Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Re-enable spark.rapids.shims-provider-override #3497

Merged
merged 1 commit into from
Sep 16, 2021

Conversation

gerashegalov
Copy link
Collaborator

@gerashegalov gerashegalov commented Sep 15, 2021

This PR closes #3495

  • The class name provided can be an arbitrary implementation of
    SparkShimServiceProvider
  • If the Shim is tested with other shims together and requires its own Parallel
    World, then it should follow the same conventions as existing Shims:
    the last component in its package name should be the root directory in
    the combined rapids-4-spark jar. If tested in isolation from other
    Shims, the standard jar layout can be used

Tested with

SPARK_HOME=~/dist/spark-3.0.1-bin-hadoop3.2 rapids.sh \
  --conf spark.rapids.shims-provider-override=com.nvidia.spark.rapids.shims.spark303.SparkShimServiceProvider

since we know from the dedupe work that Spark 3.0.x until 3.0.4 can be served with a single shim

Signed-off-by: Gera Shegalov gera@apache.org

- The class name provided can be an arbitrary implementation of
  SparkShimServiceProvider
- If the Shim is tested with other shims together and requires its own Parallel
  World, then it should follow the same conventions as existing Shims:
  the last component in its package name should be the root directory in
  the combined rapids-4-spark jar. If tested in isolation from other
  Shims, the standard jar layout can be used

Signed-off-by: Gera Shegalov <gera@apache.org>
@gerashegalov gerashegalov added the bug Something isn't working label Sep 15, 2021
@gerashegalov gerashegalov added this to the Sep 13 - Sep 24 milestone Sep 15, 2021
@gerashegalov gerashegalov self-assigned this Sep 15, 2021
@gerashegalov
Copy link
Collaborator Author

build

@tgravescs
Copy link
Collaborator

would be great to test on cdh

@tgravescs
Copy link
Collaborator

13:53:59 An unexpected error has occurred: CalledProcessError: command: ('/usr/bin/git', 'fetch', 'origin', '--tags')

@tgravescs
Copy link
Collaborator

build

@tgravescs
Copy link
Collaborator

14:00:13 stderr: fatal: unable to access 'https://github.com/NVIDIA/spark-rapids.git/': The requested URL returned error: 429

@tgravescs
Copy link
Collaborator

build

@gerashegalov gerashegalov merged commit d6ab561 into NVIDIA:branch-21.10 Sep 16, 2021
@gerashegalov gerashegalov deleted the shimLoaderOverride branch September 16, 2021 00:32
@gerashegalov
Copy link
Collaborator Author

did not test on CDH yet. will fix in a dedicated PR if it's broken there.

@gerashegalov
Copy link
Collaborator Author

Thanks @abellina for testing on CDH

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Fix shim override config
2 participants