Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
Impala 2.1
-
None
-
None
-
None
Description
Environment: 3 Node mini cluster (local)
Dataset: TPCH
File Format: parquet
Git Commit: 429c689
Query: TPCH-Q5
When mem_limit is set to 79m, the exception is:
ImpalaBeeswaxException: Query aborted:Memory limit exceeded Query did not have enough memory to get the minimum required buffers in the block manager.
When mem_limit is set to 118m, the exception is:
Query aborted:Metadata states that in group hdfs://localhost:20500/test-warehouse/tpch.lineitem_parquet/5e4880b09fe4c900-d4b208d329436fbc_623556170_data.0.parq[0] there are 2142543 rows, but only 0 rows were read.
The problem is not enough memory, but the ImpalaBeeswaxException text in the 118m case does not state this (most of the time).
This also applies to TPCH-Q6 with 15m mem_limit.