Why Recovery of Events Sent to Mini Is Not Supported

Overview

Occasionally, events intended for a production environment may mistakenly be sent to a Mini (development) instance due to configuration errors. This article explains why Snowplow does not support recovery of such events and the limitations involved.

Understanding the Issue

When events are directed to Mini instead of the production pipeline:

These events are not “failed”, they are successfully processed but in the wrong environment.
As a result, standard recovery tooling designed for failed event reprocessing cannot be applied.
The events typically reside in Elasticsearch (or Mini-specific storage), not in a pipeline stage from which they can be re-ingested.

Technical Constraints

Events stored in Mini are retained in Elasticsearch using a different format and structure from what the main pipeline requires. Specifically:

The data is not enriched or formatted in a way that supports reprocessing or direct injection into production.
Extracting and transforming this data into a suitable format would require significant bespoke development, with no out-of-the-box tooling available.
Any such effort would be technically risky, resource-intensive, and time-bound by Mini’s retention policy (typically 7 days).

Why This Is Out of Scope

Snowplow classifies this type of request as out-of-scope for both standard Support and Professional Services due to:

Lack of tooling and documentation to support the process reliably.
Historical attempts that resulted in failed efforts, despite heavy involvement from Engineering and Data teams.
No clear or supported recovery path, either internally or through customer-led efforts.

Our Recommendation

We recommend taking a snapshot of the Mini instance as an immediate backup measure if this situation arises. This can help preserve data while further discussions take place.

For prevention, we’re happy to assist in validating your pipeline configuration and ensuring events are correctly routed to avoid future mishaps.