-
Notifications
You must be signed in to change notification settings - Fork 4k
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
pyfunc.spark_udf: conda env dir / cache dir isolation and NFS optimization #5561
Conversation
Signed-off-by: Weichen Xu <weichen.xu@databricks.com>
Signed-off-by: Weichen Xu <weichen.xu@databricks.com>
Signed-off-by: Weichen Xu <weichen.xu@databricks.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@WeichenXu123 Looks great! Tried it out on Databricks & it seemed to work well! Left a few small comments.
mlflow/pyfunc/__init__.py
Outdated
# Create individual package cache dir "pkgs" under the conda_env_root_dir | ||
# for each python process. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If the user forks a new process after running this method, _CONDA_ENV_ROOT_DIR
will still be defined and the same cache dir will be reused. Shall we handle this case?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good question... This might be hard to address.
But what's the case user need to fork a process ? Is it a common case ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could be! Can't we store and check the process id? https://docs.python.org/3/library/os.html#os.getpid. If the pid doesn't match the one used to configure _CONDA_ENV_ROOT_DIR
the first time, can we overwrite it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Addressed! See cache_return_value_per_process
decorator and it is applied on get_or_create_nfs_tmp_dir
, get_or_create_tmp_dir
, _get_or_create_env_root_dir
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM once the remaining comments are addressed.
What changes are proposed in this pull request?
For pyfunc.spark_udf:
How is this patch tested?
Manually:
NFS enabled: https://e2-dogfood.staging.cloud.databricks.com/?o=6051921418418893#notebook/2674920562695789/command/2674920562695795
NFS disabled: https://e2-dogfood.staging.cloud.databricks.com/?o=6051921418418893#notebook/2674920562695847/command/2674920562695848
Does this PR change the documentation?
ci/circleci: build_doc
check. If it's successful, proceed to thenext step, otherwise fix it.
Details
on the right to open the job page of CircleCI.Artifacts
tab.docs/build/html/index.html
.Release Notes
Is this a user-facing change?
For pyfunc.spark_udf:
Make conda env dir / cache dir isolated for different python processes.
NFS optimization: Create conda env once in driver side (write env dir into NFS mounted directory), and executor side directly read the env set up in driver side.
What component(s), interfaces, languages, and integrations does this PR affect?
Components
area/artifacts
: Artifact stores and artifact loggingarea/build
: Build and test infrastructure for MLflowarea/docs
: MLflow documentation pagesarea/examples
: Example codearea/model-registry
: Model Registry service, APIs, and the fluent client calls for Model Registryarea/models
: MLmodel format, model serialization/deserialization, flavorsarea/projects
: MLproject format, project running backendsarea/scoring
: MLflow Model server, model deployment tools, Spark UDFsarea/server-infra
: MLflow Tracking server backendarea/tracking
: Tracking Service, tracking client APIs, autologgingInterface
area/uiux
: Front-end, user experience, plotting, JavaScript, JavaScript dev serverarea/docker
: Docker use across MLflow's components, such as MLflow Projects and MLflow Modelsarea/sqlalchemy
: Use of SQLAlchemy in the Tracking Service or Model Registryarea/windows
: Windows supportLanguage
language/r
: R APIs and clientslanguage/java
: Java APIs and clientslanguage/new
: Proposals for new client languagesIntegrations
integrations/azure
: Azure and Azure ML integrationsintegrations/sagemaker
: SageMaker integrationsintegrations/databricks
: Databricks integrationsHow should the PR be classified in the release notes? Choose one:
rn/breaking-change
- The PR will be mentioned in the "Breaking Changes" sectionrn/none
- No description will be included. The PR will be mentioned only by the PR number in the "Small Bugfixes and Documentation Updates" sectionrn/feature
- A new user-facing feature worth mentioning in the release notesrn/bug-fix
- A user-facing bug fix worth mentioning in the release notesrn/documentation
- A user-facing documentation change worth mentioning in the release notes