Schema Evolution Issue in Streaming

Sandy21
New Contributor III

‎11-03-2022 03:17 AM

When there is a schema change while reading and writing to a stream, will the schema changes be automatically handled by spark

or do we need to include the option(mergeschema=True)?

Eg:

df.writeStream

 .option("mergeSchema", "true")

 .format("delta")

 .outputMode("append")

 .option("path","/data/")

 .option("checkpointLocation","/checkpoint/")

 .start()

 .awaitTermination()

  

Including the option(mergeschema=True) still throws the error :

ERROR: A schema mismatch detected when writing to the Delta table

To enable schema migration, please set:

'.option("mergeSchema", "true")'.

Do any additional option /changes needs to be done to the above query? Could you please advise to resolve this issue?

Labels:

Hubert-Dudek
Databricks MVP

‎11-03-2022 03:30 AM

mergeSchema doesn't support all operations. In some cases .option("overwriteSchema", "true") is needed. MergeSchema doesn't support:

  • Dropping a column
  • Changing an existing column's data type (in place)
  • Renaming column names that differ only by case (e.g., “Foo” and “foo”)

more on that topic here https://www.databricks.com/blog/2019/09/24/diving-into-delta-lake-schema-enforcement-evolution.html


My blog: https://databrickster.medium.com/

Sandy21
New Contributor III
In response to Hubert-Dudek

‎11-03-2022 11:13 PM

Thanks @Hubert Dudek​ . In the case of writing to a streaming table, do we need to change the checkpoint location as well in addition to adding the option("mergeSchema", "true")

if there is an addition of a new column in the incoming data?