Hi, Facing the below error:

Metaflow 2.12.20+netflix-ext(1.2.2) executing CombinedBELModelFlow for user:gauravmudbhatkal
Validating your flow...
  The graph looks good!
Bootstrapping Conda environment... (this could take a few minutes)
Pretty-printed STDERR:
ERROR: Ignored the following versions that require a different python version: 8.19.0 Requires-Python >=3.10; 8.20.0 Requires-Python >=3.10; 8.21.0 Requires-Python >=3.10; 8.22.0 Requires-Python >=3.10; 8.22.1 Requires-Python >=3.10; 8.22.2 Requires-Python >=3.10; 8.23.0 Requires-Python >=3.10; 8.24.0 Requires-Python >=3.10; 8.25.0 Requires-Python >=3.10; 8.26.0 Requires-Python >=3.10; 8.27.0 Requires-Python >=3.10
ERROR: Could not find a version that satisfies the requirement pyod>=1.1.3 (from pycaret) (from versions: none)
ERROR: No matching distribution found for pyod>=1.1.3

Using the latest version of metaflow & metaflow netflix extension.

@pypi_base(
    packages={
        "sentence-transformers": "3.0.1",
        "matplotlib": "",
        "pandas": "1.5.3",
        "numpy": "",
        "psycopg2-binary": "2.9.9",
        "transformers": "4.41.2",
        "tensorflow": "2.11.0",
        "pycaret": "3.3.2",
        "torch": "",
        "scikit-learn": "",
        "seaborn": "0.13.2",
    },
    python="3.9.13",
)
class MLModelFlow(DBMixin, Config, FlowSpec):
      ....

FYI - Using micromamba

Sorry for the delay. I tried your environment and it seems to work ok. I get this:

Environment of type pypi-only full hash 0a30699962bc110cda178213e3a5dd85552d7c0e:88c4eaade6b24ad988b35d30712edf0a203fb4aa
Arch linux-64
Available on linux-64

Resolved on 2024-09-24 22:01:05.289297
Resolved by rcledat

User-requested packages sys::__glibc==2.35=0, pypi::boto3==>=1.14.0, pypi::cffi==>=1.13.0,!=1.15.0, pypi::fastavro==>=1.6.0, pypi::ipykernel, pypi::matplotlib, pypi::numpy, pypi::pandas==>=0.24.0,==1.5.3, pypi::psycopg2-binary==2.9.9, pypi::pyarrow==>=0.17.1, pypi::pycaret==3.3.2, conda::python==3.9.13, pypi::requests==>=2.21.0, pypi::scikit-learn, pypi::seaborn==0.13.2, pypi::sentence-transformers==3.0.1, pypi::tensorflow==2.11, pypi::torch, pypi::transformers==4.41.2
User sources conda::conda-forge, pypi::<https://pypi.netflix.net/simple>

Conda Packages installed _libgcc_mutex==0.1-conda_forge, _openmp_mutex==4.5-2_gnu, bzip2==1.0.8-h4bc722e_7, ca-certificates==2024.8.30-hbcca054_0, ld_impl_linux-64==2.43-h712a8e2_0, libffi==3.4.2-h7f98852_5, libgcc==14.1.0-h77fa898_1, libgcc-ng==14.1.0-h69a702a_1, libgomp==14.1.0-h77fa898_1, libnsl==2.0.1-hd590300_0, libsqlite==3.46.1-hadc24fc_0, libuuid==2.38.1-h0b41bf4_0, libzlib==1.3.1-h4ab18f5_1, ncurses==6.5-he02047a_1, openssl==3.3.2-hb9d3cd8_0, pip==24.2-pyh8b19718_1, python==3.9.13-h2660328_0_cpython, readline==8.2-h8228510_1, setuptools==74.1.2-pyhd8ed1ab_0, sqlite==3.46.1-h9eae976_0, tk==8.6.13-noxft_h4845f30_101, tomli==2.0.1-pyhd8ed1ab_0, tzdata==2024a-h8827d51_1, wheel==0.42.0-pyhd8ed1ab_0, xz==5.2.6-h166bdaf_0
Pypi Packages installed absl-py==2.1.0, asttokens==2.4.1, astunparse==1.6.3, attrs==24.2.0, blinker==1.8.2, boto3==1.35.26, botocore==1.35.26, cachetools==5.5.0, category-encoders==2.6.3, certifi==2024.8.30, cffi==1.17.1, charset-normalizer==3.3.2, click==8.1.7, cloudpickle==3.0.0, comm==0.2.2, contourpy==1.3.0, cycler==0.12.1, cython==3.0.11, dash==2.18.1, dash-core-components==2.0.0, dash-html-components==2.0.0, dash-table==5.0.0, debugpy==1.8.6, decorator==5.1.1, deprecation==2.1.0, exceptiongroup==1.2.2, executing==2.1.0, fastavro==1.9.7, fastjsonschema==2.20.0, filelock==3.16.1, flask==3.0.3, flatbuffers==24.3.25, fonttools==4.54.1, fsspec==2024.9.0, gast==0.4.0, google-auth==2.35.0, google-auth-oauthlib==0.4.6, google-pasta==0.2.0, grpcio==1.66.1, h5py==3.11.0, huggingface-hub==0.25.1, idna==3.10, imbalanced-learn==0.12.3, importlib-metadata==8.5.0, importlib-resources==6.4.5, ipykernel==6.29.5, ipython==8.18.1, ipywidgets==8.1.5, itsdangerous==2.2.0, jedi==0.19.1, jinja2==3.1.4, jmespath==1.0.1, joblib==1.3.2, jsonschema==4.23.0, jsonschema-specifications==2023.12.1, jupyter-client==8.6.3, jupyter-core==5.7.2, jupyterlab-widgets==3.0.13, kaleido==0.2.1, keras==2.11.0, kiwisolver==1.4.7, libclang==18.1.1, lightgbm==4.5.0, llvmlite==0.43.0, markdown==3.7, markupsafe==2.1.5, matplotlib==3.7.5, matplotlib-inline==0.1.7, mpmath==1.3.0, nbformat==5.10.4, nest-asyncio==1.6.0, networkx==3.2.1, numba==0.60.0, numpy==1.26.4, nvidia-cublas-cu12==12.1.3.1, nvidia-cuda-cupti-cu12==12.1.105, nvidia-cuda-nvrtc-cu12==12.1.105, nvidia-cuda-runtime-cu12==12.1.105, nvidia-cudnn-cu12==9.1.0.70, nvidia-cufft-cu12==11.0.2.54, nvidia-curand-cu12==10.3.2.106, nvidia-cusolver-cu12==11.4.5.107, nvidia-cusparse-cu12==12.1.0.106, nvidia-nccl-cu12==2.20.5, nvidia-nvjitlink-cu12==12.6.68, nvidia-nvtx-cu12==12.1.105, oauthlib==3.2.2, opt-einsum==3.3.0, orjson==3.10.7, packaging==24.1, pandas==1.5.3, parso==0.8.4, patsy==0.5.6, pexpect==4.9.0, pillow==10.4.0, platformdirs==4.3.6, plotly==5.24.1, plotly-resampler==0.10.0, pmdarima==2.0.4, prompt-toolkit==3.0.47, protobuf==3.19.6, psutil==6.0.0, psycopg2-binary==2.9.9, ptyprocess==0.7.0, pure-eval==0.2.3, pyarrow==17.0.0, pyasn1==0.6.1, pyasn1-modules==0.4.1, pycaret==3.3.2, pycparser==2.22, pygments==2.18.0, pyod==2.0.2, pyparsing==3.1.4, python-dateutil==2.9.0.post0, pytz==2024.2, pyyaml==6.0.2, pyzmq==26.2.0, referencing==0.35.1, regex==2024.9.11, requests==2.32.3, requests-oauthlib==2.0.0, retrying==1.3.4, rpds-py==0.20.0, rsa==4.9, s3transfer==0.10.2, safetensors==0.4.5, schemdraw==0.15, scikit-base==0.7.8, scikit-learn==1.4.2, scikit-plot==0.3.7, scipy==1.11.4, seaborn==0.13.2, sentence-transformers==3.0.1, setuptools==75.1.0, six==1.16.0, sktime==0.26.0, stack-data==0.6.3, statsmodels==0.14.3, sympy==1.13.3, tbats==1.1.3, tenacity==9.0.0, tensorboard==2.11.2, tensorboard-data-server==0.6.1, tensorboard-plugin-wit==1.8.1, tensorflow==2.11.0, tensorflow-estimator==2.11.0, tensorflow-io-gcs-filesystem==0.37.1, termcolor==2.4.0, threadpoolctl==3.5.0, tokenizers==0.19.1, torch==2.4.1, tornado==6.4.1, tqdm==4.66.5, traitlets==5.14.3, transformers==4.41.2, triton==3.0.0, tsdownsample==0.1.3, typing-extensions==4.12.2, urllib3==1.26.20, wcwidth==0.2.13, werkzeug==3.0.4, wheel==0.44.0, widgetsnbextension==4.0.13, wrapt==1.16.0, wurlitzer==3.1.1, xxhash==3.5.0, yellowbrick==1.5, zipp==3.20.2

Dry-run -- not caching or aliasing

Could you try with METAFLOW_DEBUG_CONDA=1 and see what you get. Given these are just pypi dependencies, the error is coming from pip directly which is somehow not able to resolve your rquirements.

Here's the logs

ah — you are doing cross platform. Unfortunately, it won’t work because pyod does not have any wheels in pypi (and cross platform requires wheels only — this is a pip requirement). You have two solutions: • move to using a

@conda

decorator (briefly looking it looks like most packages are available there) • using a separate node to build your environment (ie: build it in batch). Let me find you a thread for that second option

Thanks @dry-beach-38304, will look into it.

I suspect this is because this is micromamba 2.0? I think they changed the syntax they accept. I need to make a small change to fix it (ie not do ==>= which used to work but seems to no longer be accepted).

Yea, its 2.0. Re-installed micromamba completely as something got messed up and there's a new version of it 😅 Thanks for the help @dry-beach-38304. Will try to go back to the older version for now

I’ll update the code to fix this. I think it’s simply removing the ==. I do this for pypi already and will just need to do it here as well. I just want to test if that new format will work on the old micromamba and the new one. I’ll do that tomorrow hopefully. Shouldn’t be hard. Thanks for pointing this out.

Is there a way to use

requirements.txt

or

environemt.yml

with a flow decorator instead of resolving and then using the namedenv? Something like:

@conda_base(file="./env.yaml")
class Flow(FlowSpec):
     .....

this has been asked before and it’s not there yet but it should be out very soon. The plan will be to support it with something like this: https://github.com/Netflix/metaflow/pull/1962 and have a

Config

that defines your environment file and we will provide the parser method so you can then do something like:

from metaflow import conda_base, Flow, Config

from metaflow.config_parsers import CondaYaml # exact syntax TBD)

@conda_base(**config_expr("env_spec"))
class Flow(FlowSpec):

  env_spec = Config("env_spec", default="./env.yaml", parser=CondaYaml)

Wow, this is amazing. Excited to try it out sometime.

does your configuration define a specific channel priority?

This is how the channels are passed to the

conda_base

decorator.

channels=["conda-forge", "defaults"],

Not passing anything else in the cmd line

do you have any .condarc or anything else there?

Nope, havent set any.

could you try:

/home/sln/.local/bin/micromamba config list -a

and let me know what it says for the channel priority thing?

Its strict

channel_priority: strict

ah ha

Will try to disable it and try it out again.

micromamba config --set channel_priority false

Is it a default value or stored somewhere in some rc/config file?

I have it to

flexible

. I think micromamba can tell you where the configs are coming from.

Got the path. Changing to flexible. Thank you!

but this is great feedback — I would have thought that the env variable would override it but clearly not. If you get a chance (and have some time), I would appreciate if you could run a few of these commands and let me know what happens (with your old

strict

setting in your config): The base command is basically what is given here:

['/home/sln/.local/bin/micromamba', 'create', '--yes', '--prefix', '/tmp/tmp2urgrafd/prefix', '--dry-run', '-c', 'defaults', 'pydantic-settings==2.1.0', 'sqlalchemy==2.0.28', 'matplotlib', 'transformers==4.41.2', 'nltk==3.8.1', 'spacy==3.7.4', 'openai-clip==1.0.1', 'numpy==1.24.0', 'python==3.9.13', 'sentence-transformers==3.0.1', 'pydantic==2.5.3', 'boto3==>=1.14.0', 'jsonschema==4.22.0', 'pytorch', 'pandas==1.5.3', 'urllib3==1.26.18', 'pillow==10.1.0', 'joblib==1.2.0', 'langchain==0.2.5', 'langchain-openai==0.1.17', 'pgvector-python==0.2.5', 'pycaret==3.3.1', 'seaborn==0.11.2', 'xgboost==2.0.3', 'requests==2.31.0,>=2.21.0', 'psycopg2-binary==2.9.9', 'fastapi', 'alembic==1.13.1', '-r', '/home/sln/micromamba', '--json']

Run three things: • Just like that • CONDA_CHANNEL_PRIORITY=flexible • adding the --channel-priority 1 to the command line I currently do the second option (I think) and I’d be curious to see if the third one works. If so I’ll switch to that for micromamba (I think mamba already respects the env var)

Sure, will try it in sometime

flexible instead of 1 if that doesn’t work.

Looks like setting in the cmd line works, but not env variable

--channel-priority flexible

Awesome. That’s the issue then. Interesting it doesn’t respect it. Annoying that these things change so much between executables. I’ll fix it.

Im hoping its not just with micromamba and happening with others like conda/mamba.

mamba respects it (well pre 2.0) — I am adding support for 2.0 and a bunch of things have changed. The good news is that now it looks like micromamba and mamba behave the same (like micromamba)

Happens sometimes with micromamba aswell. It works almost immediately.