Joe McDonnell created IMPALA-14951:
--------------------------------------
Summary: Iceberg delete can hang when using the tuple cache
Key: IMPALA-14951
URL: https://issues.apache.org/jira/browse/IMPALA-14951
Project: IMPALA
Issue Type: Bug
Components: Backend
Affects Versions: Impala 5.0.0
Reporter: Joe McDonnell
With mt_dop, IcebergDeleteNode has the same independent builder as
PartitionedHashJoinNode. It needs the same logic as used for IMPALA-13660 to
notify the build side when a probe side thread closes before probing.
Interestingly enough, this is hard to demonstrate with a select. The
coordinator cancels the query when it receives all the rows, which posts the
builder out of its wait.
In a delete DML, this is not true, so it can hang indefinitely. This reproduces
via these steps:
0. Use mt_dop>2 and enable_tuple_cache=true
1. Create a partitioned Iceberg v2 tables with some rows
2. Delete some rows
3. Delete more rows - this hangs
A more specific reproducing case is attached.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
