Behavioral changes for Cloudera Runtime 7.1.9 SP2 CHF 3 introduce functional adjustments
to Cloudera Runtime components.
Impala
- CDPD-111492: Adding a catch-up mode to EventProcessor
- Previous behavior:
- Previously, when Hive Metastore (HMS) event processing experienced a
large lag or during a catalogd High Availability (HA) failover, the EventProcessor component
performed full table reloads for table-level events. This could cause slow recovery times when
processing large event backlogs.
- New behavior:
- Added a configurable catch-up mode by using the
hms_event_catchup_threshold_s property for HMS event processing. This
mode invalidates tables instead of performing expensive reloads when the event lag exceeds a
configurable threshold, which improves recovery from large event backlogs.
Apache Jira: IMPALA-14230
