You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 6f21b06
Browse filesBrowse the repository at this point in the historyBrowse files
Xichao Wang
committed
chore(core): sync the bundled SageMaker service model with botocore
Refreshes sagemaker-core's codegen input from the public botocore model and
regenerates the layers built from it. The model file is the upstream bytes
verbatim; everything else is the output of
``python -m sagemaker.core.tools.codegen``.
Model: 403 -> 409 operations, 3005 -> 3084 shapes.
What the refresh brings in
--------------------------
* Six new operations: AttachClusterNodeNetworkInterface, and the five
training-session operations (Create, Describe, List, Stop, Delete).
* 79 new shapes, of which 19 belong to the training-session operations and 60
to other parts of the API.
* 66 existing shapes whose definitions have moved upstream, spanning
AIBenchmark, AIRecommendation, App and Cluster among others.
* NotebookInstance gains the PendingMaintenance and InMaintenance states.
* MlflowApp gains kms_key_id in the configuration schema.
* A TrainingSession resource, derived from the standard CRUD naming with no
registration needed: create / get / get_all, and delete / refresh / stop /
wait_for_delete / wait_for_status.
No caller is switched over to any newly generated resource here. This change
only brings the generated layer back in line with the model.
Notes for anyone regenerating
-----------------------------
Two things about the generator are worth knowing, since both produce a diff
that looks plausible but is wrong:
* ``codegen`` needs two passes to converge. ``generate_resources()`` reads the
existing config_schema.py to decide which resources get a
``populate_inputs_decorator``, and the schema is regenerated after it, so a
single pass emits new resources without one. Those resources then silently
ignore sagemaker_config defaults. A third pass is a no-op, so two is enough.
* ``resources_extractor`` writes resource_plan.csv to the current working
directory, so running the unit tests from sagemaker-core/ overwrites the
tracked file with a partial plan. Regenerate before diffing, not after
running tests.
Separately, codegen does not emit the module docstring at the top of
config_schema.py, which is not excluded from flake8 or docstyle. It is
preserved here.
0 commit comments