Drill 1.8.0 Release Notes
The following release notes apply to the 1.8.0 version of the Apache Drill component included
in the MapR Converged Data Platform.
Version | 1.8.0 |
Release Date | September 12, 2016 |
MapR Version Interoperability | MapR Drill 1.8.0 is certified on the MapR v5.1.0 converged data platform. See Interoperability Matrix and Drill Support Matrix. |
Packages |
|
Noteworthy New Features in the MapR Distribution of Drill
This release provides enhanced query improvements with the following bug fixes and
improvements:
- Drill-on-YARN (mapr-drill-yarn package)
- Metadata cache pruning - See Partition Pruning Introduction and Optimizing Parquet Metadata Reading.
- IF EXISTS parameter with the DROP TABLE and DROP VIEW commands - See DROP TABLE and DROP VIEW.
- DESCRIBE SCHEMA command - See DESCRIBE.
- Multi-byte delimiter support - See List of Attributes and Definitions.
- New parameters for filter selectivity estimates - See System Options.
Additional bug fixes and enhancements listed in the Apache Drill 1.7.0 Release Notes and the Apache Drill 1.8.0 Release Notes.
NOTE Currently, you cannot use the
MapR Installer to install or upgrade to Drill 1.8. Support for this feature will be available
soon.
Resolved Issues
The following table lists resolved issues in Drill 1.8.0:
Issue | Description |
MD-1047 | Query time is improved for simple queries on INFORMATION_SCHEMA.TABLES when there are many views. |
MD-1044 | Queries no longer get stuck in the CANCELLATION_REQUESTED state. |
MD-1042 | CTAS queries that contain an INNER JOIN and UNION no longer produce an inefficient query plan and resource errors. |
MD-1030 | Partition pruning now works correctly when optimization is applied to queries that contain an IN list with more multiple lines. |
MD-949 | Flatten on CONVERT_FROM no longer fails with a ClassCastException. |
MD-904 | Drill no longer throws an array index out of bound exception when reading a parquet file written by a mapreduce program. |
MD-901 | Drill no longer produces a Java compilation error for large queries with many CASE and CAST expressions. |
MD-895 | Drill queries no longer produce memory leaks. |
MD-888 | Queries no longer fail with a "Connection timed out" error. |
MD-878 | The RANK() window function no longer returns the wrong results for large datasets. |
MD-746 | Queries no longer hang when fragment 0 is running. |
MD-727 | Drill returns null values for non-existent columns in HPE Ezmeral Data Fabric Database JSON tables. |
MD-603 | Drill now returns the directory structure associated with a workspace. |
MD-509 | Drill no longer returns a NumberFormatException error when casting an empty string to int in HBase/HPE Ezmeral Data Fabric Database. |
Known Issues
Issue | Description |
---|---|
MD-1076 | Drill drops NULL fields when writing to a JSON file. |
MD-1073 | Queries on ORC tables no longer take an unusually long time when Drill runs for multiple days without restart. |
MD-1061 | Queries on Hive ORC tables fail with a "does not have access to maprfs:///" error. |
MD-892 | The drill-env.sh script does not set the file client read ahead throttle environment variable. |
MD-833 | Queries submitted from Tableau produce IndexOutOfBoundsException errors. |