HIVE-29670: Avoid overwriting read-only DB_NOTIFICATION_EVENT_ID_KEY_NAME in DbNotificationListener when processing batched partition events

[Aggarwal-Raghav]

What changes were proposed in this pull request?

Check JIRA HIVE-29670 for stacktrace.

Why are the changes needed?

Alter partitions fails/warns to update stats for more than 1 batch (metastore.jdbc.max.batch.size) of partitions. local-setup-hs2.log

Does this PR introduce any user-facing change?

No

How was this patch tested?

Using q file and local hive setup testing

mvn clean test -Dtest=TestMiniLlapLocalCliDriver -Dqfile=db_notification_batch_insert.q -Drat.skip -Pitests -pl itests/qtest

[Aggarwal-Raghav]

CC @dengzhhu653 , can you please help with the feedback here! Related to HIVE-27746

On debugging, I found the same event object is being overridden in while loop in DbNotificationListener#onAlterPartitions() .

1. For 1st batch of partitions process() is triggered and set's the DB_NOTIFICATION_EVENT_ID_KEY_NAME ✅

2. But for next batch same event object is passed to process() and putParameterIfAbsent() throws IllegalStateException ❌

Before:

After:

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[dengzhhu653]

+1 looks good to me. Thanks for the PR!
The DB_NOTIFICATION_EVENT_ID_KEY_NAME looks no sense to me, even I went through the HIVE-16164.

[Aggarwal-Raghav]

Thanks for the review @dengzhhu653 .

More context, this was reported by the user, where they were performing INSERT INTO SELECT * and dynamically created more than 1000 partitions (more than default value of metastore.jdbc.max.batch.size) and the WARN log was observed on the beeline as well during stats gathering (hive.stats.autogather)