Fix script mode training hang with logging enabled #77
Conversation
This file contains 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
laurenyu
approved these changes
Dec 3, 2019
ericangelokim
approved these changes
Dec 4, 2019
aws-patlin
added a commit
to aws-patlin/sagemaker-xgboost-container
that referenced
this pull request
Dec 5, 2019
aws-patlin
added a commit
to aws-patlin/sagemaker-xgboost-container
that referenced
this pull request
Dec 5, 2019
aws-patlin
added a commit
that referenced
this pull request
Dec 5, 2019
edwardjkim
pushed a commit
to edwardjkim/sagemaker-xgboost-container
that referenced
this pull request
Mar 17, 2021
edwardjkim
pushed a commit
to edwardjkim/sagemaker-xgboost-container
that referenced
this pull request
Mar 17, 2021
edwardjkim
pushed a commit
to edwardjkim/sagemaker-xgboost-container
that referenced
this pull request
Mar 17, 2021
edwardjkim
pushed a commit
to edwardjkim/sagemaker-xgboost-container
that referenced
this pull request
Mar 17, 2021
edwardjkim
pushed a commit
that referenced
this pull request
Mar 17, 2021
… image (#179) * Bump Python to 3.7.10 * Merge commits from 0.90-1 back to reverted master * Fix CSV Pipe parsing argument to use weight instead of weights. Fix requirements for tox. (#81) * Fix script mode training hang with logging enabled. (#77) * Fix training unit test to match PR #77. (#84) * Fix label concatenation for RecordIO-protobuf dmatrix (#85) Closes #83 * Add verbosity to hyperparameter validation. (#87) * Add verbosity to hyperparameter validation. * Set scipy requirement to 1.2.2 for sagemaker-containers. * Add missing eval_metrics to hp validation. (#82) * Added aucpr and cox-nloglik to eval_metric hp validation. * Add two separate list for MAXIMIZE and MINIMIZE metrics. Co-authored-by: ericangelokim <39601338+ericangelokim@users.noreply.github.com> Co-authored-by: Patrick Lin <52252844+aws-patlin@users.noreply.github.com> Co-authored-by: rizwangilani <rizwan.gl@gmail.com>
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.
Description of changes:
A user discovered that after a recent release, their training script stops to work when adding a stdout stream handler to the logger. The issue was traced back to this commit in sagemaker-containers.
capture_error=Trueappends stderr to the error message that gets thrown if training fails. For context, this was specifically a workaround for PyTorch, which can throw a specific error even if training succeeds, so I don't believe this is necessary for XGBoost.Corresponding PR for 0.90-2: #78
Testing:
Using the prod image with capture_error enabled, the training script would hang with no log output. With capture_error disabled on a custom image, I was able to complete the training job successfully with the expected log output.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.