AgentRules.org
RulesStacksClientsAuditorPrunerSelf-HealRFCsBenchmarkCLICI/CDAuthorsSubmit
This specification is maintained by the AgentRules Working Group. Have enhancements or want to claim maintainership?Submit Specification →
backendcritical PriorityID: postgres-17-vector-streaming-io-memory-lock-standard

Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing

Enforces strict session-level buffer ring isolation, deterministic maintenance work memory capping, and asynchronous block-level streaming I/O configurations during concurrent high-dimensional HNSW index builds in PostgreSQL 17. Prevents shared buffer pool cache evictions and Linux dirty page kernel thrashing from degrading latency-sensitive co-located OLTP transactions.

Prohibited Anti-Pattern: Global Maintenance Memory Bloat and Shared Buffer Eviction Cascades Under Unbounded Vector Indexing

Export Ready-to-Drop Configuration

Select format below to copy code or download file directly.

---
description: Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing - Enforces strict session-level buffer ring isolation, deterministic maintenance work memory capping, and asynchronous block-level streaming I/O configurations during concurrent high-dimensional HNSW index builds in PostgreSQL 17. Prevents shared buffer pool cache evictions and Linux dirty page kernel thrashing from degrading latency-sensitive co-located OLTP transactions.
globs: src/database/**/*.ts, src/migrations/**/*.ts, src/jobs/**/*.ts, src/services/indexing/**/*.ts
alwaysApply: true
verified_by: "agentrules.org"
license: "CC-BY-4.0"
source: "https://agentrules.org/rules/postgres-17-vector-streaming-io-memory-lock-standard"
---

# Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing

> **Severity**: CRITICAL | **Category**: backend
> **Anti-Pattern Warning**: Global Maintenance Memory Bloat and Shared Buffer Eviction Cascades Under Unbounded Vector Indexing

## System Constraints
You are a Principal Database Architect and Systems Reliability Engineer enforcing the PostgreSQL 17 Block-Level Streaming I/O and Vector Indexing Memory Lock Standard. Your mandate is to prevent buffer pool cache eviction storms, page cache thrashing, and transaction stalling during high-dimensional pgvector (HNSW) index builds executing concurrently with mission-critical OLTP workloads. Mandate explicit session-scoped GUC isolation, ring buffer containment via CREATE INDEX CONCURRENTLY, strict maintenance_work_mem boundaries relative to container cgroup limits, and PG17 asynchronous block streaming parameter enforcement.

## Strict Guidelines
- Engineers MUST ALWAYS isolate vector index creation within dedicated connection sessions using non-persistent 'SET LOCAL' or connection-scoped 'SET' configurations, explicitly preventing modified GUC parameters from leaking into connection pools.
- Engineers MUST ALWAYS set 'maintenance_work_mem' to a deterministic threshold that fits entirely within physical RAM while preventing OS page thrashing (e.g., maintaining at least 30% headroom below container cgroup limits when combined with shared_buffers).
- Engineers MUST ALWAYS execute vector index operations using 'CREATE INDEX CONCURRENTLY' with pgvector HNSW to utilize ring-buffer streaming patterns and avoid global table lock acquisitions.
- Engineers MUST configure Postgres 17 vector build sessions with optimized streaming I/O parameters ('SET io_combine_limit = ...' and appropriate 'max_parallel_maintenance_workers') to prevent saturation of the OS page writeback subsystem.
- Engineers MUST NOT execute HNSW index builds on high-dimensional vectors (e.g., 1536d+) inside ambient application transaction blocks, standard connection pool checkouts, or without transiently pinning 'maintenance_work_mem' per worker.
- Services MUST monitor pg_stat_activity, pg_statio_user_tables, and pg_stat_io during vector re-indexing to trigger emergency circuit breakers if OLTP buffer cache hit ratios drop below 99%.

## Bad vs Good Implementations
### Unbounded Direct Indexing vs Isolated Streaming Session Context

#### Bad Antipattern (AI Common Hallucination)
> **Why it fails**: Executes an unbounded, non-concurrent HNSW index creation inside a general pooled connection. Lacks session-isolated maintenance_work_mem limits, causes massive shared_buffers evictions of active OLTP hot pages, and locks the 'items' table against concurrent writes.
```typescript
import { Pool } from 'pg';

export async function rebuildVectorIndex(pool: Pool): Promise<void> {
  // ANTI-PATTERN: Executes index build on ambient shared connection without memory gating,
  // saturating shared_buffers and triggering kernel page writeback storms on active OLTP nodes.
  const client = await pool.connect();
  try {
    await client.query('CREATE INDEX IF NOT EXISTS items_embedding_idx ON items USING hnsw (embedding vector_cosine_ops);');
  } finally {
    client.release();
  }
}
```

#### Production Standard (Correct)
> **Standard**: Guarantees session-scoped memory allocation limits, leverages PG17 streaming block I/O parameterization ('io_combine_limit'), prevents table locks using 'CONCURRENTLY', and cleans up session GUC mutations via 'RESET ALL' before returning the client to the pool.
```typescript
import { Pool, PoolClient } from 'pg';

interface VectorIndexOptions {
  maintenanceWorkMemMb: number;
  maxParallelWorkers: number;
  ioCombineLimitPages: number;
}

export async function rebuildVectorIndexIsolated(
  pool: Pool,
  options: VectorIndexOptions = { maintenanceWorkMemMb: 2048, maxParallelWorkers: 2, ioCombineLimitPages: 16 }
): Promise<void> {
  const client: PoolClient = await pool.connect();
  try {
    // Guardrail: Apply localized session GUCs to isolate memory allocation and leverage PG17 streaming I/O
    await client.query(`
      SET maintenance_work_mem = '${options.maintenanceWorkMemMb}MB';
      SET max_parallel_maintenance_workers = ${options.maxParallelWorkers};
      SET io_combine_limit = ${options.ioCombineLimitPages};
      SET synchronous_commit = OFF;
    `);

    // Invariant: Non-blocking concurrent execution utilizing streaming ring buffers
    await client.query(`
      CREATE INDEX CONCURRENTLY IF NOT EXISTS items_embedding_hnsw_idx
      ON items USING hnsw (embedding vector_cosine_ops)
      WITH (m = 16, ef_construction = 64);
    `);
  } finally {
    // Guardrail: Reset connection state before releasing back to the shared application pool
    await client.query('DISCARD SEQUENCES; DISCARD TEMP; RESET ALL;');
    client.release();
  }
}
```
Recommended MCP Servers (mcp-serverlist.com)

Equip agent with system tools to verify code for this stack.

Postgres MCP Server
Inspects real-time buffer cache usage, pg_stat_io metrics, and running backend memory consumption.
Recommended Agent Skills (askills.dev)

Ready-to-use executable skill functions tailored to this stack.

Postgres Index Health Auditor
Audits pgvector HNSW configurations, buffer hit ratios, and background writer metrics under high write concurrency.
Architectural Specification & Invariant Rationale

Architectural Invariant Design: Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing

Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing defines an industrial-grade behavioral standard for backend engineering. Enforces strict session-level buffer ring isolation, deterministic maintenance work memory capping, and asynchronous block-level streaming I/O configurations during concurrent high-dimensional HNSW index builds in PostgreSQL 17. Prevents shared buffer pool cache evictions and Linux dirty page kernel thrashing from degrading latency-sensitive co-located OLTP transactions. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: Global Maintenance Memory Bloat and Shared Buffer Eviction Cascades Under Unbounded Vector Indexing Without an explicit system instruction file, developers spend hours debugging silent failures.
Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure ModeAI Symptom (The Anti-Pattern)SeverityProduction Impact & Remedy
AI Hallucination & Deprecation DriftGlobal Maintenance Memory Bloat and Shared Buffer Eviction Cascades Under Unbounded Vector IndexingCritical
Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production.
Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture InconsistencyAI assistants produce mixed paradigms across files, creating fragmented codebase conventions.Medium
Increases cognitive load during code reviews and leads to maintenance debt across engineering teams.
Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.
Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

1

Engineers MUST ALWAYS isolate vector index creation within dedicated connection sessions using non-persistent 'SET LOCAL' or connection-scoped 'SET' configurations, explicitly preventing modified GUC parameters from leaking into connection pools.

Engineers MUST ALWAYS isolate vector index creation within dedicated connection sessions using non-persistent 'SET LOCAL' or connection-scoped 'SET' configurations, explicitly preventing modified GUC parameters from leaking into connection pools.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
2

Engineers MUST ALWAYS set 'maintenance_work_mem' to a deterministic threshold that fits entirely within physical RAM while preventing OS page thrashing (e.g., maintaining at least 30% headroom below container cgroup limits when combined with shared_buffers).

Engineers MUST ALWAYS set 'maintenance_work_mem' to a deterministic threshold that fits entirely within physical RAM while preventing OS page thrashing (e.g., maintaining at least 30% headroom below container cgroup limits when combined with shared_buffers).

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
3

Engineers MUST ALWAYS execute vector index operations using 'CREATE INDEX CONCURRENTLY' with pgvector HNSW to utilize ring-buffer streaming patterns and avoid global table lock acquisitions.

Engineers MUST ALWAYS execute vector index operations using 'CREATE INDEX CONCURRENTLY' with pgvector HNSW to utilize ring-buffer streaming patterns and avoid global table lock acquisitions.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
4

Engineers MUST configure Postgres 17 vector build sessions with optimized streaming I/O parameters ('SET io_combine_limit = ...' and appropriate 'max_parallel_maintenance_workers') to prevent saturation of the OS page writeback subsystem.

Engineers MUST configure Postgres 17 vector build sessions with optimized streaming I/O parameters ('SET io_combine_limit = ...' and appropriate 'max_parallel_maintenance_workers') to prevent saturation of the OS page writeback subsystem.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
5

Engineers MUST NOT execute HNSW index builds on high-dimensional vectors (e.g., 1536d+) inside ambient application transaction blocks, standard connection pool checkouts, or without transiently pinning 'maintenance_work_mem' per worker.

Engineers MUST NOT execute HNSW index builds on high-dimensional vectors (e.g., 1536d+) inside ambient application transaction blocks, standard connection pool checkouts, or without transiently pinning 'maintenance_work_mem' per worker.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
6

Services MUST monitor pg_stat_activity, pg_statio_user_tables, and pg_stat_io during vector re-indexing to trigger emergency circuit breakers if OLTP buffer cache hit ratios drop below 99%.

Services MUST monitor pg_stat_activity, pg_statio_user_tables, and pg_stat_io during vector re-indexing to trigger emergency circuit breakers if OLTP buffer cache hit ratios drop below 99%.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
Technical Questions & Answers

Frequently Asked Technical Questions

Common questions developers face when configuring AI coding assistants for this stack:

Q:Why is Postgres 17 Block-Level Streaming I/O and Isolated Memory Context for HNSW Vector Re-indexing critical for AI coding assistants?

Enforces strict session-level buffer ring isolation, deterministic maintenance work memory capping, and asynchronous block-level streaming I/O configurations during concurrent high-dimensional HNSW index builds in PostgreSQL 17. Prevents shared buffer pool cache evictions and Linux dirty page kernel thrashing from degrading latency-sensitive co-located OLTP transactions. Without explicit behavioral boundaries, AI coding assistants (Cursor, Claude Code, Windsurf, Codex) rely on training data averages, which frequently produce deprecated, inefficient, or vulnerable code patterns.

Q:How does adopting this rule prevent AI hallucinations?

This rule explicitly declares what the AI must avoid (Global Maintenance Memory Bloat and Shared Buffer Eviction Cascades Under Unbounded Vector Indexing) and prescribes verified positive examples. Grounding the LLM with negative and positive pairs eliminates guesswork.

Q:Which AI coding environments are supported?

This specification is multi-platform compatible: Cursor (.cursor/rules/*.mdc & .cursorrules), Anthropic Claude Code (CLAUDE.md), Windsurf (.windsurfrules), Codex / Cline (AGENTS.md), and GitHub Copilot (.github/copilot-instructions.md).

Q:Can I combine this rule with other technology stacks?

Yes. Use the Interactive Stack Builder or download an Agent Pack to merge this rule with other frontend, backend, database, security, and DevOps rules without syntax collision.

Open Specification · CC BY 4.0·Free for Personal, Team & Commercial In-Repo Use

Citation & Open Source Attribution Guidelines

This specification is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). You are completely free to adapt and share this rule in your workspace. If citing, including in public AI benchmark datasets, or redistributing, attribution to AgentRules.org is required.

Related BACKEND Rules

View all rules →
backend
42

Fastify 5 Schema Compilers & Encapsulation Production Standard

Fastify relies on lexical encapsulation contexts to sandbox decorators, hooks, and compiled validation/serialization schemas without memory leaks or cross-tenant bleed. In Fastify 5, misuse of global compilers or failing to encapsulate plugin boundaries causes compiled `fast-json-stringify` serializers and Ajv validators to either crash on duplicate IDs, leak private data across sibling scopes, or bypass JIT compilation optimizations. Production services must enforce strict encapsulation hierarchies and type-safe schema compilation across every routing subtree.

backend
68

Bun Embedded SQLite WAL Mode & Transaction Synchronization Standard

Production database standard for bun:sqlite. Mandates enabling Write-Ahead Logging (PRAGMA journal_mode = WAL), busy timeouts, and using db.transaction() for atomic operations without lock starvation.

backend
68

Drizzle ORM Batch API & Interactive Transaction Isolation Standard

Production database standard for Drizzle ORM (PostgreSQL & MySQL). Mandates using db.batch() over HTTP drivers to prevent round-trip waterfalls, explicit db.transaction() with serializable/read-committed isolation, and strictly avoids unawaited query promises.

1 Rule in Stack

Postgres