Fixed issues in Cloudera Runtime 7.1.9 SP1 CHF 16

Know more about the list of fixes that are shipped for CDP Private Cloud Base version 7.1.9 SP1 CHF 16.

CDPD-72320: Insufficient EC pipelines in the container pipeline cache: Previously, the Ozone Manager pipeline cache did not store all DataNode information for a container if, for example, some DataNodes did not report in time. This caused issues for Erasure Coded (EC) files. If the number of nodes was less than the required number of EC data blocks, the following error occured during data reads:
There are insufficient datanodes to read the EC block
This error persisted until the cache expired after 6 hours, even if all relevant DataNodes became available. The issue is now fixed. EC pipelines with incomplete DataNode information are not cached.
CDPD-95213: Hive TLS hostname verification in version 7.1.9 SP1 CHF16: Previously, the Socket Appender in Apache Log4j Core did not perform TLS hostname verification, even when the property was enabled.; This issue is resolved for the Hive service by upgrading the Log4j Core version to 2.25.3, which ensures that hostname verification is correctly performed to secure log traffic. See, CVE-2025-68161.
note
This fix applies only to the Hive service and default configurations generated by Cloudera Manager in the 7.1.9 SP1 CHF16 release. If you implemented custom loggers by using the hive_log4j2_xml_role_safety_valve safety valve or similar configurations, you must manually update your custom properties to ensure all loggers and appenders are correctly defined.

Go to Cloudera Manager

Select the Hive Service > Configuration tab and search forlog4j_safety_valve

Add the missing logger or appender entry to the loggers or appenders list to fix the mismatch.

Alternatively, remove the mismatched definition entirely.

Save the changes and restart the Hive service.
IMPALA-14271: Fragments wait for runtime filters during tuple caching: Previously, some fragments became stuck waiting for remote runtime filters, leading to performance regressions and timeouts.; This issue is resolved by re-enabling the logic to cancel backends as soon as a query transitions to the results-returned state.
Apache Jira: IMPALA-14271
IMPALA-13479: Maximum thread cache size limit for TCMalloc is increased: Previously, the TCMalloc max_total_thread_cache_bytes property was limited to 1 GB.; This issue is resolved by updating the toolchain to a version that removes the 1 GB limit. You can now set the thread cache size to a value higher than 1 GB to improve performance on high-concurrency systems.
Apache Jira: IMPALA-13479
IMPALA-13487: Missing memory allocation metrics in execution profiles: Previously, execution profiles did not capture detailed metrics for memory allocation and wait times.; This issue is resolved by adding profile counters to track memory allocation duration and bytes across various components, such as Parquet scanners and final row batches of the fragment instances. These new metrics, including page faults and scratch batch allocation durations, allow you to detect resource contention and better understand query execution delays.
Apache Jira: IMPALA-13487
IMPALA-12146: Fix stale memory reserved values after abnormal executor termination: Previously, if an executor backend terminated abnormally, the reserved memory accounting was not updated correctly. This caused the admission controller to maintain an incorrect view of the reserved memory for that specific backend.; This issue is resolved by ensuring that the coordinator accurately reflects the current memory state after an executor quits unexpectedly.
Apache Jira: IMPALA-12146

Fixed Common Vulnerabilities and Exposures in Cloudera Runtime 7.1.9 SP1 CHF 16

There are no CVEs fixed in this release.