自动刷新 Apache Iceberg™ 表¶

Set the severity level to capture events¶

To capture automated refresh events, you must set the LOG_LEVEL parameter at the Iceberg table level or account level. The LOG_LEVEL determines which events to capture based on the following values:

• ERROR: Events that signal a change requiring human intervention to resolve.

• WARN: Events that signal an issue that can be resolved without human intervention.

• DEBUG: High-volume events.

For example, to capture DEBUG-level automated refresh events for a specific Iceberg table, use the following command:

ALTER ICEBERG TABLE <my_table_name> SET LOG_LEVEL = DEBUG;

有关更多信息，请参阅 为日志、指标和跟踪设置级别。

Query your event table for automated refresh events¶

Before you can query for automated refresh events, you must set up an event table and set the severity level for event capture.

The following example shows how to retrieve Iceberg automated refresh events that are generated during snapshot processing:

SELECT record_type,
       record:"name" event_name,
       record:"severity_text" log_level,
       resource_attributes:"snow.database.name" database_name,
       resource_attributes:"snow.schema.name" schema_name,
       resource_attributes:"snow.table.name" table_name,
       resource_attributes:"snow.catalog.integration.name" catalog_integration_name,
       record_attributes:"snow.snapshot.id" snapshot_id,
       parse_json(value):metadata_file_location metadata_file_location,
       parse_json(value):snapshot_state snapshot_state
       parse_json(value):created created
       parse_json(value):detected detected
  FROM my_active_event_table
  WHERE record_type='EVENT' AND event_name='iceberg_auto_refresh_snapshot_lifecycle';

输出：

+-------------+-----------------------------------------+-----------+---------------+-------------+------------+--------------------------+---------------+------------------------+----------------+--------------------------+--------------------------+
| RECORD_TYPE | EVENT_NAME                              | LOG_LEVEL | DATABASE_NAME | SCHEMA_NAME | TABLE_NAME | CATALOG_INTEGRATION_NAME | SNAPSHOT_ID   | METADATA_FILE_LOCATION | SNAPSHOT_STATE | CREATED                  | DETECTED                 |
+-------------+-----------------------------------------+-----------+---------------+-------------+------------+--------------------------+---------------+------------------------+----------------+--------------------------+--------------------------+
| EVENT       | iceberg_auto_refresh_snapshot_lifecycle | DEBUG     | TESTDB        | TESTSH      | TESTTABLE  | glue_integration         | 4281775564368 | metadata.json          | detected       | 2025-05-21T20:38:06.822Z | 2025-10-17T21:01:49.695Z |
| EVENT       | iceberg_auto_refresh_snapshot_lifecycle | DEBUG     | TESTDB        | TESTSH      | TESTTABLE  | glue_integration         | 4281775564368 | metadata.json          | started        | 2025-05-21T20:38:06.822Z | 2025-10-17T21:01:49.695Z |
| EVENT       | iceberg_auto_refresh_snapshot_lifecycle | DEBUG     | TESTDB        | TESTSH      | TESTTABLE  | glue_integration         | 4281775564368 | metadata.json          | completed      | 2025-05-21T20:38:06.822Z | 2025-10-17T21:01:49.695Z |
+-------------+-----------------------------------------+-----------+---------------+-------------+------------+--------------------------+---------------+------------------------+----------------+--------------------------+--------------------------+

Query your event table for stale automated refresh events¶

You can query your event table for tables whose last successful refresh is older than a threshold you define.

1. The following example defines a threshold of 20 minutes:

   SET STALENESS_THRESHOLD_MINUTES = 20;
   

   Copy

2. Query for tables whose last successful refresh is older than the threshold you defined:

   WITH last_successful_refresh AS (
     -- Find the most recent 'completed' event for each table
     SELECT
       resource_attributes:"snow.table.name"::STRING AS table_name,
       MAX(timestamp) AS last_success_timestamp
     FROM
       <my_active_event_table>
     WHERE
       record:"name" = 'iceberg_auto_refresh_snapshot_lifecycle'
       AND parse_json(value):snapshot_state::STRING = 'completed'
     GROUP BY
       table_name
   )
   
   -- Select tables whose the last successful refresh was longer ago than our threshold
   SELECT
     table_name,
     last_success_timestamp
   FROM
     last_successful_refresh
   WHERE
     last_success_timestamp < DATEADD(minute, -$STALENESS_THRESHOLD_MINUTES, CURRENT_TIMESTAMP())
   ORDER BY
     last_success_timestamp ASC;
   

   Copy

• Where my_active_event_table is your active event table.

输出：

+------------+-------------------------+
| TABLE_NAME | LAST_SUCCESS_TIMESTAMP  |
+------------+-------------------------+
| my_table   | 2025-10-10 07:24:30.854 |
+------------+-------------------------+