Greenplum Database 4.2.5 Release Notes

posted Oct 26, 2013, 6:35 AM by Sachchida Ojha
Greenplum Database 4.2.5 is a service pack release that enhances support database backup and the query execution management.

Supported Platforms
Greenplum Database 4.2.5 runs on the following platforms:
• Red Hat Enterprise Linux 64-bit 5.5, 5.6, 5.7, 6.1, and 6.2
• SUSE Linux Enterprise Server 64-bit 10 SP4, 11 SP1
• Solaris x86 64-bit v10 U7, U8, U9, U10
• Oracle Unbreakable Linux 64-bit 5.5
• CentOS 64-bit 5.5, 5.6, 5.7, 6.1, and 6.2
Greenplum Database 4.2.5 supports Data Domain Boost on Red Hat Enterprise Linux.
Greenplum Database 4.2.5 supports Data Domain Boost SDK version with DDOS, 5.1 and 5.2.

GPDB was certified against the following DCA versions:
Note: For the most recent server platform support information for Greenplum Database, check the EMC Support Matrix (ESM).
To access the ESM, go to the Support Zone home page and click the link E-Lab Interoperability Navigator.
In the E-Lab Interoperability Navigator, search for Greenplum Database and add one or more search results to the search cart. Click Get Results to display links to 
EMC Support Statements.

Changed and Deprecated Features
These Greenplum Database features are deprecated with Greenplum Database 4.2.5 and will be removed in a future release:
• gp_dump utility
• gp_restore utility
• The -y option of the gpcrondump utility
For Greenplum Database 4.2.5, if the gpcrondump utility -y option is specified, the utility returns a warning message stating that the -y option is deprecated and 
creates a copy of the job log file in the specified directory. See the Greenplum Database Utilities Guide for information about the utilities.

Product Enhancements 
Greenplum Database 4.2.5 includes enhancements in these areas:

Incremental Backup
The Greenplum Database utilities gpcrondump and gpdbrestore support incremental backup of append-only tables, including partitioned tables. An incremental backup is similar to a full backup. An incremental backup creates backup files for the master and hosts. Similar to a full backup, an incremental backup backs up regular, heap-storage tables. The difference between an incremental backup and a full backup is that append-only tables are backed up only if one of the following operations was performed on the table after the last backup:
DROP and then re-create the table
For a partitioned, append-only tables, only the changed table partitions are backed up. If an incremental backup is specified when restoring data, gpdbrestore ensures that the correct incremental backups and full backup are used to restore the data. See the Greenplum Database System Administrators Guide for information about 
incremental backups. See the Greenplum Database Utilities Guide for information about gpcrondump and gpdbrestore.

Managed File Replication with Data Domain Boost 
When Data Domain Boost is used to back up a database to a Data Domain system, you can configure a local Data Domain system for backing up the database and a remote system for disaster recovery of backup images. The Greenplum Database utility gpmfr uses Data Domain Boost managed file replication to manage the Greenplum Database backup sets that are on the primary and a remote Data Domain system. The Greenplum Database utility gpcrondump supports configuring Data Domain systems to be used for managed file replication. See the Greenplum Database Utilities Guide for information about gpmfr and gpcrondump. See the Greenplum Database System Administrators Guide for information about configuring Data Domain systems for managed file replication.

Restoring Backups from a User Specified Directory
The Greenplum Database utility gpdbrestore supports restoring a backup from a user specified directory. The -u option specifies the directory. You use the 
gpdbrestore option to restore a backup after you have used the gpcrondump -u option to specify a directory for backup files.  See the Greenplum Database Utilities Guide for information about gpcrondump.

Query Workfile Size Management
The new Greenplum Database server configuration parameters gp_workfile_limit_per_query and gp_workfile_limit_per_segment control the maximum size of the temporary workfiles created on disk during query execution. You can create gp_workfile_* views with Greenplum Database 4.2.5 that contain information on all the queries that are currently using disk spill space for workfiles.  This information can be used for troubleshooting, tuning queries, and setting the workfile configuration parameters. You can create and use the following views: 
• The gp_workfile_entries view contains one row for each operator using disk space for workfiles on a segment at the current time.
• The gp_workfile_usage_per_query view contains one row for each query using disk space for workfiles on a segment at the current time. 
• The gp_workfile_usage_per_segment view contains one row for each segment. Each row displays the total amount of disk space used for workfiles on the segment at the current time.

See the Greenplum Database Reference Guide for information about the configuration parameters. See the Greenplum Database System Administrators Guide for 
information about creating the gp_workfile_* views.

Query Execution Plan Management
The new server configuration parameter gp_max_plan_size controls the maximum size of query plans that are generated during query execution. See the Greenplum Database Reference Guide for information about the configuration parameter. 

Support for Hadoop 2.0 distribution Pivotal HD 1.0
Greenplum Database 4.2.5 adds gNet support for Pivotal HD 1.0. Greenplum Database provides an enhancement for the server configuration parameter 
gp_hadoop_target_version. The parameter specifies the preferred Hadoop distribution. 

To enable support for Pivotal HD 1.0, set gp_hadoop_target_version to gphd-2.0.

Parameter and Option Updates
These parameters and options have been updated:
• For the Greenplum Database utility gpfdist, the maximum timeout value for the -t option is increased to 600 seconds.
• For Greenplum Database configuration parameter gp_resqueue_priority_cpucores_per_segment, the maximum value is increased to 512.0.
See the Greenplum Database Reference Guide for information about the configuration parameters. 

Catalog Table Changes
In the pg_stat_last_operation catalog table, the TRUNCATE statistics were combined with the VACUUM statistics. The TRUNCATE statistics are now tracked 
separately from the VACUUM statistics.