Describe the issue:

Code execution stops in the example below with

PerformanceWarning: The input Dask array will be rechunked along axis 0 with chunk size 8, but a chunk size divisible by 5 is required for Dask to write safely to the Zarr array <Array memory://4933780544 shape=(8,) dtype=float64>. To avoid risk of data loss when writing to this Zarr array, set the "array.chunk-size" configuration parameter to at least the size in bytes of a single on-disk chunk (or shard) of the Zarr array, which in this case is 40 bytes. E.g., dask.config.set({"array.chunk-size": 40})

Problems:

1. What probably should be a warning is raised as an exception and stops execution:

2. In the shown example I think the warning doesn't make sense, since region= is used and in this case the dask and zarr chunks align perfectly.

Minimal Complete Verifiable Example:

import dask.array as da
import zarr

output_zarr_array = zarr.create(
    shape=(8,),
    chunks=(5,),
    overwrite=True,      # Allows overwriting if the path exists
)

x = da.ones((5,), chunks=(5,))

da.to_zarr(
    x,
    output_zarr_array,
    region=tuple([slice(0, 5)]),
    )

# performance warning RAISED

Anything else we need to know?:

Environment:

• Dask version: 2025.11.0
• Python version: 3.12
• Operating System: macos
• Install method (conda, pip, source): pip