Updating Table Metadata with REFRESH TABLE

Function

Spark caches Parquet metadata to improve performance. If you update a Parquet table, the cached metadata is not updated. Spark SQL cannot find the newly inserted data and an error similar with the following is reported:

DLI.0002: FileNotFoundException: getFileStatus on  error message

You can use REFRESH TABLE to solve this problem. REFRESH TABLE reorganizes files of a partition and reuses the original table metadata information to detect the increase or decrease of table fields. This statement is mainly used when the metadata in a table is not modified but the table data is modified.

Syntax

1	REFRESH TABLE [db_name.]table_name;

Keyword

None

Parameter

**Table 1** Parameter description
Parameter	Description
db_name	Database name that contains letters, digits, and underscores (_). It cannot contain only digits or start with an underscore (_).
table_name	Table name of a database that contains letters, digits, and underscores (_). It cannot contain only digits or start with an underscore (_). The matching rule is *^(?!_)(?![0-9]+$)[A-Za-z0-9_$]$**. If special characters are required, use single quotation marks ('') to enclose them.