E.1. Postgres Pro Enterprise 15.10.1
Release Date: 2024-12-09
E.1.1. Overview
This release is based on PostgreSQL 15.10 and Postgres Pro Enterprise 15.8.2. All changes inherited from PostgreSQL 15.10 are listed in PostgreSQL 15.10 Release Notes. As compared with Postgres Pro Enterprise 15.8.2, this version also provides the following changes:
Added the ability to form the
MergeAppend
operation instead ofAppend
, which allows omitting additional sorting needed to ensure that the data is in fact sorted. NowMergeAppend
is formed specifying the columns to sort by. The new ability applies only when the enable_appendorpath option is set toON
.Improved performance of CFS by reducing the number of invalidation messages for tables, which previously caused multiple file reopen events. Now invalidation is targeted at the specific segments affected by the garbage collector.
In JSON files that store frozen plans, arrays of parameter types are now displayed in the text format instead of numeric, making it easier to edit an incorrect parameter type.
Allowed transformation of correlated
IN
subqueries into joins.Implemented the ability to dump the state of a single backend process by sending the dump signal.
Added a validation step to check for non-zero values in unused header fields of
*.cfm
files by pg_upgrade.Changed the error level to calculate the statistics on vacuum interrupts to
ERROR
(PGERROR
) only. Calculating vacuum interrupts for thePANIC
error level could cause unpredictable system behavior, and the statistics could not be calculated anyway because of the server crash.Fixed
cfs_gc_relation
to return the actual number of processed segments of the relation.Fixed an issue with upgrading from PostgreSQL or Postgres Pro Standard to Postgres Pro Enterprise using pg_upgrade when a multitransaction offset counter overflow occurred.
Fixed the processing of 64-bit transaction IDs with proper encoding and decoding of the transaction ID.
Fixed an issue with upgrading from PostgreSQL or Postgres Pro Standard to Postgres Pro Enterprise using pg_upgrade that could occur when attempting to convert heap pages to the
double xmax
page format in cases where there was insufficient free space to accommodate the special area. In the presence of multitransactions, this could lead to errors like “MultiXactId XXXXX has not been created yet” or “could not access status of transaction XXXXXX”, depending on the server version. No data loss is expected as the actual conversion did not occur. Pages will now be safely converted upon first access.Fixed a memory leak that could occur when the
EXPLAIN
command's output produced over 64 columns.Fixed a segmentation fault, which could occur while executing the
COPY TO
command when using the online_analyze module.Fixed an issue where
EXPLAIN ANALYZE
showed incorrect counts of inserted and conflicted tuples during row inserts with conflicting primary keys when usingON CONFLICT DO NOTHING
.Fixed an issue that could slow down query execution. The reason was that the optimizer selected a suboptimal index due to the lack of its cost estimation after removing redundant clauses with the enable_appendorpath option set to
ON
.Fixed an issue in the parsing of prepared statement names when the statement name contained an SQL command name, and removed the no-longer-needed
sr_plan.max_consts_len
.Added support for ARM architecture for Red OS Murom 8.
Removed support for legacy data loading, previously required for reading obsolete sr_plan binary plans and converting them to the JSON format. A minimal functionality is retained to detect legacy files, back them up, and log a message.
Upgraded aqo to include the following optimizations and bug fixes:
Enhanced aqo wide search (
aqo.wide_search
=on
) to utilize data from multiple neighbors simultaneously. Also, revised its interaction withaqo.min_neighbors_for_predicting
to allow queries with fewer than the specified number of data samples to participate in wide search. Previously, such queries were excluded from consideration.Fixed an array index out-of-bounds issue in the aqo smart statement timeout update logic, as well as incorrect statistics insertion into
aqo_query_stats
when the arrays were full. In previous releases, usingaqo_query_stat_update
with arrays of size 20 or more could lead to subsequent statistics updates being written to uncontrolled memory segments. Make sure to install the Postgres Pro on both primary and standby servers to avoid similar memory corruption problems.
Upgraded mamonsu to version 3.5.9, which provides optimizations and bug fixes. Notable changes are as follows:
Added support for systems where setuptools version greater than 67.7.2 is installed.
Ended the use of dotted
user:group
specification at the RPM pre-install stage.
Upgraded multimaster to version 1.2.0. Significantly increased catchup performance by enabling asynchronous commit operations on a lagging node syncing with the donor node, controlled by the
multimaster.enable_async_3pc_on_catchup
parameter.Upgraded the ODBC driver to version 17.00.0002.
Upgraded oracle_fdw to version 2.7.0.
Upgraded orafce to version 4.13.5.
Upgraded pg_filedump to version 17.1, which specifically fixed an issue that could cause pg_filedump to crash with a segmentation fault when handling incomplete pages.
Upgraded pg_hint_plan to version 1.5.2.
Upgraded pg_integrity_check to display actual checksum values, if the checksums differ, only with the
--verbose
option.Upgraded pg_portal_modify to version 0.3.4.
Upgraded pg_proaudit to provide the following optimizations and bug fixes:
Added logging of security events related to operations with profiles:
CREATE PROFILE
,ALTER PROFILE
, andDROP PROFILE
.Changed the distribution of security events related to functions and stored procedures among the groups of security events: the
ALL_DDL
group now includes no events related to functions and stored procedures,ALL_PROC
includesCREATE
,ALTER
, andDROP
, whileALL_DML
includesEXECUTE
.Fixed compatibility with the connection pooler. Previously, issues could be encountered when logging
AUTHENTICATE
andDISCONNECT
security events.Fixed logging of security events on partitioned tables by pg_proaudit. Now the
SELECT/UPDATE/INSERT/DELETE/TRUNCATE
events on a partitioned table get logged if logging rules imply logging of appropriate events for this table. Previously, only events on individual partitions were logged.
Upgraded pg_probackup to version 2.8.5 Enterprise, which provides the following optimizations and bug fixes:
Enabled the output of the
version
command in the JSON format using--format
=json
.Added the
--all
option to theamcheck
command, which allows performing all the checks in a single command.Added support for the
include
parameter in the configuration file. The contents of the file specified ininclude
gets added to the configuration file instead of this parameter.Added a possibility to specify an individual number of threads for execution of a backup and its subsequent validation by specifying the
--backup-threads
and--validate-threads
option, respectively.Changed the priority of setting
PGDATA
andBACKUP_PATH
. Now the values from the command-line options take precedence.Added validation of WAL files for backups to be merged by the
merge
command.An error message that was issued when the
validate
command run with the--wal
option found an archive in theDEGRADED
status has been replaced with a warning since backups in this status do not affect the restore correctness as a whole.Added support for the
--s3-config-file
option ofrestore_command
.postgresql.auto.conf
must include therestore_command
with the--s3-config-file
option if the backup was created with this option.The default value of
PG_PROBACKUP_S3_IGNORE_CERT_VER
is set toon
according to the documentation.Fixed an issue that could occur if the command line was too long.
Fixed an issue that could occur during validation or restore when the PITR from a previous timeline was performed.
Fixed an issue that could occur when the S3 connection port was specified both in
PG_PROBACKUP_S3_PORT
andPG_PROBACKUP_S3_HOST
. Now the value specified inPG_PROBACKUP_S3_HOST
, together with the address, has higher priority.
Upgraded pgpro_controldata to version 17.1.0.
Upgraded pgpro_stats to version 1.8, which supports Postgres Pro 17. Notable changes are as follows:
Updated
pgpro_stats_statements
andpgpro_stats_totals
views to include fields added to pg_stat_statements. Related functions were updated accordingly.Streamlined access to views and functions. Specifically, access to the
pgpro_stats_archiver
,pgpro_stats_vacuum_database
,pgpro_stats_vacuum_tables
, andpgpro_stats_vacuum_indexes
views was granted to all users. Previously, these views required explicitly granting access rights. Access to execution of thepgpro_stats_trace_reset
function, which could previously be executed by any user, was restricted to superusers.Implemented an optimization that lowers the locking time in some cases when copying metrics from the local memory to the shared memory.
Eliminated an excessive check of holding the lock when accessing the hash table of session-tracing filters.
Fixed processing of the
pgpro_stats.stats_temp_directory
configuration parameter. Previously after the server restart for changes to this parameter value to take effect, the fatal error “pfree called with invalid pointer” could occur when processing this parameter.Fixed an issue that could occur during session tracing and cause errors “could not write file ""pg_stat/pgpro_stats_filters.trace.tmp"": No such file or directory”. Concurrent writing from different sessions to a trace file is fixed now, so processes can write to the trace file in parallel without collisions.
Upgraded pgpro_pwr to version 4.7, which provides new features, optimizations and bug fixes. Notable changes are as follows:
A subsample feature to collect relatively fast changing data.
New report tables, specifically regarding session states.
Support for new Postgres Pro 17 statistics.
A possibility not to reset statistics of the statistics collecting extension during taking a sample.
The change of the type for the field that tracks transaction IDs in a certain table from
xid
totext
. The use of thexid
type could previously cause a failure of pg_upgrade when upgrading from Postgres Pro Standard to Postgres Pro Enterprise.
Upgraded pgpro_scheduler to version 2.10, which provides the following changes:
Fixed an issue with automatic restart of one-time jobs after job interruption.
Fixed
schedule.get_active_jobs
to properly return the list of currently executed cron jobs.Fixed an issue where errors occurring during job execution were not properly logged due to a primary key constraint preventing multiple log entries.
Fixed an issue where
at job executor
would fail when thedatabase manager
prematurely released the shared memory segment before the worker had connected to it.Fixed an issue where
schedule.timetable
did not display the next execution time of cron jobs with eithernext_time_statement
or an array of dates. Fixed an issue with calculating next execution time for cron jobs that caused job instances to be skipped.Fixed an issue causing database crashes due to segmentation faults in the job executor background worker.
Upgraded pg_repack to version 1.5.1.
Upgraded pg_wait_sampling to provide tracking of subquery IDs and utility statements and to add the new
pg_wait_sampling.sample_cpu
parameter, which allows specifying the sampling mode that determines whether to perform sampling of on-CPU backends.Upgraded pljava to version 1.6.8.
Upgraded the PLV8 extension to version 3.2.3.
Ensured compatibility of
postgres_fdw
with autonomous transactions.Upgraded PTRACK to prevent possible issues with PTRACK backups by setting the
ptrack.map
file for automatic deletion when PTRACK is disabled.Upgraded tds_fdw to version 2.0.4.
E.1.2. Migration to Version 15.10.1
If you are upgrading from a Postgres Pro Enterprise release based on the same PostgreSQL major version, it is enough to install the new version into your current installation directory.
ABI versions may change between minor releases of Postgres Pro. If this is the case, and you see the ABI mismatch
error when trying to run your extension, make sure to install a new version of the extension supplied with a new release of Postgres Pro, or recompile your third-party extension to be able to use it with the current version of Postgres Pro.
If you take backups using pg_probackup and you have previously upgraded it to version 2.8.0 Enterprise or 2.8.1 Enterprise, make sure to upgrade it to version 2.8.2 Enterprise or higher and retake a full backup after upgrade, since backups taken using those versions might be corrupted. If you suspect that your backups taken with versions 2.8.0 or 2.8.1 may be corrupted, you can validate them using version 2.8.2.
Starting from Postgres Pro Enterprise 15.5.1, the privileges of CREATEROLE
are restricted, and modification of other roles requires the role requesting the change to have the ADMIN OPTION
permission.
It is strongly recommended to avoid migrating a cluster with 32-bit transaction IDs to Postgres Pro Enterprise versions below 15.4.2, if there are multitransactions in your database, as it may cause data corruption.
When upgrading to Postgres Pro Enterprise versions starting with 15.3.1, make sure to upgrade pg_probackup Enterprise to version 2.6.4 or higher since lower versions are incompatible with these Postgres Pro Enterprise versions.
To migrate from PostgreSQL, as well as Postgres Pro Standard or Postgres Pro Enterprise based on a previous PostgreSQL major version, see the migration instructions for version 15.