topic Re: MongoDB Spark Connector UC command is not supported without recommendation in Data Engineering

MongoDB Spark Connector UC command is not supported without recommendation

shan-databricks — Thu, 04 Jun 2026 15:00:49 GMT

I am trying to read a MongoDB collection using spark.read.format("mongodb"). However, when I attempt to display the collection, I receive the error: "UC command is not supported without recommendation." Please help resolve this issue.

Re: MongoDB Spark Connector UC command is not supported without recommendation

szymon_dybczak — Thu, 04 Jun 2026 15:46:49 GMT

Hi @shan-databricks ,

Try to use dedicated access mode instead of standard (shared). Probably mongodb connector requires access to some API which is blocked in standard access mode (i.e data source v2), but is allowed in dedicated access mode.

If my answer was helpful, please consider marking it as solution

Re: MongoDB Spark Connector UC command is not supported without recommendation

Ashwin_DSA — Thu, 04 Jun 2026 19:10:45 GMT

Hi @shan-databricks,

This is expected behaviour on Unity Catalog shared or standard compute. The mongodb Spark data source path can fail there with UC_COMMAND_NOT_SUPPORTED.WITHOUT_RECOMMENDATION / SQLSTATE 0AKUC. The public docs for that are UC_COMMAND_NOT_SUPPORTED and Standard compute requirements and limitations.

If the goal is to keep using the MongoDB Spark connector, the simplest fix is to move the workload to dedicated compute.

df = (
    spark.read.format("mongodb")
    .option("connection.uri", MONGO_URI)
    .option("database", MONGO_DB)
    .option("collection", MONGO_COLLECTION)
    .load()
)

display(df)

If the requirement is to stay on Unity Catalog compute, the alternative is a JDBC connection, which supports Spark read/write across Unity Catalog compute types once the JDBC driver and connection object are configured.

df = (
    spark.read.format("jdbc")
    .option("databricks.connection", "my_jdbc_connection")
    .option("query", "select * from my_table")
    .load()
)

display(df)

If your flow needs both MongoDB IO and UC-governed table operations, you can also consider splitting it into two stages. MongoDB read/write on dedicated compute, and the UC-governed transformations on the compute type best suited for those policies.

Hope this helps.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.