Airflow 2.8.3.0 - 2404 (EEP 9.2.2) Release Notes

The following notes relate specifically to the HPE Data Fabric Distribution for Apache Airflow. You may also be interested in the Apache Airflow home page.
Airflow Version 2.8.3.0
Release Date April 2024
HPE Version Interoperability See Ecosystem Pack Components and OS Support.
Source on GitHub https://github.com/mapr/airflow
GitHub Release Tag 2.8.3.0-eep-922
Maven Artifacts https://repository.mapr.com/maven/
Package Names Navigate to http://package.ezmeral.hpe.com/releases/MEP/, and select your EEP(MEP) and OS to view the list of package names.
Documentation

New in This Release

  • This release updates the Airflow component to version 2.8.3.0.
  • Introduced 2 new options:
    • admin_only_cli_access: A property to limit Airflow CLI access to only the admin cluster user. Set to true to limit Airflow CLI only for the admin cluster user. This property disables impersonation functionality.
    • admin_cli_with_impersonation: A property to limit Airflow CLI access to only the admin cluster user, except for airflow tasks commands. Supports impersonation when the property is set to true. This property has lower priority than admin_only_cli_access.
  • If the logrotation tool is installed on the cluster, Airflow copies its own configuration to the logrotation conf files. Then the webserver and scheduler log files are rotated daily by default.

Fixes

  • Fix incorrect serialization of FixedTimezone (#38139)

  • Fix excessive permission changing for log task handler (#38164)

  • Fix task instances list link (#38096)

  • Fix a bug where scheduler heartrate parameter was not used (#37992)

  • Add padding to prevent grid horizontal scroll overlapping tasks (#37942)

  • Fix hash caching in ObjectStoragePath (#37769)

Known Issues and Limitations

  • The Installer can install Airflow, but cannot set up MySQL as the backend database for Airflow. The default Airflow database is SQLite.​
  • Apache PySpark has many CVEs and is removed from the default Airflow dependencies. To use the Spark JDBC operator/hook from Apache, install PySpark as follows:
    1. Run <airflow_home>/build/env/bin/activate.
    2. Run pip install pyspark==3.3.2.
    3. Run deactivate.
    4. NOTE
      This process does not affect the Ezmeral Spark provider.
  • If the repair_pip_depends.sh script failed with the following error, you must run the script again:
    subprocess.CalledProcessError: Command 'krb5-config --libs gssapi' returned non-zero exit      status 127. 
 [end of output]

Resolved Issues

None.