[
https://issues.apache.org/jira/browse/HUDI-6219?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17825953#comment-17825953
]
Geser Dugarov commented on HUDI-6219:
-------------------------------------
Fixed in the master branch by commit ea547e5681a007e546b8ca8cb1399da0a4cd5012.
> Ensure consistency between Spark catalog schema and Hudi schema
> ---------------------------------------------------------------
>
> Key: HUDI-6219
> URL: https://issues.apache.org/jira/browse/HUDI-6219
> Project: Apache Hudi
> Issue Type: Bug
> Reporter: Wechar
> Priority: Major
> Labels: pull-request-available
>
> [HUDI-4149|https://github.com/apache/hudi/pull/5672] fix the drop table error
> if table directory moved, but it will make the Spark catalog table schema not
> consistent with Hudi schema if some column types are not Avro data types.
> *Root cause:*
> Hudi schema is Avro types, but Spark catalog table schema is not. There are
> two steps to record schema when create a hudi table:
> Step1: record the Avro compatible schema to .hoodie/hoodie.properties,
> Step2: record table in Spark catalog
> The Step2 will use HoodieCatalog.tableSchema, which is table.schema now and
> cause this issue.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
