Release date: 2026-04-30
This release is based on Postgres Pro AXE 1.0.0 and provides the following changes:
New features:
Added support for Hive partitioning when creating analytical tables
using the metastore.add_table stored procedure.
You can now specify the optional fourth parameter
to partition an analytical table.
For more information, refer to Chapter 30 and Section 25.1.1.
Added the
axe.use_postgres_snapshot
configuration parameter to avoid holding the transaction horizon.
When set to false, Postgres Pro snapshots are not used
for queries that do not work with heap tables, which allows
VACUUM to delete dead tuples, preventing
table bloat.
For more information, refer to Chapter 31.
Added support for UNION operations
between Postgres Pro heap tables and views created for
Parquet files in regular SELECT commands,
as well as between two heap tables in the
metastore.copy_table stored procedure.
Added the ability to retrieve statistics for analytical tables, including overall table statistics and column-level statistics, from the metadata catalog.
For more information, refer to Chapter 39.
Split the expire_snapshot stored procedure
into two separate procedures:
metastore.expire_snapshot_on_time, for marking
snapshots as expired based on their creation date and time,
and metastore.expire_snapshot_on_ids, for marking
snapshots as expired based on their IDs.
For more information, refer to Section 29.4.
Added the ability to configure Parquet file storage parameters, such as the compression algorithm, compression level, and the number of rows in a row group, when creating Parquet files.
For more information, refer to Section 29.3.
Added force and cascade
parameters to the metastore.remove_table stored procedure.
The force parameter specifies whether to drop
the Postgres Pro view created for the analytical table.
The cascade parameter specifies whether to drop
all views based on this view.
For more information, refer to Section 25.4.
Added the optional third parameter to the
metastore.create_view stored procedure that specifies
whether to drop an existing Postgres Pro view with the same name
before creating a new view.
For more information, refer to Section 25.3.
Added the metastore.pgpro_build() stored
procedure for retrieving the pgpro_metastore build number.
For more information, refer to Chapter 20.
Added functions for working with simple secrets and improved compatibility with S3 storages.
For more information, refer to Chapter 23 and Chapter 24.
Added the metastore.mgrant and
metastore.mrevoke stored procedures
for granting and revoking privileges on
pgpro_metastore objects.
For more information, refer to Section 22.4.
Optimized the pgpro_axe columnar-vectorized analytics engine to provide high performance on virtual machines with SSE4.2 and later vector CPU instructions. SSE2 is the minimum requirement.
For more information, refer to Chapter 13.
Added support for Red Hat Enterprise Linux (RHEL) 10.
For more information, refer to Chapter 13.
Bug fixes:
Fixed the following errors that occurred when working with S3 storages:
Errors when reading the S3 data.
Errors when using encrypted connections.
Errors when granting privileges on Parquet files.
Errors when reading the metadata of Parquet files.
Errors when using the metastore.delete_expired_files
stored procedure.
Fixed errors that occurred when using the COPY
command with pgpro_scheduler.
Known issues:
The recommended Parquet file size is from 100 MB to 10 GB. If the file size is outside this range, query performance may be significantly affected. For files smaller than 100 MB, consider keeping the data in heap tables until a sufficient volume is accumulated. For files larger than 10 GB due to Hive partitioning, consider reducing the partition size.