Hybrid Manager 1.3.1 release notes v1.3.2

Release Date: October 28, 2025

This release includes bug fixes and security updates.

HM console

• The total number of Postgres clusters on the HM console home page now correctly excludes individual cluster nodes (e.g., EFM or CNP replicas) to prevent inflated and inaccurate totals.
• Fixed an issue where the database connection information in the HM console's Connect tab incorrectly displayed 'edb_admin' instead of the required 'bdrdb' for Advanced High Availability and Distributed High Availability (PGD) clusters.
• The database size in the Monitoring tab of a cluster's detailed page now correctly displays storage metrics for self-managed clusters.
• Tags are now correctly displayed for active clusters on the Deployment Details page of the HM console.

Advanced High Availability and Distributed High Availability (PGD)

• Corrected validation logic for max_worker_processes to ensure that GUC values for replica groups aren't incorrectly validated as lower than the primary group.
• Fixed an issue where submitting an empty GUCs configuration payload didn't preserve the existing settings, instead incorrectly triggering a reset to default values.
• Ensured that when duplicating a data group, the GUCs (such as max_connections) are properly included in the payload, preventing unnecessary validation failures.
• Resolved a bug that prevented the successful addition of a new PGD group.
• Fixed an issue that could cause PGD-X cluster creation to stall during the reconcile phase.
• When adding a new PGD data group, the database settings are now correctly duplicated from the first data group to maintain consistency.
• PGD cluster status visibility has been enhanced with new control plane resource (CR) status panels for cluster, clusterwrapper, and pgdgroup resources.
• The console now correctly reflects the effective max_wal_size of 8 GB in the HM console, resolving a discrepancy where the console displayed 1024 MB.
• The required shared_preload_libraries setting for PGD now correctly includes pgaa, or the HM console provides appropriate indication when the setting is incorrect.

Backup, recovery & storage

• Fixed a bug where cluster metadata wasn't completely removed after the retention period, leaving orphaned Barman files in S3/object storage.
• WAL files marked as completed are no longer incorrectly retained beyond the configured limits.
• Resolved an issue with the upm-beacon agent component to stabilize the backup deletion process.
• The cross-location cluster restore functionality now correctly uses pg_basebackup as the method.

Core platform functionality & API

• Fixed a bug that caused the accm-server to crash due to an invalid pod name format.
• Resolved an issue that could prevent cluster creation due to upper bounds limits on EKS.
• Fixed a bug that prevented the secondary location beacon dispatcher runner from connecting to the primary beacon-server.
• The Model Profile Paths field in the HM console is now correctly marked as optional during configuration.
• Fixed an API bug that caused certain status fields, such as status.backupStatus, to appear empty.
• Improved cluster metadata availability across the platform for enhanced reliability.
• Fixed an issue that caused an error when calling an LLM remotely, ensuring successful external API communication.
• Increased the default memory allocation for the beacon agent and server and made the limits configurable in the bootstrap values.
• Fixed a bug where a process group would cease retrying to connect after an hour, ensuring continuous attempts for connection and stability.