Hi Airbyte team, We’re running Airbyte OSS with a MariaDB 10.6.22 source using the MySQL source connector v3.51.5 (CDC / ROW-based binlog) into Snowflake destination v4.0.31, and we’re stuck on a CDC failure we’d like guidance on fixing correctly. We’re consistently hitting:

MissingTableMapEventException: No TableMapEventData has been found for table id:100
eventType=UPDATE_ROWS
nextPosition=5606243
timestamp=2025-12-18 00:30:00 UTC

We’ve investigated the binlog and confirmed: •

table id 100

=

mubi.film_group_availabilities

• A valid

TABLE_MAP

exists immediately before the

UPDATE_ROWS

• The failure occurs because Debezium resumes mid logical event group, missing the

TABLE_MAP

Refreshing the stream or removing the table does not change the behaviour — the connector always resumes from the same binlog position. Could you please confirm the recommended way (in Airbyte OSS) to reset or clear the persisted Debezium CDC state / cursor for a connection, so that the source restarts from a valid binlog boundary? We want to be sure we’re using the correct mechanism (UI vs API vs internal DB) and not risking unintended data loss.

You’re hitting a known Debezium/binlog pattern: the connector is resuming from a position inside a logical event group (after

TABLE_MAP

), which triggers

MissingTableMapEventException

. The MySQL connector issue you referenced shows the same root cause: Debezium starts reading “within the logical event group (e.g. from WRITE_ROWS and not preceding TABLE_MAP)”. [[MySQL CDC issue](https://github.com/airbytehq/airbyte/issues/51607)] From the Airbyte side, the part you can control is the connection state (which holds the CDC cursor) and, if needed, a Clear of the destination data. ### 1. Safest, supported way in OSS: Clear + resync Airbyte’s “Clear” operation wipes the destination tables for a connection (or a single stream) but does not touch your source. After a Clear, the next sync re‑backfills from the beginning according to the source’s replication method. [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] UI steps (OSS): 1. Go to Connections → select the problematic connection. 2. To reset all streams: - Open the Settings tab → click Clear data → confirm. [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] 3. To reset only the failing table: - On the Status (connection details) page, find the stream

mubi.film_group_availabilities

. - Click the three dots (

⋯

) next to the stream → Clear data → confirm. [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] 4. After the Clear finishes successfully, trigger a new sync; Airbyte will re‑backfill that stream and rebuild CDC state from a clean point. Impact: - Destination rows for the cleared stream(s) will be deleted; tables remain but are emptied. [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] - If your source doesn’t retain full history in binlog or tables, this can result in data loss (i.e., you may not be able to reconstruct the full past). [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] ### 2. Editing the CDC cursor / state directly (advanced) Airbyte lets you manually edit the connection state, which contains the CDC cursor (e.g., last binlog filename/position). This can force Debezium to resume from an earlier, valid binlog position without doing a full Clear, but must be done very carefully. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)] UI steps: 1. Go to Connections → select the connection. 2. Open the Settings tab. 3. Click the Advanced dropdown. 4. In Connection State, click into the JSON to edit it, then click Update state to save. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)] Docs explicitly warn: > Updates to connection state should be handled with extreme care. Updates may break your syncs, requiring a full historical sync of your data to fix. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)] In your case, you would: - Identify a safe earlier binlog point (before the problematic

UPDATE_ROWS

/

TABLE_MAP

pair) in

mubi.film_group_availabilities

. - Edit the corresponding state fields (typically the binlog file and position) back to that point. - Run the sync and verify that Debezium now resumes from a correct boundary. Risks: - If you move the cursor backward, any changes between the old state and the new state may be emitted again (duplicate changes). - If the cursor is set incorrectly, the connector may fail or require a full Clear + resync to recover. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)] ### 3. When to prefer which - If you can tolerate a full re‑backfill of the affected stream(s) and your source retains full history: → Use Clear data for that stream or the whole connection, then resync. This is the most straightforward and supported recovery path in OSS. [[Clearing data](https://docs.airbyte.com/platform/operator-guides/clear)] - If you must avoid a full re‑backfill and understand the binlog positions well: → Carefully edit the connection state JSON to move the CDC cursor back to a known‑good binlog boundary, then sync. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)] Beyond these two mechanisms (Clear + state edit), the knowledge sources do not describe any other recommended way (such as directly modifying Airbyte’s internal DB) to reset Debezium CDC state. For anything more invasive than the UI/API mechanisms above, the docs advise doing it only under direct Airbyte guidance. [[Modifying state](https://docs.airbyte.com/platform/cloud/managing-airbyte-cloud/manage-connection-state)]

can I clear and resync one particular stream? There are too many tables in this connection to clear all and start again

A message from kapa.ai

the clear one table failed with this

sourceStats must be set with recordCount for state message id=1

@kapa.ai

A message from kapa.ai