Hi All, I am using

bentoml==1.1.10

. I made a small change to the service to deploy a weekly trained model. On the bentoml build step, I received the following error:

Error: [bentoml-cli] `build` failed: Failed to import module "my_service": Dask dataframe requirements are not installed.

Please either conda or pip install as follows:

  conda install dask                     # either conda install
  python -m pip install "dask[dataframe]" --upgrade  # or python -m pip install

Anyone encounters this problem? Is there a (transient) dependency change on Dask? Any help is appreciated! Edit: I downgraded from

dask-2024.7.0

to

dask==2024.6.2

and the error went away 🤔 The newer version of dask removed pandas 2 support. Not sure if that is what

bentoml==1.1.10

depends on. BentoML 1.2 upgrade is on our roadmap.

doing

build

with lock package might help you prevent problems like this.

Hi @Jian Shen Yap I didn't see the lock package option in

bentoml build -h

.

https://docs.bentoml.org/en/latest/guides/build-options.html#pypi-package-locking ah i should be more specific. you can specify it here in the bentofile. as a results it will take a longer build time because of the dependencies traversingh