[ https://issues.apache.org/jira/browse/HUDI-6199?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
ASF GitHub Bot updated HUDI-6199: --------------------------------- Labels: pull-request-available (was: ) > CDC payload with op field for deletes do not work > ------------------------------------------------- > > Key: HUDI-6199 > URL: https://issues.apache.org/jira/browse/HUDI-6199 > Project: Apache Hudi > Issue Type: Bug > Reporter: Ethan Guo > Assignee: Ethan Guo > Priority: Blocker > Labels: pull-request-available > Fix For: 0.13.1 > > > Delete operation in custom payload after RFC-46: while looking into a 0.13.1 > release [blocker|https://github.com/apache/hudi/pull/8573], I found that > custom payload implementation like AWS DMS payload and Debezium payload are > not properly migrated to the new APIs introduced by RFC-46, causing the > delete operation to fail. Our tests did not catch this. > > It is currently assumed that delete records are marked by > "_hoodie_is_deleted"; however, custom CDC payloads use op field to mark > deletes. > > Impact: > OverwriteWithLatest payload(also OverwriteNonDefaultsWithLatestAvroPayload) > are not affected. > for any other custom payloads: (AWSDMSAvropayload, All debezium payloads) > deletes are broken. > If someone is using "_is_hoodie_deleted" to enforce deletes, there are no > issues w/ custome payloads. > COW: > deleting a non-existant will break if not using "_is_hoodie_deleted" way. > MOR: > any deletes will break if not using "_is_hoodie_deleted" way. > Writer: > all writers(spark, flink) except spark-sql. > DefaultHoodieRecordPayload delete marker support in 0.14.0 is also affected. -- This message was sent by Atlassian Jira (v8.20.10#820010)