[DRILL-7225] Merging of columnTypeInfo for file with different schema throws NullPointerException during refresh metadata

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.16.0
• Fix Version/s: 1.17.0
• Component/s: Metadata
• Labels: ready-to-commit
• Assignee: Venkata Jyothsna Donapati
• Reporter: Venkata Jyothsna Donapati
• Reviewer: Aman Sinha
• Created: 29/Apr/19 19:16
• Updated: 23/May/19 23:42
• Resolved: 03/May/19 22:48

Description

Merging of columnTypeInfo from two files with different schemas throws nullpointerexception. For example if a directory Orders has two files:

• orders.parquet (with columns order_id, order_name, order_date)
• orders_with_address.parquet (with columns order_id, order_name, address)

When refresh table metadata is triggered, metadata such as total_null_count for columns in both the files is aggregated and updated in the ColumnTypeInfo. Initially ColumnTypeInfo is initialized with the first file's ColumnTypeInfo (i.e., order_id, order_name, order_date). While aggregating, the existing ColumnTypeInfo is looked up for columns in the second file and since some of them don't exist in the ColumnTypeInfo, a npe is thrown. This can be fixed by initializing ColumnTypeInfo for columns that are not yet present.

 

Issue Links

• links to: GitHub Pull Request #1773