[
https://issues.apache.org/jira/browse/SPARK-51351?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17931346#comment-17931346
]
Jungtaek Lim commented on SPARK-51351:
--------------------------------------
Going to file a PR sooner.
> TWS PySpark implementation materializes the entire output iterator in python
> worker
> -----------------------------------------------------------------------------------
>
> Key: SPARK-51351
> URL: https://issues.apache.org/jira/browse/SPARK-51351
> Project: Spark
> Issue Type: Bug
> Components: Structured Streaming
> Affects Versions: 4.0.0
> Reporter: Jungtaek Lim
> Priority: Critical
>
> I figured out the implementation of dump_stream is materializing the output
> iterator.
> This means all the outputs will be materialized when JVM signals to Python
> worker that there is no further input (at task completion), which brings up
> two critical issues:
> # output is seriously delayed to be produced for downstream operator
> # memory usage on Python worker would be problematic
> We need to make this be "lazy evaluated", mostly iterator/generator.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
