[
https://issues.apache.org/jira/browse/SPARK-59358?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Anish Mahto updated SPARK-59358:
--------------------------------
Description:
Once the version map is constructed per incoming row in the microbatch, we need
to start actually reconciling columns in the incoming microbatch that have been
deduced as unauthored - they should instead inherit the last known authored
value.
While reconciling the affected existing rows from the aux/target table, it's
also possible out of order received events in the current microbatch affect
what authored value existing ignored-null columns inherit; these need to be
updated too.
was:
During SCD2 microbatch reconciliation, it's possible an incoming microbatch
does not contain a column that the target table already contains - this can
happen if the upstream change data feed undergoes reductive schema evolution
and drops a column.
When this happens, in `Scd2ForeachBatchHandler` we union the microbatch with
the target/aux tables using `allowMissingColumns=true`. The microbatch then
desirably gains null entries for columns it does not contain but the target
table does contain (and vice versa).
With ignore-null, we need to do the microbatch projection specifically before
the version map is constructed, so that those projected null columns can obey
the active ignore-null selection during version map construction.
In isolation, this is a no-op change; we project null columns onto the
microbatch the same as before, we just do it earlier during microbatch
reconciliation.
> SCD2 Ignore-null support; Coalesce ignored-null values
> ------------------------------------------------------
>
> Key: SPARK-59358
> URL: https://issues.apache.org/jira/browse/SPARK-59358
> Project: Spark
> Issue Type: Sub-task
> Components: Declarative Pipelines
> Affects Versions: 4.4.0
> Reporter: Anish Mahto
> Priority: Major
>
> Once the version map is constructed per incoming row in the microbatch, we
> need to start actually reconciling columns in the incoming microbatch that
> have been deduced as unauthored - they should instead inherit the last known
> authored value.
> While reconciling the affected existing rows from the aux/target table, it's
> also possible out of order received events in the current microbatch affect
> what authored value existing ignored-null columns inherit; these need to be
> updated too.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
