topic Liquid clustering with boolean columns in Data Engineering

Liquid clustering with boolean columns

Karin — Wed, 08 Nov 2023 12:53:03 GMT

Hi community 👋

Is it possible to use boolean columns as cluster keys for liquid clustering on Delta Tables?

I've been trying to set a boolean column as cluster key since it's one of my most common queries when reading from the table. I'm getting the error "DeltaAnalysisException: Liquid clustering requires clustering columns to have stats. Couldn't find clustering column *boolean_column_name* in stats schema" - which doesn't make sense since I've made sure to collect statistics on all columns (delta.dataSkippingNumIndexedCols = -1) and can see stats on the boolean columns as well.

I came across this post by Miles Cole, saying that it's not possible to have boolean data types as clustering keys but it's not stated anywhere in the official documentation.

Any input is appreciated!

Re: Liquid clustering with boolean columns

jeroenvs — Wed, 13 Dec 2023 13:18:05 GMT

Can confirm that boolean columns are note allowed for liquid clustering. This seems to be undocumented and the error message is not helpful: "couldn't find clustering column in stats schema"

Re: Liquid clustering with boolean columns

SFDataEng — Tue, 09 Sep 2025 16:15:49 GMT

I saw that boolean columns are not supported by liquid clustering, too. Why? In any case:

By now there is an error that can get raised called DELTA_CLUSTERING_COLUMNS_DATATYPE_NOT_SUPPORTED.

There is also by now documentation of column datatypes that are supported by liquid clustering:

Clustering supports the following data types for clustering keys:
Date
Timestamp
TimestampNTZ (requires Databricks Runtime 14.3 LTS or above)
String
Integer
Long
Short
Float
Double
Decimal
Byte