[IMPALA-10914] Iceberg query create inconsistent scan ranges between consecutive runs

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 4.0.0
• Fix Version/s: Impala 4.1.0
• Component/s: Distributed Exec
• Labels: impala-iceberg
• Epic Color: ghx-label-10
• Assignee: Zoltán Borók-Nagy
• Reporter: Riza Suminto
• Created: 10/Sep/21 16:18
• Updated: 08/Oct/21 08:36
• Resolved: 08/Oct/21 08:36

Description

When running a same query several times against hive table, scan ranges assignment between query runs are identical. This helps towards efficiently hitting already cached file handles and data.

However, for iceberg table query, consecutive runs of the same query yield varying scan range assignment. This lead to more cache miss and slower scans.