Release notes

Added support for SSL with a new Manage SSL cluster action

Added HTTPS auth to Patroni REST API

Added SSL for client-server PostgreSQL connections

Added SSL to PgBouncer

Added SSL to HAProxy

Added SSL to etcd

Added SSL to pgBackRest

Added HTTPS support for Enterprise Tools

Upgraded timescaledb to 2.15.0

Balancer was removed from the Import cluster page

Added precheck that certificates exist on all nodes

Added precheck that certificates are placed outside the ADPG data directory (PGDATA) and are specified with absolute paths

Added check for pgbouncer-exporter that the ADPG service is a prerequisite on the same host

Upgraded Ansible to 2.9

Added check and UI restriction that the replicatefrom Patroni tag value is chosen from the list of Patroni members

Fixed when some config strings with special characters could be pre-processed by Ansible for safety reasons

Fixed when trust auth for a local connection on ADPG CE was ignored

Fixed when Prometheus/Grafana failed while trying to remap to a new host in ADPG CE

Fixed when pgbackrest-exporer was not stopped after the Monitoring service’s Stop and cluster’s Stop actions

Fixed when the Etcd installation failed with uppercase hostnames

Fixed when Grafana remapping to a new host failed

The Metrics storage service was migrated from the Postgres storage to the Prometheus engine

Added basic auth to Patroni REST API

Added the pg_orphaned extension

Added the pg_repack extension

Added the timescaledb_toolkit extension

Added the pg_vector extension

Added the check that the input value of the replicatefrom Patroni tag does not refer to the host itself for which the tag is specified

Added a link to the ADPG documentation to the cluster Info section

Added the check that the required parameter port is specified in the postgresql.conf field

The global_adcm_precheck_fail check was refactored to display a list of all occurred problems

The SYS_UID*/SYSGID* variables were pre-assigned to ensure that UIDs/IDs of accounts created for ADPG will not match the reserved ones in the range 1-499

The Patroni instance is now restarted on SIGKILL

ADPG cluster member names have been simplified to Patroni member names

Fixed when listen_addresses was incorrectly parsed in case of using several IPs

Fixed when the Check cluster action failed because it expected IPs in HAProxy file and not FQDNs

Fixed when the Abort upgrade action failed after successful Upgrade leader during the major upgrade

Fixed when the Complete upgrade action failed if the ADPG service was already stopped before the major upgrade

Fixed when the Patroni restart was used if reload was enough during the Reconfigure & Restart action

Fixed when Balancer packages were not upgraded during the major upgrade EE

Fixed when the Save button was locked without reason on the cluster’s Configuration page

Fixed when the PG_HBA field was replaced with its default value after the Reconfigure & Restart action

Fixed when the Reinstall action failed with Patroni tags

Fixed when the Info tab of the ADPG service was not updated with new ADPG member names after the major upgrade

Fixed when the logrotate.d folder contained the adpg14 data instead of adpg16 after the major upgrade to ADPG 16

Added support for RED OS 7.3 Certified edition

Added Patroni member names to the ADPG service Info section

Added the allow_to_terminate property to ADPG major upgrade jobs

ADCM min version was limited to 2.3.0 to ensure introduced ADPG features that use ADCM structs, ADCM configuration groups, job termination

Introduced Patroni tags (see Configuration parameters → ADPG → Patroni ADPG tags) and ability to pinpoint them to different ADPG nodes

Changed the Patroni default log path to /var/log/adpg16-patroni

Changed the display name of the use_custom_location parameter to use_custom_patroni_log_dir

HAProxy was rebuilt to support SSL on Ubuntu, Alt Linux, Astra Linux, RED OS

Simplified the ADPG Configuration section to a plaintext sheet

Changed timeout to wait replica infinitely during major upgrade and the Expand action

Added check that all services are running and healthy on the Precheck action of the major upgrade

Refactored etcd3 client to retry connecting to other nodes with etcd servers in case of unresponsive service

Implemented Patroni config check: ttl >= loop_wait + 2 * retry_timeout

Added the Run backup after upgrade checkbox to boost EE major upgrade

The Reconfigure & Reload action is now included into the Reconfigure & Restart action, which is able to apply changes without the ADPG service restart if possible

Fixed when stanza upgrade was not applied for all replicas during the Complete upgrade action

Fixed when PgBouncer was not updated during minor upgrade on Centos and Alt Linux

Fixed when the Complete upgrade action failed for cluster with external monitoring during major upgrade to ADPG v16.x.x

Fixed when the Reconfigure & Reload action rendered incorrect configs and broke replication

Fixed when the repetitive failed Upgrade replica step was retried with all replicas in a cluster during major upgrade to ADPG v16.x.x

Fixed when the Upgrade leader step failed in case the adpg16 stanza existed

Fixed when stopping postgres failed on timeout in the Upgrade leader step during major upgrade to ADPG v16.x.x

Fixed when the Abort upgrade step did not check that at least one adpg14 replica was still alive and existed during major upgrade to ADPG v16.x.x

Upgraded the PostgreSQL version to 16.3

Added support for Ubuntu 22.04

Upgraded Patroni to 3.2.1

Upgraded etcd to 3.5.11

Upgraded HAProxy to 2.9.0

Upgraded postgis to 3.4.1

Upgraded Grafana to 9.5.8

Upgraded PgBouncer to 1.23.1

Added the pg_walinspect extension

Added dynamic URLs to ADPG services

Removed the Run SQL action of the ADPG service

Upgraded pgwatch2 to 1.13.0

Upgraded timescaledb to 2.14.2

Upgraded confd to 0.16.1

Added UUID to the ADPG cluster

Refactored ADPG to use core part as a submodule in other Arenadata products

Changed the Enable maintenance mode and Disable maintenance mode action names to Pause Cluster and Resume Cluster respectively

Improved EE upgrade with Patroni Pause/Resume

Added locale=en_US.UTF-8 to initdb to validate logs in English

Updated the default value of the wal_keep_size PostgreSQL parameter to 128 MB

Set log level to info in confd haproxy template

Removed the following extensions: hstore_plperl, hstore_plperlu, jsonb_plperl, and jsonb_plperlu

Fixed when Grafana lost the connection to a database if the ADPG service port had been changed

Fixed when stanza creation failed if the ADPG service non-default port was used

Fixed when adpg-monitoring-agent could not start in case of a numeric ADPG cluster name

Fixed when the Reinstall statuschecker cluster action did not update .sh scripts

Fixed when the Reconfigure & Restart action of the Balancer service failed after changing the Grafana port

Fixed when Grafana did not display metrics on dashboards with database filter for an external cluster

Fixed when upgrade failed if ADPG has an extension with dash (e.g. uuid-ossp)

Added the ability to create backups using the functionality of the pgBackRest utility. The Backup: Start, Backup: Restore cluster, and Backup: Info actions are implemented for the ADPG service. ADCM UI now contains the following settings of the ADPG service: Archive timeout(s), Backup stanza, Backup repo, Repo type, S3 URI style, S3 Region, S3 Bucket, S3 Endpoint, S3 Key, S3 Key Secret, Retention full type, Retention full, Retention diff, Enable compression, Compress type, Compress level, Log path, Log level, Use custom config, Global options, and Custom options

Astra Linux 1.7 SE "Orel" is supported in ADPG Enterprise edition

A connection pooler is implemented based on PgBouncer in ADPG Enterprise edition. ADCM UI now contains the PgBouncer settings for the ADPG and Balancer services. For ADPG: Enable PgBouncer, listen_port, pool_mode, max_client_conn, default_pool_size, min_pool_size, reserve_pool_size, reserve_pool_timeout, max_db_connections, max_user_connections, Enable all users, Users list, Enable all databases, Databases list, Use custom pg_hba.conf и Custom pg_hba.conf. For Balancer: pgbouncer_leader_port, pgbouncer_replica_port.

The Force reconfigure and restart checkbox is added to the Reconfigure & Restart action settings. Force reconfigure and restart disables validation of the configuration settings before restarting ADPG

The new monitoring system is implemented. Added the Metrics storage service that can collect metrics from multiple ADPG clusters. For information on the installation of the new monitoring system, refer to Install monitoring

ADCM UI now contains the following monitoring cluster settings: Use external ADPG cluster for monitoring, External ADPG cluster [ip_address:port], Monitoring db name, Monitoring username, and Monitoring user password

The following settings of the new Metrics storage service are available in ADCM UI: Monitoring db name, Monitoring username, Monitoring user’s password, Grafana TCP port, and Grafana admin’s password

Load balancing is implemented in ADPG Enterprise edition. The new Balancer service performs balancing based on HAProxy

ADCM UI now contains Balancer service settings: leader as replica, leader_port, replica_port, and balancer_stats_port

The contrib and PostGIS extensions are included in ADPG and do not require installation

Since the contrib and PostGIS extensions are included in the ADPG distribution, the settings and actions related to their installation are removed

You can upgrade to version 14.3.3 from version 14.3.2. If you need to upgrade from an earlier version, upgrade first to 14.3.2 and then to 14.3.3

High availability (HA) feature based on Patroni is implemented in ADPG Enterprise Edition

The Etcd service required for HA is added

The ADCM UI now contains External Etcd settings that allow you to use an external etcd cluster as an alternative of the built-in Etcd service

The Check, Disable maintenance mode, Enable maintenance mode, and Expand actions are implemented for services

The ADCM UI now includes the following Patroni ADPG configuration settings: synchronous_mode, synchronous_node_count, synchronous_mode_strict, and maximum_lag_on_failover

The Enable monitoring setting and the Install monitoring agents action are added for the ADPG cluster. These options install monitoring agents on all hosts of the cluster

Offline installation is implemented in ADPG Enterprise Edition

Upgrade an ADPG cluster from a previous version to version 14.3.2 does not preserve the settings specified in postgresql.conf custom section. It is necessary to restore them manually

ADPG migrates to PostgreSQL 14.3. For information about changes and migration to PostgreSQL 14.3, see PostgreSQL release notes

Alt Linux 8.4 SP is supported in ADPG Enterprise edition

Base monitoring system is implemented

The following performance tuning options are added to the ADCM UI: effective_cache_size, maintenance_work_mem, work_mem, min_wal_size, max_wal_size, wal_keep_size, huge_pages, and superuser_reserved_connections

The ADCM integration for the ADPG installation and administration is implemented

The Install, Reinstall, Delete, and Uninstall actions are implemented for the Chrony service in ADCM

The ability to perform minor updates of the ADPG bundle version via ADCM is implemented

The ability to add lines to the pg_hba.conf file via ADCM is added. To do this, it is necessary to fill in the PG_HBA custom section in the ADPG service configuration window and then apply the Reconfigure & Reload action

The Run SQL action is implemented for the ADPG service in ADCM

The Install postgis action is implemented for the ADPG service in ADCM

The Install contrib action is implemented for the ADPG service in ADCM

The Start and Stop actions are implemented for the ADPG service and cluster in ADCM

The Check action is implemented for the ADPG cluster in ADCM

The Precheck action is implemented for the ADPG cluster in ADCM

The Install and Reinstall actions are implemented for the ADPG service in ADCM