You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "Piotr Nowojski (Jira)" <ji...@apache.org> on 2025/02/07 16:47:00 UTC
[jira] [Created] (FLINK-37280) Inconsistent content of TableConfig in planner when mini batch is enabled
Piotr Nowojski created FLINK-37280:
--------------------------------------
Summary: Inconsistent content of TableConfig in planner when mini batch is enabled
Key: FLINK-37280
URL: https://issues.apache.org/jira/browse/FLINK-37280
Project: Flink
Issue Type: Improvement
Components: Table SQL / Planner
Affects Versions: 2.0-preview
Reporter: Piotr Nowojski
Attachments: image-2025-02-07-17-42-29-104.png
In planner when checking {{table.exec.mini-batch.enabled}}, it's value might be inconsistent. For example as described [here|https://github.com/apache/flink/pull/26051#issuecomment-2637506032], when checking for that config value in {{FlinkChangelogModeInferenceProgram}} during planning, the returned value might be different compared to when translating plan into the exec nodes, leading to potential inconsistencies.
The problem is visible here:
!image-2025-02-07-17-42-29-104.png!
note configuration and rootConfiguration having different values.
I've tracked it down to this code: https://github.com/apache/flink/blob/b1544e4e513d2b75b350c20dbb1c17a8232c22fd/flink-table/flink-table-planner/src/main/scala/org/apache/flink/table/planner/plan/optimize/StreamCommonSubGraphBasedOptimizer.scala#L106:L125
introduced in FLINK-32622.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)