Closed lampajr closed 1 month ago
Locally I am experiencing a failure running bentoml
tests, in particular while issueing the curl
to the fraud-detection
pod. From the logs I see:
2024-05-07T08:38:31+0000 [ERROR] [api_server:1] Exception on /is_fraud [POST] (trace=fe8f5861ac58e7997bf04620e99e67ba,span=62350ba5fd974236,sampled=0)
Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/bentoml/_internal/server/http_app.py", line 324, in api_func
output = await api.func(input_data)
File "/home/bentoml/bento/src/service.py", line 34, in is_fraud
results = await asyncio.gather(*[_is_fraud(p, r, input_df) for p, r in zip(fraud_detection_preprocessors, fraud_detection_runners)])
File "/home/bentoml/bento/src/service.py", line 28, in _is_fraud
input_features = preprocessor.transform(input_df)
File "/usr/local/lib/python3.10/site-packages/sklearn/utils/_set_output.py", line 140, in wrapped
data_to_wrap = f(self, X, *args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/sklearn/compose/_column_transformer.py", line 800, in transform
Xs = self._fit_transform(
File "/usr/local/lib/python3.10/site-packages/sklearn/compose/_column_transformer.py", line 652, in _fit_transform
transformers = list(
File "/usr/local/lib/python3.10/site-packages/sklearn/compose/_column_transformer.py", line 349, in _iter
transformers = [
File "/usr/local/lib/python3.10/site-packages/sklearn/compose/_column_transformer.py", line 350, in <listcomp>
replace_passthrough(*trans) for trans in self.transformers_
File "/usr/local/lib/python3.10/site-packages/sklearn/compose/_column_transformer.py", line 345, in replace_passthrough
if name not in self._name_to_fitted_passthrough:
AttributeError: 'ColumnTransformer' object has no attribute '_name_to_fitted_passthrough'
@juliusvonkohout any idea?
Anyway, let's see if the same error happens here in the GHA
@yubozhao you need to maintain bentoml or we need to consider removing it from Kubeflow
Maybe we need to upgrade from https://github.com/bentoml/Fraud-Detection-Model-Serving
In the worst case you have to drop some test parts until it is working again. This is fine for /contrib stuff.
Maybe we need to upgrade from https://github.com/bentoml/Fraud-Detection-Model-Serving
Yeah, maybe yes as it looks like it is using docker.io/bentoml/fraud_detection:o5smnagbncigycvj
and from https://github.com/bentoml/Fraud-Detection-Model-Serving I don't see any change since last year
In the worst case you have to drop some test parts until it is working again. This is fine for /contrib stuff.
If this is acceptable, I think it is enough to skip the curl
request.. all other checks should work. See https://github.com/kubeflow/manifests/pull/2704/commits/deaf47b77cf640541a04deb43157ec5266348ae5
This is just /contrib which shall not block the release, so
/lgtm /approve
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: juliusvonkohout, lampajr
The full list of commands accepted by this bot can be found here.
The pull request process is described here
/lgtm
Which issue is resolved by this Pull Request: As suggested https://github.com/kubeflow/manifests/pull/2693#issuecomment-2097665911 I am proposing to update
betnoml
manifestsDescription of your changes:
Following the
UPGRADE.md
instructions I updatedcontrib/bentoml
with the latest version content.Checklist:
make generate-changed-only
make test