Carl Niblaeus
06/22/2026, 2:28 PMkapa.ai
06/22/2026, 2:28 PMlatest and earliest as a troubleshooting step. If set to latest, events produced while the connector is not running could be missed permanently. [Kafka 0 records]
3. Destination BigQuery Version Bugs One forum post notes that certain BigQuery destination versions have known bugs (e.g., BigQuery destination 1.21 was listed as bugged in the changelog). Using an outdated or buggy destination connector version could contribute to data loss. [Shopify to BigQuery]
4. BigQuery Destination Failing to Commit All Records There is an open issue where large syncs to BigQuery complete the load job successfully but fail with a TransientErrorException (missing terminal stream status), causing only the first batch to be committed to the final table. This could explain a consistent small percentage of lost records. [BigQuery TransientError]
Recommendations • Check your connector versions: Upgrade both the Kafka source and BigQuery destination connectors to the latest versions.
• Review Kafka consumer permissions and connectivity: Ensure the Airbyte consumer is consistently connecting and has the right permissions.
• Check Auto Offset Reset: Make sure it is set to earliest if you cannot afford to miss events.
• Review sync logs: Look for TransientErrorException or partial commit errors in the BigQuery destination logs.
• Open a GitHub issue: Given the specificity of your problem (consistent ~0.6% loss), it may be worth filing a detailed bug report at the Airbyte GitHub repository with logs and connector versions.
The knowledge sources do not provide a definitive root cause for your specific scenario, so I'd strongly recommend sharing detailed sync logs with the Airbyte community or support team for further diagnosis. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.