Alice Sayutina created SPARK-45093:
--------------------------------------
Summary: AddArtifacts should give proper error messages if it fails
Key: SPARK-45093
URL: https://issues.apache.org/jira/browse/SPARK-45093
Project: Spark
Issue Type: Improvement
Components: PySpark
Affects Versions: 3.5.0
Reporter: Alice Sayutina
I've been trying to do some testing of udf's using code in other module, so
that AddArtifact is necessary.
I got the following error:
{code:java}
Traceback (most recent call last):
File "/Users/alice.sayutina/db-connect-playground/udf2.py", line 5, in
<module>
spark.addArtifacts("udf2_support.py", pyfile=True)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/pyspark/sql/connect/session.py",
line 744, in addArtifacts
self._client.add_artifacts(*path, pyfile=pyfile, archive=archive, file=file)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/pyspark/sql/connect/client/core.py",
line 1582, in add_artifacts
self._artifact_manager.add_artifacts(*path, pyfile=pyfile, archive=archive,
file=file)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/pyspark/sql/connect/client/artifact.py",
line 283, in add_artifacts
self._request_add_artifacts(requests)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/pyspark/sql/connect/client/artifact.py",
line 259, in _request_add_artifacts
response: proto.AddArtifactsResponse = self._retrieve_responses(requests)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/pyspark/sql/connect/client/artifact.py",
line 256, in _retrieve_responses
return self._stub.AddArtifacts(requests, metadata=self._metadata)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/grpc/_channel.py",
line 1246, in __call__
return _end_unary_response_blocking(state, call, False, None)
File
"/Users/alice.sayutina/db-connect-venv/lib/python3.10/site-packages/grpc/_channel.py",
line 910, in _end_unary_response_blocking
raise _InactiveRpcError(state) # pytype: disable=not-instantiable
grpc._channel._InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
status = StatusCode.UNKNOWN
details = "Exception iterating requests!"
debug_error_string = "None"
{code}
Which doesn't give any clue about what happens.
Only after noticeable investigation I found the problem: I'm specifying the
wrong path and the artifact fails to upload. Specifically what happens is that
ArtifactManager doesn't read the file immediately, but rather creates iterator
object which will generate requests to send. This iterator is passed to grpc's
stream_unary to consume and actually send, and while grpc catches the error
(see above), it suppresses the underlying exception.
I think we should improve pyspark user experience. One of the possible ways to
fix this is to wrap ArtifactsManager._create_requests with an iterator wrapper
which would log the throwable into spark connect logger so that user would see
something like at least when the debug mode is on.
{code:java}
FileNotFoundError: [Errno 2] No such file or directory:
'/Users/alice.sayutina/udf2_support.py' {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
