[GLUTEN-9801] Clean up Write Files commit Protocol logic #9844
+103
−144
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
|
@ulysses-you Can you help to review this PR? Thanks. |
JkSelf
force-pushed
the
delete-failed-task
branch
from
40f2dd8 to
a538513
Compare
|
@RushabhK, could you have a try again? |
|
@JkSelf just curious, why we can't let spark handle this for Gluten? |
ulysses-you
reviewed
Jun 4, 2025
| @@ -119,6 +122,8 @@ class VeloxColumnarWriteFilesRDD( | |||
| val targetFileName = fileWriteInfo.targetFileName | |||
| val outputPath = description.path | |||
|
|
|||
| fileNames += targetFileName | |||
There was a problem hiding this comment.
it seems writePath does not contain partition partitionFragment ? we should collect partitionFragment + "/" + targetFileName
There was a problem hiding this comment.
@ulysses-you Yes. Updated.
JkSelf
force-pushed
the
delete-failed-task
branch
from
a538513 to
5a8d954
Compare
RushabhK
reviewed
Jun 4, 2025
backends-velox/src/main/scala/org/apache/spark/sql/execution/VeloxColumnarWriteFilesExec.scala
Outdated
Show resolved
Hide resolved
|
@JkSelf I tested this change on my setup. It's still giving the same exception, is not a Parquet file. Expected magic number at tail, but found [2, 0, 0, 0]. This file is ~250 MB size. This is the complete stack trace: |
@RushabhK Ok. Can you help to provide the reproduced code? Thanks. |
@JkSelf I can elaborate how I am testing this in the following steps
|
|
@RushabhK I followed the steps you provided using the main branch code, and the test code is as follows: Here is the Spark history UI for reference.
However, I was unable to reproduce the issue you mentioned. Am I missing something? |
@JkSelf I am running this on Ubuntu, what OS are you running it on? Can you share the setup for your spark image? |
Also @JkSelf I am writing the parquet in overwrite mode. I am using the following code for write: |
|
@JkSelf I added some logs for better visibility around what all files the abortTask is deleting. So is the code in the catchBlock suggests 0 files: https://github.com/RushabhK/incubator-gluten/blob/v1.3.0-fixes/backends-velox/src/main/scala/org/apache/spark/sql/execution/VeloxColumnarWriteFilesExec.scala#L244 I had added more logs to check the fileNames status at every point. This suggest the fileNames size to be 1 in all the logs: https://github.com/RushabhK/incubator-gluten/blob/v1.3.0-fixes/backends-velox/src/main/scala/org/apache/spark/sql/execution/VeloxColumnarWriteFilesExec.scala#L139 The problem is 0 files are being collected while calling the abortTask. This is the issue which needs to be addressed, this is why it's not deleting any files when the abort task is being called. |
|
@RushabhK My OS is Ubuntu 20.04.6 and spark version is 3.5.2.
|
Are you saying that |
Yes, |
JkSelf
force-pushed
the
delete-failed-task
branch
from
5a8d954 to
fb9ec0c
Compare
JkSelf
force-pushed
the
delete-failed-task
branch
from
fb9ec0c to
e14720b
Compare
JkSelf
force-pushed
the
delete-failed-task
branch
from
e14720b to
663107b
Compare
|
@RushabhK Could you help to try the latest commit again? Thanks. |
JkSelf
force-pushed
the
delete-failed-task
branch
2 times, most recently
from
8694073 to
7ce2fe2
Compare
JkSelf
changed the title
…ark's FileCommitProtocol
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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?
Velox supports third-party integration for passing the fileName, eliminating the need to define SparkWriteFilesCommitProtocol in Gluten for managing the write task commit process. Instead, Spark's FileCommitProtocol can be used directly.
With this PR, the file written by the failed task can be deleted in abortTask().
Fix #9801 (comment)
How was this patch tested?
Existing unit tests