[
https://issues.apache.org/jira/browse/SPARK-61250?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hyukjin Kwon updated SPARK-61250:
---------------------------------
Fix Version/s: connect-rust-4.3.0
> The sync Rust API panics inside a current_thread Tokio runtime, and stop()
> strands the session
> ----------------------------------------------------------------------------------------------
>
> Key: SPARK-61250
> URL: https://issues.apache.org/jira/browse/SPARK-61250
> Project: Spark
> Issue Type: Bug
> Components: Connect
> Affects Versions: connect-rust-4.3.0
> Reporter: Hyukjin Kwon
> Priority: Major
> Fix For: connect-rust-4.3.0
>
>
> Every synchronous call (collect, count, schema, sql, stop) goes through
> runtime::block_on. Inside a multi-thread runtime it uses block_in_place.
> Inside a current_thread runtime (the default of #[tokio::test]) it calls
> Runtime::block_on, which panics with "Cannot start a runtime from within a
> runtime". stop() marks the session stopped before it releases it, so the
> panic leaves the server session alive until its idle timeout. Repro: a
> #[tokio::test] that calls range(3).count(). Official PySpark has no such
> limit: the call works from any thread.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
