tests.system.aiplatform.test_experiment_model.TestExperimentModel: test_sklearn_model failed

[flaky-bot]

This test failed!

To configure my behavior, see the Flaky Bot documentation.

If I'm commenting on this issue too often, add the flakybot: quiet label and
I will stop commenting.

---

commit: 440ca5b
buildURL: Build Status, Sponge
status: failed

Test output

self = 
shared_state = {'bucket': , 'staging_bucket_name': 'test-ver...ent-model-955215c8-01d3-4726-a088-25', 'storage_client': }
def test_sklearn_model(self, shared_state):
    aiplatform.init(
        project=e2e_base._PROJECT,
        location=e2e_base._LOCATION,
        staging_bucket=shared_state["staging_bucket_name"],
    )

    train_x = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
    train_y = np.dot(train_x, np.array([1, 2])) + 3
    model = LinearRegression()
    model.fit(train_x, train_y)

    # Test save sklearn model


  aiplatform.save_model(model, "sk-model")


tests/system/aiplatform/test_experiment_model.py:53:

google/cloud/aiplatform/metadata/_models.py:508: in save_model

gcs_utils.upload_to_gcs(temp_dir, uri)

google/cloud/aiplatform/utils/gcs_utils.py:78: in upload_to_gcs

destination_blob = storage.Blob.from_string(

cls = <class 'google.cloud.storage.blob.Blob'>

uri = 'test-vertex-sdk-e2e-experiment-model-955215c8-01d3-4726-a088-25/2023-01-24-00-42-27-03ace-sklearn-model/model.pkl'

client = <google.cloud.storage.client.Client object at 0x7f1d63cb5370>
@classmethod
def from_string(cls, uri, client=None):
    """Get a constructor for blob object by URI.

    .. code-block:: python

        from google.cloud import storage
        from google.cloud.storage.blob import Blob
        client = storage.Client()
        blob = Blob.from_string("gs://bucket/object", client=client)

    :type uri: str
    :param uri: The blob uri pass to get blob object.

    :type client: :class:`~google.cloud.storage.client.Client`
    :param client:
        (Optional) The client to use.  Application code should
        *always* pass ``client``.

    :rtype: :class:`google.cloud.storage.blob.Blob`
    :returns: The blob object created.
    """
    from google.cloud.storage.bucket import Bucket

    scheme, netloc, path, query, frag = urlsplit(uri)
    if scheme != "gs":


      raise ValueError("URI scheme must be gs")


E           ValueError: URI scheme must be gs
.nox/system-3-8/lib/python3.8/site-packages/google/cloud/storage/blob.py:408: ValueError

[flaky-bot]

commit: 24cdecf
buildURL: Build Status, Sponge
status: failed

Test output

args = (parent: "projects/ucaip-sample-tests/locations/us-central1/metadataStores/default"
artifact {
  uri: "gs://test-verte...{
      key: "modelFile"
      value {
        string_value: "model.pkl"
      }
    }
  }
}
artifact_id: "sk-model"
,)
kwargs = {'metadata': [('x-goog-request-params', 'parent=projects/ucaip-sample-tests/locations/us-central1/metadataStores/default'), ('x-goog-api-client', 'model-builder/1.21.0 gl-python/3.8.13 grpc/1.47.0 gax/1.32.0 gapic/1.21.0')]}
@six.wraps(callable_)
def error_remapped_callable(*args, **kwargs):
    try:


      return callable_(*args, **kwargs)


.nox/system-3-8/lib/python3.8/site-packages/google/api_core/grpc_helpers.py:67:

self = <grpc._channel._UnaryUnaryMultiCallable object at 0x7f47919d11c0>

request = parent: "projects/ucaip-sample-tests/locations/us-central1/metadataStores/default"

artifact {

uri: "gs://test-vertex...s {

key: "modelFile"

value {

string_value: "model.pkl"

}

}

}

}

artifact_id: "sk-model"
timeout = None

metadata = [('x-goog-request-params', 'parent=projects/ucaip-sample-tests/locations/us-central1/metadataStores/default'), ('x-goog-api-client', 'model-builder/1.21.0 gl-python/3.8.13 grpc/1.47.0 gax/1.32.0 gapic/1.21.0')]

credentials = None, wait_for_ready = None, compression = None
def __call__(self,
             request,
             timeout=None,
             metadata=None,
             credentials=None,
             wait_for_ready=None,
             compression=None):
    state, call, = self._blocking(request, timeout, metadata, credentials,
                                  wait_for_ready, compression)


  return _end_unary_response_blocking(state, call, False, None)


.nox/system-3-8/lib/python3.8/site-packages/grpc/_channel.py:946:

state = <grpc._channel._RPCState object at 0x7f4791964730>

call = <grpc._cython.cygrpc.SegregatedCall object at 0x7f4791969140>

with_call = False, deadline = None
def _end_unary_response_blocking(state, call, with_call, deadline):
    if state.code is grpc.StatusCode.OK:
        if with_call:
            rendezvous = _MultiThreadedRendezvous(state, call, None, deadline)
            return state.response, rendezvous
        else:
            return state.response
    else:


      raise _InactiveRpcError(state)


E           grpc._channel._InactiveRpcError: <_InactiveRpcError of RPC that terminated with:

E           	status = StatusCode.ALREADY_EXISTS

E           	details = "Artifact with name projects/580378083368/locations/us-central1/metadataStores/default/artifacts/sk-model already exists"

E           	debug_error_string = "{"created":"@1674529088.936859322","description":"Error received from peer ipv4:74.125.197.95:443","file":"src/core/lib/surface/call.cc","file_line":966,"grpc_message":"Artifact with name projects/580378083368/locations/us-central1/metadataStores/default/artifacts/sk-model already exists","grpc_status":6}"

E           >
.nox/system-3-8/lib/python3.8/site-packages/grpc/_channel.py:849: _InactiveRpcError
The above exception was the direct cause of the following exception:
self = <tests.system.aiplatform.test_experiment_model.TestExperimentModel object at 0x7f4794660220>

shared_state = {'bucket': <Bucket: test-vertex-sdk-e2e-experiment-model-8e14b52b-111f-4fc4-93e8-ea>, 'staging_bucket_name': 'test-ver...ent-model-8e14b52b-111f-4fc4-93e8-ea', 'storage_client': <google.cloud.storage.client.Client object at 0x7f47939eeca0>}
def test_sklearn_model(self, shared_state):
    aiplatform.init(
        project=e2e_base._PROJECT,
        location=e2e_base._LOCATION,
        staging_bucket=f"gs://{shared_state['staging_bucket_name']}",
    )

    train_x = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
    train_y = np.dot(train_x, np.array([1, 2])) + 3
    model = LinearRegression()
    model.fit(train_x, train_y)

    # Test save sklearn model


  aiplatform.save_model(model, "sk-model")


tests/system/aiplatform/test_experiment_model.py:53:

google/cloud/aiplatform/metadata/_models.py:520: in save_model

model_artifact.create(

google/cloud/aiplatform/metadata/schema/base_artifact.py:186: in create

new_artifact_instance = artifact.Artifact.create(

google/cloud/aiplatform/metadata/artifact.py:352: in create

return cls._create(

google/cloud/aiplatform/metadata/artifact.py:203: in _create

resource = cls._create_resource(

google/cloud/aiplatform/metadata/artifact.py:112: in _create_resource

return client.create_artifact(

google/cloud/aiplatform_v1/services/metadata_service/client.py:1159: in create_artifact

response = rpc(

.nox/system-3-8/lib/python3.8/site-packages/google/api_core/gapic_v1/method.py:145: in call

return wrapped_func(*args, **kwargs)

.nox/system-3-8/lib/python3.8/site-packages/google/api_core/grpc_helpers.py:69: in error_remapped_callable

six.raise_from(exceptions.from_grpc_error(exc), exc)

value = None

from_value = <_InactiveRpcError of RPC that terminated with:

status = StatusCode.ALREADY_EXISTS

details = "Artifact with name pro...ojects/580378083368/locations/us-central1/metadataStores/default/artifacts/sk-model already exists","grpc_status":6}"



???

E   google.api_core.exceptions.AlreadyExists: 409 Artifact with name projects/580378083368/locations/us-central1/metadataStores/default/artifacts/sk-model already exists

:3: AlreadyExists

[flaky-bot]

commit: 2eb88a7
buildURL: Build Status, Sponge
status: failed

Test output

self = 
shared_state = {'bucket': , 'resources': [}
def test_sklearn_model(self, shared_state):
    aiplatform.init(
        project=e2e_base._PROJECT,
        location=e2e_base._LOCATION,
        staging_bucket=f"gs://{shared_state['staging_bucket_name']}",
    )

    train_x = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
    train_y = np.dot(train_x, np.array([1, 2])) + 3
    model = LinearRegression()
    model.fit(train_x, train_y)

    # Test save sklearn model
    aiplatform.save_model(model, "sk-model")

    # Test get ExperimentModel with aritfact id
    model_artifact = aiplatform.get_experiment_model("sk-model")
    assert model_artifact.uri.endswith("sklearn-model")

    shared_state["resources"] = [model_artifact]

    # Test get model info from ExperimentModel
    model_info = model_artifact.get_model_info()
    assert model_info == {
        "model_class": "sklearn.linear_model._base.LinearRegression",
        "framework_name": "sklearn",
        "framework_version": sklearn.__version__,
    }

    # Test load model and make prediction
    loaded_model = model_artifact.load_model()
    preds = loaded_model.predict(train_x)
    assert isinstance(preds, np.ndarray)

    # Test register model


  registered_model = model_artifact.register_model()


tests/system/aiplatform/test_experiment_model.py:75:

google/cloud/aiplatform/metadata/schema/google/artifact_schema.py:989: in register_model

return _models.register_model(

google/cloud/aiplatform/metadata/_models.py:803: in register_model

serving_container_image_uri = helpers._get_closest_match_prebuilt_container_uri(

framework = 'sklearn', framework_version = <Version('1.2.1')>, region = 'us'

accelerator = 'cpu'
def _get_closest_match_prebuilt_container_uri(
    framework: str,
    framework_version: str,
    region: Optional[str] = None,
    accelerator: str = "cpu",
) -> str:
    """Return a pre-built container uri that is suitable for a specific framework and version.

    If there is no exact match for the given version, the closest one that is
    higher than the input version will be used.

    Args:
        framework (str):
            Required. The ML framework of the pre-built container. For example,
            `"tensorflow"`, `"xgboost"`, or `"sklearn"`
        framework_version (str):
            Required. The version of the specified ML framework as a string.
        region (str):
            Optional. AI region or multi-region. Used to select the correct
            Artifact Registry multi-region repository and reduce latency.
            Must start with `"us"`, `"asia"` or `"europe"`.
            Default is location set by `aiplatform.init()`.
        accelerator (str):
            Optional. The type of accelerator support provided by container. For
            example: `"cpu"` or `"gpu"`
            Default is `"cpu"`.

    Returns:
        A string representing the pre-built container uri.

    Raises:
        ValueError: If the framework doesn't have suitable pre-built container.
    """
    URI_MAP = prediction._SERVING_CONTAINER_URI_MAP
    DOCS_URI_MESSAGE = (
        f"See {prediction._SERVING_CONTAINER_DOCUMENTATION_URL} "
        "for complete list of supported containers"
    )

    # If region not provided, use initializer location
    region = region or initializer.global_config.location
    region = region.split("-", 1)[0]
    framework = framework.lower()

    if not URI_MAP.get(region):
        raise ValueError(
            f"Unsupported container region `{region}`, supported regions are "
            f"{', '.join(URI_MAP.keys())}. "
            f"{DOCS_URI_MESSAGE}"
        )

    if not URI_MAP[region].get(framework):
        raise ValueError(
            f"No containers found for framework `{framework}`. Supported frameworks are "
            f"{', '.join(URI_MAP[region].keys())} {DOCS_URI_MESSAGE}"
        )

    if not URI_MAP[region][framework].get(accelerator):
        raise ValueError(
            f"{framework} containers do not support `{accelerator}` accelerator. Supported accelerators "
            f"are {', '.join(URI_MAP[region][framework].keys())}. {DOCS_URI_MESSAGE}"
        )

    framework_version = version.Version(framework_version)
    available_version_list = [
        version.Version(available_version)
        for available_version in URI_MAP[region][framework][accelerator].keys()
    ]
    try:
        closest_version = min(
            [
                available_version
                for available_version in available_version_list
                if available_version >= framework_version
                # manually implement Version.major for packaging < 20.0
                and available_version._version.release[0]
                == framework_version._version.release[0]
            ]
        )
    except ValueError:


      raise ValueError(


            f"You are using `{framework}` version `{framework_version}`. "
            f"Vertex pre-built containers support up to `{framework}` version "
            f"`{max(available_version_list)}` and don't assume forward compatibility. "
            f"Please build your own custom container. {DOCS_URI_MESSAGE}"
        ) from None

E           ValueError: You are using sklearn version 1.2.1. Vertex pre-built containers support up to sklearn version 1.0 and don't assume forward compatibility. Please build your own custom container. See https://cloud.google.com/vertex-ai/docs/predictions/pre-built-containers for complete list of supported containers
google/cloud/aiplatform/helpers/container_uri_builders.py:210: ValueError