anonymizes old `LogEntry` objects.
   * The event record is kept, but personal data is replaced with anonymous values.
3. **Important Note**:
   * **No data is deleted.**
   * The database size does not decrease, since anonymization only replaces values without removing rows.
   * The benefit is **compliance with data protection regulations (e.g. GDPR)**, not storage optimization.

## **Version Information**

* The anonymization mechanism has been present in the system for several releases.
* In the **Rapid Rabbit** release, related fixes improved the overall reliability of scheduled tasks, ensuring anonymization runs consistently.