Files
hyperframes/packages/cli/src/telemetry/system.ts
T
Miguel Ángel 2ce5b421f1 fix(engine): respect cgroup memory limits in low-memory detection (#1373)
getSystemTotalMb returned os.totalmem() — the host's physical RAM — so a
4GB Docker container on a 32GB host never auto-flagged as low-memory and
the low-memory render profile didn't activate exactly where it's needed
most. Read the cgroup v2 limit (/sys/fs/cgroup/memory.max, with the v1
fallback and its no-limit sentinel handled) and use min(host, cgroup).
The probe is best-effort and non-Linux platforms never touch /sys.

Review follow-ups: worker sizing (calculateOptimalWorkers) and the
getSystemResources diagnostics previously read os.totalmem() directly
and now use getSystemTotalMb(), so container limits actually govern
parallel spawn decisions; CLI telemetry reports the effective total as
well. The cgroup probe result is cached for the process lifetime (the
limit is immutable per process) with a test reset hook; a detected limit
logs once so operators can see which source governs, and a
present-but-unreadable cgroup file warns once instead of failing
silently — absence stays silent. The root-path-vs-/proc/self/cgroup
trade-off is documented at the path constants. cli/tsconfig.json gains
the gcp-cloud-run/sdk source alias (matching the existing producer and
aws-lambda entries) so the cli typecheck resolves from source in a
fresh checkout.

Refs #1193, #1194, #1195, #1236
2026-06-12 12:21:43 -04:00

214 lines
6.9 KiB
TypeScript

import { cpus, freemem, platform, release } from "node:os";
import { existsSync, readFileSync, statfsSync } from "node:fs";
import { execSync } from "node:child_process";
import { getSystemTotalMb } from "@hyperframes/engine";
import {
detectAgentRuntime,
detectSandboxRuntime,
type AgentRuntime,
type SandboxRuntime,
} from "./agent_runtime.js";
import { detectWSL } from "./platform.js";
// ---------------------------------------------------------------------------
// System metadata collected once per CLI session and attached to all events.
// Follows the same patterns as Next.js, Turborepo, and Gatsby telemetry.
// No PII — only hardware/environment characteristics useful for debugging.
// ---------------------------------------------------------------------------
/** Convert bytes to whole megabytes. */
export function bytesToMb(bytes: number): number {
return Math.trunc(bytes / (1024 * 1024));
}
export interface SystemMeta {
os_release: string;
cpu_count: number;
cpu_model: string | null;
cpu_speed: number | null;
memory_total_mb: number;
is_docker: boolean;
is_ci: boolean;
ci_name: string | null;
is_wsl: boolean;
is_tty: boolean;
/**
* Managed sandbox runtime hosting this invocation, when one is detectable
* (gvisor / firecracker / docker / kvm / wsl). null on a normal dev
* machine. Lets us distinguish "real laptop" from "ephemeral cloud
* sandbox driving the CLI" without geo guesswork.
*/
sandbox_runtime: SandboxRuntime;
/**
* Coding-agent vendor that spawned this process, if any (see the
* `AgentRuntime` union in agent_runtime.ts for the full, current set).
* Most rules check env-var existence only — values are never read; a few
* use filesystem/kernel markers (e.g. the Gemini managed-agent mount).
* Every rule keys on a marker with a source citation in agent_runtime.ts;
* unverified guesses are deliberately omitted (false-negative > guess).
* null when no agent is detected.
*/
agent_runtime: AgentRuntime;
}
let cached: SystemMeta | null = null;
/**
* Collect system metadata. Cached after first call.
* Only includes static values — use `freemem()` directly for volatile readings.
*/
export function getSystemMeta(): SystemMeta {
if (cached) return cached;
const cpuInfo = cpus();
const firstCpu = cpuInfo[0] ?? null;
cached = {
os_release: release(),
cpu_count: cpuInfo.length,
cpu_model: firstCpu?.model?.trim() ?? null,
cpu_speed: firstCpu?.speed ?? null,
memory_total_mb: getSystemTotalMb(),
is_docker: detectDocker(),
is_ci: detectCI(),
ci_name: getCIName(),
is_wsl: detectWSL(),
is_tty: Boolean(process.stdout?.isTTY),
sandbox_runtime: detectSandboxRuntime(),
agent_runtime: detectAgentRuntime(),
};
return cached;
}
// ---------------------------------------------------------------------------
// Environment detectors
// ---------------------------------------------------------------------------
function detectDocker(): boolean {
// Standard detection: /.dockerenv file or "docker" in /proc/1/cgroup
try {
if (existsSync("/.dockerenv")) return true;
if (platform() === "linux") {
const cgroup = readFileSync("/proc/1/cgroup", "utf-8");
if (cgroup.includes("docker") || cgroup.includes("containerd")) return true;
}
} catch {
// Ignore — not in Docker
}
return false;
}
// Named providers come first so getCIName() picks the most specific match.
// `truthy` accepts 'true' or '1'; `presence` matches any non-null value.
type CIProvider =
| { name: string | null; envVar: string; mode: "truthy" }
| { name: string | null; envVar: string; mode: "presence" };
const CI_PROVIDERS: CIProvider[] = [
{ name: "github_actions", envVar: "GITHUB_ACTIONS", mode: "truthy" },
{ name: "gitlab_ci", envVar: "GITLAB_CI", mode: "truthy" },
{ name: "circleci", envVar: "CIRCLECI", mode: "truthy" },
{ name: "jenkins", envVar: "JENKINS_URL", mode: "presence" },
{ name: "buildkite", envVar: "BUILDKITE", mode: "truthy" },
{ name: "travis", envVar: "TRAVIS", mode: "truthy" },
{ name: null, envVar: "CONTINUOUS_INTEGRATION", mode: "truthy" },
{ name: null, envVar: "CI", mode: "truthy" },
];
function matchesProvider(p: CIProvider): boolean {
const v = process.env[p.envVar];
if (p.mode === "presence") return v != null;
return v === "true" || v === "1";
}
function detectCI(): boolean {
return CI_PROVIDERS.some(matchesProvider);
}
function getCIName(): string | null {
for (const provider of CI_PROVIDERS) {
if (provider.name && matchesProvider(provider)) return provider.name;
}
return detectCI() ? "unknown" : null;
}
// ---------------------------------------------------------------------------
// Extended hardware checks (for doctor command and detailed render events)
// ---------------------------------------------------------------------------
/**
* Get /dev/shm size in MB (Linux only). Chrome uses shared memory heavily;
* Docker's default 64MB limit causes crashes.
*/
export function getShmSizeMb(): number | null {
if (platform() !== "linux") return null;
try {
const stats = statfsSync("/dev/shm");
return bytesToMb(stats.bsize * stats.blocks);
} catch {
return null;
}
}
/**
* Get available disk space in MB at a given path.
*/
export function getFreeDiskMb(path: string = "."): number | null {
try {
const stats = statfsSync(path);
return bytesToMb(stats.bsize * stats.bavail);
} catch {
return null;
}
}
/**
* Get available memory in MB, accounting for OS-level page caching.
*
* `os.freemem()` on macOS returns only truly free pages — ignoring
* inactive/purgeable/speculative pages that the kernel reclaims on demand.
* On a 24 GB Mac this reports ~0.1 GB "free" when ~5 GB is actually
* available. Linux has a similar (milder) issue; its kernel exposes the
* correct value via `MemAvailable` in /proc/meminfo.
*/
export function getAvailableMemoryMb(): number {
const fallback = bytesToMb(freemem());
if (platform() === "darwin") {
try {
const raw = execSync("vm_stat", { encoding: "utf-8", timeout: 5000 });
const pageSize = parseInt(raw.match(/page size of (\d+)/)?.[1] ?? "0", 10);
if (!pageSize) return fallback;
const pages = (key: string) =>
parseInt(raw.match(new RegExp(`${key}:\\s+(\\d+)`))?.[1] ?? "0", 10);
const available =
(pages("Pages free") +
pages("Pages inactive") +
pages("Pages purgeable") +
pages("Pages speculative")) *
pageSize;
return available > 0 ? bytesToMb(available) : fallback;
} catch {
return fallback;
}
}
if (platform() === "linux") {
try {
const meminfo = readFileSync("/proc/meminfo", "utf-8");
const match = meminfo.match(/MemAvailable:\s+(\d+)\s+kB/);
if (match) {
return Math.trunc(parseInt(match[1]!, 10) / 1024);
}
return fallback;
} catch {
return fallback;
}
}
return fallback;
}