Fixed Issues in Hive

Cloudera Runtime 7.1.9 SP2 resolves identified Hive functional errors and includes technical patches to improve service stability and performance.

CDPD-98415: Hive ACID incremental replication no longer fails with "MetaException: invalid write id"

Previously, during Hive ACID incremental replication involving large volumes of events (such as bulk DDLs, partition updates, or TRUNCATE TABLE operations), the replication load task failed with the following error:

Caused by: org.apache.hadoop.hive.metastore.api.MetaException: invalid write id <ID> for table <db_name>.<table_name>

This occurred because the background task responsible for cleaning up dangling transactions (CLEAR_DANGLING_TXNS) executed prematurely at the end of individual task batch iterations instead of waiting for the full incremental load operation to complete. As a result, the cleanup process aborted in-flight target transactions before replaying their corresponding commit events in subsequent batches, which invalidated the target write ID.

Cloudera Manager and Cloudera Runtime now defer the CLEAR_DANGLING_TXNS task to execute only after all task iterations in the incremental load operation complete successfully. This fix prevents premature transaction aborts and ensures seamless incremental replication.

Apache Jira: HIVE-29459

CDPD-92478: Timestamp processing in MetaStoreUtils

Previously, Hive Metastore utilities used local time zone settings to convert between timestamps and strings.

This issue is now fixed by using UTC time zone and the java.time.Instant class to process timestamps, which ensures that time points are represented accurately regardless of local time zone rules.

Apache Jira: HIVE-28337

CDPD-100039: Hive Query Language compilation error

Previously, Hive Query Language failed with a compilation error because the org.apache.commons.configuration package was not found.

This issue is resolved by upgrading the Atlas-related code to synchronize with recent Atlas changes.

CDPD-98866: Timestamp processing in MetaStoreUtils

Previously, Hive Metastore utilities used local time zone settings to convert between timestamps and strings.

This issue is now fixed by using UTC time zone and the java.time.Instant class to process timestamps, which ensures that time points are represented accurately regardless of local time zone rules.

Apache Jira: HIVE-28337

CDPD-98287/CDPD-97222: HiveServer2 hangs during JVM shutdown

Previously, the PrivilegeSynchronizer thread did not run as a daemon thread. This caused the HiveServer2 process to remain running but stop listening to the port during OutOfMemory (OOM) situations, preventing the JVM from destroying properly.

This issue is resolved by setting the PrivilegeSynchronizer thread as a daemon thread, ensuring the process exits correctly during failure events.

Apache Jira: HIVE-20676

CDPD-93597: SHOW COMPACTIONS output filtering

Previously, the SHOW COMPACTIONS output included all historical information, which caused the display to become unwieldy when many partitions or history lines existed.

This issue is resolved by adding the ability to filter the command output by database, table, partition, and compaction type or state.

Apache Jira: HIVE-13353