Database Technology

Christophe Pettus highlights two new GUCs, log_startup_progress_interval and log_recovery_conflict_waits, designed to surface startup diagnostics previously buried in server logs. These settings allow operators to explicitly track WAL replay progress on primaries or diagnose standby stalls caused by recovery conflicts. By making these metrics configurable, DBAs can now monitor replication health without parsing raw log files.

• Use log_startup_progress_interval to monitor WAL replay rates directly.
• Enable log_recovery_conflict_waits to identify standby stalls from conflicts.
• Eliminate need for complex log parsing to detect replication lag.
• Improve observability for HA failover and maintenance windows.

Amazon Linux 2023 has been added as a first-class target for the PostgreSQL Global Development Group (PGDG) YUM repository. This eliminates the previous reliance on older base OS packages or manual source builds for EC2 instances. The new package tree aligns AL2023 with other major distributions, providing day-one minor releases and the full extension ecosystem.

• AL2023 now receives day-one PostgreSQL minor releases via yum.postgresql.org
• Access the complete PGDG extension ecosystem without manual compilation
• Standardizes package layout across RHEL, Fedora, SUSE, and AL2023
• Removes dependency on potentially outdated Amazon Linux base repo versions

The session_replication_role parameter controls whether triggers and rules fire, but it does not disable all integrity checks. As demonstrated, foreign key constraints enforced by internal triggers remain active, yet user-defined check constraints continue to validate data even when the role is set to replica. This distinction means that setting the role to replica does not provide a blanket exemption from all database constraints.

• session_replication_role only suppresses triggers and rules, not all constraint enforcement.
• Foreign key checks often rely on internal triggers and may still block invalid data.
• Check constraints remain active regardless of the session replication role setting.
• Do not rely on replica mode to bypass data validation during bulk loads.

PostgreSQL 19 Beta 3 shipped on August 13, 2026, with general availability expected in the coming weeks based on the project's typical September or October release cadence. The release notes highlight specific compatibility breaks that could disrupt existing applications if ignored during upgrade planning. Additionally, new SQL-level features have been introduced that offer tangible utility for developers and database administrators.

• PostgreSQL 19 GA is imminent, likely landing in September or October 2026.
• Review compatibility breaks carefully to prevent upgrade failures in production.
• New SQL features in Beta 3 warrant testing for potential performance gains.
• Beta 3 codebase is stable enough for preliminary feature evaluation.

Amazon Aurora MySQL-Compatible Edition 8.4.8 is now generally available, introducing support for post-quantum TLS key exchange to secure data in transit. The release adds transaction timeout controls to prevent long-running transactions from blocking InnoDB purge operations. It also enables multi-source replication for consolidating data from multiple sources and configurable delayed replication to protect against accidental data loss.

• Enable PQ-TLS key exchange for post-quantum cryptography on data in transit.
• Configure transaction timeouts to mitigate InnoDB purge blocking from long queries.
• Use multi-source replication to consolidate shards or merge data into a single replica.
• Set delayed replication lag to guard against accidental writes or deletes.
• Review security enhancements and bug fixes in this general availability release.

Fleet impact: Aurora MySQL 8.4.8 on AWS Aurora PostgreSQL/MySQL + RDS: Enable PQ-TLS for future-proofing in-transit encryption and configure transaction timeouts to stabilize InnoDB purge performance. Verify multi-source replication topology if consolidating shards, and test delayed replication settings to ensure they meet RPO requirements for accidental deletion protection.