Deterministic Heap-Capped Worker Isolation & Lifecycle Eviction in Bun 1.2

--- description: Deterministic Heap-Capped Worker Isolation & Lifecycle Eviction in Bun 1.2 - Enforces strict memory boundaries and tenant isolation for high-density multi-tenant workloads in Bun 1.2. Mandates deterministic per-tenant worker recycling based on post-execution heap inspection and strict memory limits, preventing out-of-memory (OOM) cascade failures across shared compute nodes. globs: src/workers/**/*.{ts,js}, src/serverless/**/*.{ts,js}, src/services/pool/**/*.{ts,js}, src/runtime/**/*.{ts,js} alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/bun-multi-tenant-worker-heap-sandboxing" --- # Deterministic Heap-Capped Worker Isolation & Lifecycle Eviction in Bun 1.2 > **Severity**: CRITICAL | **Category**: backend > **Anti-Pattern Warning**: Unbounded In-Process Tenant Execution and Unmonitored Worker Reuse Leading to Host OOM Cascades ## System Constraints You are a Principal Systems Engineer enforcing strict tenant isolation and memory bounding under Bun 1.2. Never allow untrusted multi-tenant code to execute directly in the main thread or within unmetered, long-lived workers. Always mandate deterministic worker lifecycle control, memory usage verification via process.memoryUsage() / heap sampling, and proactive worker termination when per-tenant memory ceilings are breached. ## Strict Guidelines - Engineers MUST ALWAYS isolate tenant-provided execution within a dedicated `Worker` instance configured with explicit execution timeouts and message protocols. - Engineers MUST NOT allow workers to persist indefinitely without evaluating cumulative allocated heap post-invocation; workers exceeding tenant heap thresholds MUST be terminated immediately. - Services MUST enforce an explicit hard cap (e.g., 64MB-128MB per tenant worker) and dispatch an abort signal via `worker.terminate()` if execution duration or memory boundaries exceed service limits. - Code SHOULD implement an active worker pool that tracks RSS and heap allocation per tenant ID, evicting warm instances when system-level memory pressure approaches threshold metrics. - Components NEVER share global module state, native resource handles, or database connection clients directly across worker boundaries without explicit IPC sanitization. ## Bad vs Good Implementations ### Unbounded Worker Reuse vs. Deterministic Capped Worker Pool with Eviction #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Shared worker executes tasks from multiple tenants concurrently without memory accounting, IPC segregation, or lifecycle bounds. A single rogue tenant allocating excessive objects causes an unrecoverable process crash or memory bleed across tenants. ```typescript import { type Worker } from "bun"; // ANTI-PATTERN: Indefinitely reusing workers across arbitrary tenants with no memory boundaries const sharedWorker = new Worker(new URL("./tenant-runner.ts", import.meta.url)); export async function executeTenantTask(tenantId: string, payload: unknown): Promise<unknown> { return new Promise((resolve, reject) => { sharedWorker.onmessage = (event) => resolve(event.data); sharedWorker.onerror = (err) => reject(err); // Cross-tenant data bleed and unbounded memory retention sharedWorker.postMessage({ tenantId, payload }); }); } ``` #### Production Standard (Correct) > **Standard**: Spawns isolated tenant runtime, enforces execution deadlines via timeouts, samples post-execution memory usage from worker IPC, and proactively terminates the worker to guarantee memory reclamation and avoid cross-tenant leaks. ```typescript import { type Worker } from "bun"; interface TaskResult { result: unknown; heapUsedBytes: number; } const MAX_TENANT_HEAP_BYTES = 64 * 1024 * 1024; // 64 MB strict tenant limit const EXECUTION_TIMEOUT_MS = 5_000; export async function executeSandboxedTenantTask( tenantId: string, payload: unknown ): Promise<unknown> { const worker = new Worker(new URL("./tenant-runner.ts", import.meta.url)); return new Promise((resolve, reject) => { const timeoutId = setTimeout(() => { worker.terminate(); reject(new Error(`[Tenant:${tenantId}] Execution timed out after ${EXECUTION_TIMEOUT_MS}ms`)); }, EXECUTION_TIMEOUT_MS); worker.onmessage = (event: MessageEvent<TaskResult>) => { clearTimeout(timeoutId); const { result, heapUsedBytes } = event.data; // Deterministic teardown if tenant exceeds memory budget worker.terminate(); if (heapUsedBytes > MAX_TENANT_HEAP_BYTES) { reject(new Error(`[Tenant:${tenantId}] Heap limit exceeded: ${heapUsedBytes} bytes`)); return; } resolve(result); }; worker.onerror = (err) => { clearTimeout(timeoutId); worker.terminate(); reject(err); }; worker.postMessage({ tenantId, payload }); }); } ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: Deterministic Heap-Capped Worker Isolation & Lifecycle Eviction in Bun 1.2

Deterministic Heap-Capped Worker Isolation & Lifecycle Eviction in Bun 1.2 defines an industrial-grade behavioral standard for backend engineering. Enforces strict memory boundaries and tenant isolation for high-density multi-tenant workloads in Bun 1.2. Mandates deterministic per-tenant worker recycling based on post-execution heap inspection and strict memory limits, preventing out-of-memory (OOM) cascade failures across shared compute nodes. 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: Unbounded In-Process Tenant Execution and Unmonitored Worker Reuse Leading to Host OOM Cascades 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 Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	Unbounded In-Process Tenant Execution and Unmonitored Worker Reuse Leading to Host OOM Cascades	Critical	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 Inconsistency	AI 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:

Engineers MUST ALWAYS isolate tenant-provided execution within a dedicated `Worker` instance configured with explicit execution timeouts and message protocols.

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.