What causes STALE_STREAM (Stream Staleness)?

The stream was not consumed (queried with DML) within the table's DATA_RETENTION_TIME_IN_DAYS window. The table's Time Travel retention was reduced after the stream was created. An ETL or task pipeline paused or failed, preventing regular stream consumption. DATA_RETENTION_TIME_IN_DAYS is set to 0 on the source table, making streams stale almost immediately. Stream consumer (Snowflake Task) was suspended and resumed after the retention window elapsed

How do I fix STALE_STREAM (Stream Staleness)?

Check stream staleness: SELECT SYSTEM$STREAM_HAS_DATA('my_stream') and SELECT STALE FROM INFORMATION_SCHEMA.STREAMS WHERE STREAM_NAME = 'MY_STREAM'. If the stream is stale, drop and recreate it — historical CDC data is unrecoverable: DROP STREAM my_stream; CREATE STREAM my_stream ON TABLE my_table. Increase the source table's retention: ALTER TABLE my_table SET DATA_RETENTION_TIME_IN_DAYS = 14. Ensure the consuming task or pipeline runs more frequently than the retention window. Set up monitoring on stream staleness before it becomes unrecoverable — check STALE status daily

High severityexecution

Power BI Refresh Error:
STALE_STREAM (Stream Staleness)

What does this error mean?

A Snowflake Stream has become stale — the change data captured by the stream is no longer available within the table's Time Travel retention period. Consuming a stale stream returns an error and data cannot be recovered.

Common causes

1The stream was not consumed (queried with DML) within the table's DATA_RETENTION_TIME_IN_DAYS window
2The table's Time Travel retention was reduced after the stream was created
3An ETL or task pipeline paused or failed, preventing regular stream consumption
4DATA_RETENTION_TIME_IN_DAYS is set to 0 on the source table, making streams stale almost immediately
5Stream consumer (Snowflake Task) was suspended and resumed after the retention window elapsed

How to fix it

1Check stream staleness: SELECT SYSTEM$STREAM_HAS_DATA('my_stream') and SELECT STALE FROM INFORMATION_SCHEMA.STREAMS WHERE STREAM_NAME = 'MY_STREAM'
2If the stream is stale, drop and recreate it — historical CDC data is unrecoverable: DROP STREAM my_stream; CREATE STREAM my_stream ON TABLE my_table
3Increase the source table's retention: ALTER TABLE my_table SET DATA_RETENTION_TIME_IN_DAYS = 14
4Ensure the consuming task or pipeline runs more frequently than the retention window
5Set up monitoring on stream staleness before it becomes unrecoverable — check STALE status daily

Frequently asked questions

Is there any way to recover data from a stale stream?

No — once a stream is stale, the underlying Change Data Capture data is gone. You must recreate the stream and perform a full reload of the target table from the source to restore consistency.

How frequently must a stream be consumed to stay fresh?

The stream must be consumed (i.e., its offset advanced by a DML transaction that reads it) at least once per DATA_RETENTION_TIME_IN_DAYS of the source table. With 1-day retention, the stream must be consumed at least once every 24 hours.