This research paper presents an original research design for tuning PostgreSQL parameters in Odoo deployments using Query Perception and Evolutionary Reinforcement Learning. Odoo is widely used as an open-source ERP platform for sales, accounting, inventory, manufacturing, procurement, and customer operations. However, as transactional data grows into millions of records, performance degradation often appears in reporting, background jobs, integrations, imports, dashboards, and API workloads. This paper argues that PostgreSQL tuning for Odoo should not rely only on static parameter templates. Odoo workloads are dynamic and change according to business time, module usage, transaction volume, reporting needs, scheduler activity, integration load, and month-end financial operations. The paper proposes an Odoo-specific research framework that captures real SQL workloads from pgₛtatₛtatements, autoₑxplain logs, and Odoo SQL logs; extracts workload features; classifies Odoo query patterns; predicts latency under PostgreSQL parameter configurations; and uses evolutionary reinforcement learning to recommend safe tuning actions. The proposed research module, thinqₚgₑrltuneₒdoo, is designed as a controlled tuning advisor rather than an uncontrolled auto-tuning engine. It collects query workload, classifies query patterns, captures PostgreSQL parameter snapshots, evaluates performance risk, generates tuning candidates, benchmarks safe configurations, records before-after results, and provides rollback plans. The contribution of this paper is a practical research direction for making PostgreSQL tuning in Odoo more adaptive, measurable, workload-aware, and production-safe. It combines database observability, Odoo-specific query perception, machine learning, evolutionary search, reinforcement learning, benchmark validation, and safety constraints into a repeatable performance optimization framework.
Pradana Mahroza (Tue,) studied this question.