What causes 000625 (57014)?

Multiple concurrent DML operations (INSERT, UPDATE, MERGE) are targeting the same table simultaneously. A long-running transaction is holding a lock while other statements queue up beyond the 20-waiter limit. An explicit transaction was opened with BEGIN and not committed or rolled back, leaving locks open. Stored procedures with scoped transactions were not properly finalised

How do I fix 000625 (57014)?

Step 1: Identify blocking transactions: SELECT * FROM TABLE(INFORMATION_SCHEMA.TRANSACTION_HISTORY()) WHERE state = 'OPEN' ORDER BY start_time.. Step 2: Terminate the blocking transaction if safe to do so: SELECT SYSTEM$ABORT_TRANSACTION(transaction_id).. Step 3: Refactor concurrent DML to use Snowflake's MERGE statement for upserts instead of separate INSERT + UPDATE.. Step 4: Ensure all stored procedures commit or roll back their transactions explicitly before returning.. Step 5: Stagger job schedules to reduce simultaneous write contention on hot tables.

High severityexecution

Power BI Refresh Error:
000625 (57014)

What does this error mean?

Snowflake aborted and rolled back a statement because it was waiting on a table-level lock held by another transaction and the queue of waiting statements exceeded the limit.

Common causes

1Multiple concurrent DML operations (INSERT, UPDATE, MERGE) are targeting the same table simultaneously
2A long-running transaction is holding a lock while other statements queue up beyond the 20-waiter limit
3An explicit transaction was opened with BEGIN and not committed or rolled back, leaving locks open
4Stored procedures with scoped transactions were not properly finalised

How to fix it

1Step 1: Identify blocking transactions: SELECT * FROM TABLE(INFORMATION_SCHEMA.TRANSACTION_HISTORY()) WHERE state = 'OPEN' ORDER BY start_time.
2Step 2: Terminate the blocking transaction if safe to do so: SELECT SYSTEM$ABORT_TRANSACTION(transaction_id).
3Step 3: Refactor concurrent DML to use Snowflake's MERGE statement for upserts instead of separate INSERT + UPDATE.
4Step 4: Ensure all stored procedures commit or roll back their transactions explicitly before returning.
5Step 5: Stagger job schedules to reduce simultaneous write contention on hot tables.

Frequently asked questions

Is Snowflake ACID-compliant, and does that protect against data corruption here?

Yes — Snowflake transactions are fully ACID-compliant. A rolled-back transaction leaves no partial writes; however, the pipeline that triggered the transaction must handle the error and retry to ensure the data eventually lands.

Can I use AUTOCOMMIT to avoid this issue?

AUTOCOMMIT is on by default in Snowflake. Problems arise when code explicitly opens a transaction with BEGIN and forgets to close it. Always pair BEGIN with a COMMIT or ROLLBACK in exception handlers.