Aglet

Investigate Duplicate Records and Delivery Repeats

Investigation should explain why two records represent one event or why they only look alike. Trace a matched pair from source through persistence, compare delivery and processing metadata, and reproduce the smallest repeat. Test retry, ordering, and identity hypotheses separately so the evidence remains useful.

Build a useful investigation brief

  1. Pair and label records

    Choose exact repeats, changed copies, and legitimate near-matches from the affected cohort. Label the expected logical relationship before comparing them. Record source identifiers, sequence or delivery markers, timestamps, payload differences, and the downstream view that exposes the suspected duplication.

  2. Trace write paths

    Follow each pair through ingestion, idempotency or grouping logic, retries, joins, and persistence. Capture the key used at each boundary and the order of writes. Reproduce with a bounded fixture or deterministic replay that preserves the relevant ordering and failure condition.

  3. Test identity hypotheses

    Compare source key reuse, missing deduplication key, reordered delivery, version handling, and legitimate event repetition as separate possibilities. Change one condition at a time and inspect both stored rows and derived totals. Do not call a retry the cause without evidence of repeated delivery.

What to carry forward

Deliver a brief containing labeled pairs, identity and write-path traces, a reproducible repeat, tested explanations, and unresolved ambiguity. Recommend a reversible prevention or correction boundary. If legitimate repetition cannot be distinguished from retries, escalate the identity contract instead of collapsing records.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow