[vpj] Throw a clear error instead of NPE on a missing timestamp.field

[minhmo1620]

Problem Statement

When a batch push sets timestamp.field (the RMD_FIELD_PROP config) so that records carry a per-record replication-metadata (RMD) timestamp, the Spark data-writer validates the input's RMD schema in RmdPushUtils.getInputRmdSchema():

return pushJobSetting.inputDataSchema.getField(pushJobSetting.rmdField).schema();

This line had no null checks, so it threw an opaque NullPointerException in two distinct situations:

1. inputDataSchema is null — the input never populated a top-level Avro record schema. For example, VSON inputs set rmdField from RMD_FIELD_PROP in DefaultInputDataInfoProvider without ever setting inputDataSchema.
2. The configured field is absent — Schema.getField() returns null when timestamp.field is not a top-level field of the input record (nested inside the value, misnamed, or simply absent).

In both cases the user got a bare NPE at RmdPushUtils.getInputRmdSchema with no hint that the real problem is a misconfigured or missing timestamp.field. The MR record-reader path (AbstractAvroRecordReader.getField) already throws VeniceSchemaFieldNotFoundException for the missing-field case, so the same misconfiguration was legible on the MR engine but surfaced as a bare NPE on the Spark engine.

Solution

Add explicit guards in RmdPushUtils.getInputRmdSchema():

• If inputDataSchema is null → throw VeniceException naming the configured timestamp.field and noting that a top-level Avro input record schema is required (VSON inputs called out as the example).
• If the looked-up field is null → throw VeniceSchemaFieldNotFoundException naming the configured timestamp.field and stating it must be a top-level field of the input record (a sibling of the key and value fields, not nested inside the value). This mirrors the existing MR-path behavior.

The success path is unchanged.

How was this PR tested?

• Modified or extended existing tests. Added two cases to TestRmdPushUtils:
  • testGetInputRmdSchemaWhenInputDataSchemaIsNull — asserts VeniceException when inputDataSchema is null.
  • testGetInputRmdSchemaWhenRmdFieldNotInInputSchema — asserts VeniceSchemaFieldNotFoundException (not NullPointerException) when the configured field is absent from the input schema.

./gradlew :clients:venice-push-job:test --tests "com.linkedin.venice.hadoop.utils.TestRmdPushUtils"

→ 6/6 tests pass.

Does this PR introduce any user-facing or breaking changes?

• Yes (error-message improvement only). A batch push that configures timestamp.field either without a populated input schema, or with a field that is not present at the top level of the input record, previously failed with a NullPointerException. It now fails with a clear VeniceException / VeniceSchemaFieldNotFoundException whose message explains the requirement. The job still fails in the same situations — only the error is now actionable. There is no change to the success path.

[Copilot]

Pull request overview

This PR improves the Spark push-job validation path for per-record RMD timestamps by replacing an opaque NullPointerException with a targeted VeniceSchemaFieldNotFoundException when the configured timestamp.field is missing from the top-level input schema, and adds a unit test to lock in the behavior.

Changes:

• Add a null-check in RmdPushUtils.getInputRmdSchema() and throw VeniceSchemaFieldNotFoundException with an actionable message when the configured field is absent.
• Add a new test case asserting the exception type when Schema.getField() returns null.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
clients/venice-push-job/src/main/java/com/linkedin/venice/spark/utils/RmdPushUtils.java	Converts the prior NPE into a clear, user-actionable exception when timestamp.field isn’t found in the input schema.
clients/venice-push-job/src/test/java/com/linkedin/venice/hadoop/utils/TestRmdPushUtils.java	Adds coverage to ensure missing timestamp.field yields VeniceSchemaFieldNotFoundException (not NPE).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

Hi there. This pull request has been inactive for 30 days. To keep our review queue healthy, we plan to close it in 7 days unless there is new activity. If you are still working on this, please push a commit, leave a comment, or convert it to draft to signal intent. Thank you for your time and contributions.