[SPARK-54452] Fix empty response from SparkConnect server for spark.sql(...) inside FlowFunction
#53156
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
dongjoon-hyun
left a comment
dongjoon-hyun
left a comment
There was a problem hiding this comment.
Hi, @SCHJonathan . FYI, Apache Spark community does not use this kind of tag, '[bugfix]' in the PR title.
SCHJonathan
changed the title
spark.sql(...) inside FlowFunctionspark.sql(...) inside FlowFunction
SCHJonathan
changed the title
spark.sql(...) inside FlowFunctionspark.sql(...) inside FlowFunction
Contributor
Author
|
fyi @vicennial |
Member
|
cc @sryza |
Member
dongjoon-hyun
left a comment
dongjoon-hyun
left a comment
There was a problem hiding this comment.
If you don't ind, please file a new JIRA issue as a bug fix instead of reusing SPARK-54020, @SCHJonathan .
...test/scala/org/apache/spark/sql/connect/pipelines/SparkDeclarativePipelinesServerSuite.scala
Outdated
Show resolved
Hide resolved
SCHJonathan
changed the title
spark.sql(...) inside FlowFunctionspark.sql(...) inside FlowFunction
|
|
||
| test( | ||
| "SPARK-54452: spark.sql() outside a pipeline flow function should return a " + | ||
| "sql_command_result") { |
Member
There was a problem hiding this comment.
Why do we expect the same behavior? According to the log, it is guarded by insidePipelineFlowFunction condition, isn't it?
dongjoon-hyun
changed the title
spark.sql(...) inside FlowFunctionspark.sql(...) inside FlowFunction
| proto.Command | ||
| .newBuilder() | ||
| .setSqlCommand( | ||
| proto.SqlCommand |
Member
There was a problem hiding this comment.
May I ask why we use different setSqlCommand in two test cases? Can we use the following like the other test case in the same way?
.setSqlCommand(
proto.SqlCommand
.newBuilder()
.setInput(proto.Relation
.newBuilder()
.setSql(proto.SQL.newBuilder().setQuery("SELECT * FROM RANGE(5)"))
.build())
.build())
.build())
Member
|
Gentle ping, @SCHJonathan . |
dongjoon-hyun
approved these changes
Nov 23, 2025
dongjoon-hyun
pushed a commit
that referenced
this pull request
Nov 23, 2025
…sql(...)` inside FlowFunction ### What changes were proposed in this pull request? In PR #53024, we added SDP support for `spark.sql(...)` inside a FlowFunction. For these calls, instead of eagerly executing the SQL, the Spark Connect server should return the raw logical plan to the client and defer execution to the flow function. However, in that PR we constructed the response object but forgot to actually return it to the Spark Connect client, so the client received an empty response. This went unnoticed in tests because, when the client sees an empty `spark.sql(...)` response, [it falls back to creating an empty DataFrame holding the raw logical plan](https://github.com/apache/spark/blob/master/python/pyspark/sql/connect/session.py#L829-L835), which happens to match the desired behavior. This PR fixes the bug by returning the proper response instead of relying on that implicit fallback. ### Why are the changes needed? ### Does this PR introduce _any_ user-facing change? This PR fixes a bug introduced in #53024 where the server did not return the constructed spark.sql(...) response to the client. ### How was this patch tested? New tests ### Was this patch authored or co-authored using generative AI tooling? No Closes #53156 from SCHJonathan/jonathan-chang_data/fix-spark-sql-bug. Authored-by: Yuheng Chang <[email protected]> Signed-off-by: Dongjoon Hyun <[email protected]> (cherry picked from commit 997525c) Signed-off-by: Dongjoon Hyun <[email protected]>
Member
|
Merged to master/4.1. |
huangxiaopingRD
pushed a commit
to huangxiaopingRD/spark
that referenced
this pull request
Nov 25, 2025
…sql(...)` inside FlowFunction ### What changes were proposed in this pull request? In PR apache#53024, we added SDP support for `spark.sql(...)` inside a FlowFunction. For these calls, instead of eagerly executing the SQL, the Spark Connect server should return the raw logical plan to the client and defer execution to the flow function. However, in that PR we constructed the response object but forgot to actually return it to the Spark Connect client, so the client received an empty response. This went unnoticed in tests because, when the client sees an empty `spark.sql(...)` response, [it falls back to creating an empty DataFrame holding the raw logical plan](https://github.com/apache/spark/blob/master/python/pyspark/sql/connect/session.py#L829-L835), which happens to match the desired behavior. This PR fixes the bug by returning the proper response instead of relying on that implicit fallback. ### Why are the changes needed? ### Does this PR introduce _any_ user-facing change? This PR fixes a bug introduced in apache#53024 where the server did not return the constructed spark.sql(...) response to the client. ### How was this patch tested? New tests ### Was this patch authored or co-authored using generative AI tooling? No Closes apache#53156 from SCHJonathan/jonathan-chang_data/fix-spark-sql-bug. Authored-by: Yuheng Chang <[email protected]> Signed-off-by: Dongjoon Hyun <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
In PR #53024, we added SDP support for
spark.sql(...)inside a FlowFunction. For these calls, instead of eagerly executing the SQL, the Spark Connect server should return the raw logical plan to the client and defer execution to the flow function.However, in that PR we constructed the response object but forgot to actually return it to the Spark Connect client, so the client received an empty response.
This went unnoticed in tests because, when the client sees an empty
spark.sql(...)response, it falls back to creating an empty DataFrame holding the raw logical plan, which happens to match the desired behavior. This PR fixes the bug by returning the proper response instead of relying on that implicit fallback.Why are the changes needed?
Does this PR introduce any user-facing change?
This PR fixes a bug introduced in #53024 where the server did not return the constructed spark.sql(...) response to the client.
How was this patch tested?
New tests
Was this patch authored or co-authored using generative AI tooling?
No