What happened:
Attempted to use a.fillna(b) on a Dask DataFrame a with another Dask DataFrame b provided as the fill value. This raised the error
ValueError: invalid fill value with a <class 'numpy.ndarray'>

What you expected to happen:
A new Dask DataFrame returned containing the non-nan values of a and nan values replaced by the equivalent values from b

Minimal Complete Verifiable Example:

import numpy as np
import pandas as pd
import dask.dataframe as dd

a=dd.from_pandas(pd.DataFrame({"A":[1,2,np.nan,4,5],"B":1}),npartitions=2)
b=dd.from_pandas(pd.DataFrame({"A":[1,2,3,4,5],"B":2}),npartitions=2)
a.fillna(b)

Anything else we need to know?:

Environment:

• Dask version: '2022.02.1'
• Python version: 3.8.10
• Operating System: Windows 10
• Install method (conda, pip, source): pip

A workaround is to instead use

a.fillna(b.compute())

but this seems sub-optimal for large datasets.