@parcha/pi-durable-disk 0.1.0-beta.12 → 0.1.0-beta.14

package/dist/watchdog.js

@@ -1,8 +1,9 @@
 // The lease watchdog on a thread of its own. A store commit is synchronous, so a FUSE daemon that stops
 // answering blocks the instance's main thread inside the kernel, and with it every timer the main thread owns. This
-// thread never touches the mount: it reads a clock reading the main thread shares, /proc and the cgroup, and sends
-// signals. Once the lease has lapsed it kills the instance's commands on every look, until the main thread runs again,
+// thread never touches the mount: it reads a clock reading the main thread shares, /proc and the cgroup (on macOS, one
+// ps listing), and sends signals. Once the lease has lapsed it kills the instance's commands on every look, until the main thread runs again,
 // sees the flag and takes the normal fence path (exit 75).
+import * as child_process from "node:child_process";
 import * as fs from "node:fs";
 import * as path from "node:path";
 import { Worker } from "node:worker_threads";
@@ -49,20 +50,49 @@ function cgroupPids(instance, proc, sys) {
   }
 }
 
-function sweep(instance, options = {}) {
-  const proc = options.proc ?? "/proc";
-  const kill = options.kill ?? ((pid, signal) => process.kill(pid, signal));
+/** The instance's group and its direct children, from /proc. */
+function procTree(instance, proc) {
   const own = stat(proc, instance);
-  if (own === undefined) return 0;
-  const targets = new Set();
+  if (own === undefined) return undefined;
+  const children = [];
   for (const entry of fs.readdirSync(proc)) {
     const pid = Number(entry);
     if (!Number.isInteger(pid) || pid <= 0 || pid === instance) continue;
     const child = stat(proc, pid);
-    if (child === undefined || child.ppid !== instance) continue;
-    targets.add(child.pgrp === own.pgrp || child.pgrp <= 0 ? pid : -child.pgrp);
+    if (child !== undefined && child.ppid === instance) children.push({ pid, pgrp: child.pgrp });
+  }
+  return { pgrp: own.pgrp, children };
+}
+
+/** The same from one ps listing, where there is no /proc (macOS). */
+function psTree(instance, ps) {
+  let text;
+  try {
+    text = ps("/bin/ps", ["-A", "-o", "pid=,ppid=,pgid="]);
+  } catch {
+    return undefined;
+  }
+  const rows = [];
+  for (const line of text.split("\n")) {
+    const f = line.trim().split(/\s+/).map(Number);
+    if (f.length === 3 && f.every(Number.isInteger)) rows.push({ pid: f[0], ppid: f[1], pgrp: f[2] });
   }
-  if (options.ownCgroup) {
+  const own = rows.find((row) => row.pid === instance);
+  if (own === undefined) return undefined;
+  return { pgrp: own.pgrp, children: rows.filter((row) => row.ppid === instance && row.pid > 0 && row.pid !== instance) };
+}
+
+const runPs = (file, args) => child_process.execFileSync(file, args, { encoding: "utf8", stdio: ["ignore", "pipe", "ignore"], timeout: 10000 });
+
+function sweep(instance, options = {}) {
+  const proc = options.proc ?? "/proc";
+  const kill = options.kill ?? ((pid, signal) => process.kill(pid, signal));
+  const ps = options.ps ?? (options.proc === undefined && process.platform === "darwin" ? runPs : undefined);
+  const tree = ps === undefined ? procTree(instance, proc) : psTree(instance, ps);
+  if (tree === undefined) return 0;
+  const targets = new Set();
+  for (const child of tree.children) targets.add(child.pgrp === tree.pgrp || child.pgrp <= 1 ? child.pid : -child.pgrp);
+  if (options.ownCgroup && ps === undefined) {
     for (const pid of cgroupPids(instance, proc, options.sys ?? "/sys/fs/cgroup")) if (pid !== instance && pid > 0) targets.add(pid);
   }
   let sent = 0;
@@ -78,6 +108,7 @@ function sweep(instance, options = {}) {
 }
 `;
 const WORKER_SOURCE = `"use strict";
+const child_process = require("node:child_process");
 const fs = require("node:fs");
 const path = require("node:path");
 const { workerData } = require("node:worker_threads");
@@ -88,14 +119,14 @@ setInterval(() => {
   Atomics.add(shared, SLOT_KILLED, BigInt(sweep(workerData.instance, { ownCgroup: workerData.ownCgroup })));
 }, workerData.checkMs);
 `;
-const compiled = new Function("fs", "path", `"use strict";${LIB}\nreturn { lapsed, sweep };`)(fs, path);
+const compiled = new Function("child_process", "fs", "path", `"use strict";${LIB}\nreturn { lapsed, sweep };`)(child_process, fs, path);
 /** One look at the shared state at `now`. True once the lease has lapsed, for good; the first look records the age. */
 export const lapsed = compiled.lapsed;
 /**
  * SIGKILL the instance's commands: the process group of each direct child (pi starts each command as its own group's
  * leader), or the child alone when it shares the instance's group; with `ownCgroup`, every other process of the cgroup.
  * Never the instance or its group. A process this user may not signal (sudo's archil helpers) is skipped. Returns the
- * number of signals sent. `stat` lines are `pid (comm) state ppid pgrp ...`, read after the last `)` since comm may hold
+ * number of signals sent. Where there is no /proc (macOS) the processes come from one ps listing, and there is no cgroup. `stat` lines are `pid (comm) state ppid pgrp ...`, read after the last `)` since comm may hold
  * spaces and parentheses.
  */
 export const sweep = compiled.sweep;
@@ -110,7 +141,9 @@ export class LeaseWatchdog {
         Atomics.store(this.#shared, SLOT_BEAT, options.beatNs);
         Atomics.store(this.#shared, SLOT_LIMIT, BigInt(Math.round(options.limitMs * 1e6)));
         const data = { buffer, checkMs: options.checkMs, instance: process.pid, ownCgroup: options.ownCgroup === true };
-        this.#worker = new Worker(WORKER_SOURCE, { eval: true, workerData: data });
+        // No inherited Node flags: the source is self-contained CommonJS and needs none of the host's, and some would make Node
+        // read it as ESM (`--input-type=module`, a host running inline ESM through -e or stdin), where `require` is undefined.
+        this.#worker = new Worker(WORKER_SOURCE, { eval: true, workerData: data, execArgv: [] });
         this.#worker.unref();
     }
     /** A heartbeat write that started at `startNs` succeeded. */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@parcha/pi-durable-disk",
-  "version": "0.1.0-beta.12",
+  "version": "0.1.0-beta.14",
   "description": "A durable pi host on a network disk (Archil): one owner per run, a fence, durable commits, and a supervisor.",
   "keywords": [
     "pi",
@@ -30,6 +30,10 @@
       "types": "./dist/lease.d.ts",
       "default": "./dist/lease.js"
     },
+    "./internals": {
+      "types": "./dist/internals.d.ts",
+      "default": "./dist/internals.js"
+    },
     "./bin/archil-scoped": "./bin/archil-scoped",
     "./package.json": "./package.json"
   },
@@ -49,7 +53,7 @@
     "build": "rm -rf dist && tsc -p tsconfig.build.json && node scripts/rewrite-dts.mjs dist",
     "prepack": "npm run build",
     "typecheck": "tsc --noEmit && tsc --noEmit -p examples",
-    "test": "node --test 'test/*.test.ts'",
+    "test": "node --test --test-timeout=240000 'test/*.test.ts' 'test/examples/*.test.ts'",
     "test:acceptance": "node --test --test-concurrency=1 'test/acceptance/*.test.ts'",
     "test:live": "node --test --test-concurrency=1 'test/live/*.test.ts'",
     "test:live:claim": "node --test --test-concurrency=1 test/live/claim.live.test.ts",
@@ -58,20 +62,21 @@
     "test:live:run": "node --test --test-concurrency=1 test/live/run.live.test.ts",
     "test:live:store": "node --test --test-concurrency=1 test/live/store.live.test.ts",
     "test:live:supervise": "node --test --test-concurrency=1 test/live/supervise.live.test.ts",
+    "test:live:supervise-slow": "node --test --test-concurrency=1 test/live/supervise-slow.live.test.ts",
     "test:live:docker": "PDA_LIVE_DOCKER=1 node --test --test-concurrency=1 test/live/docker.live.test.ts test/live/docker-park.live.test.ts",
     "verify:tarball": "node scripts/verify-tarball.mjs"
   },
   "peerDependencies": {
-    "@earendil-works/chord": "~1.0.4",
-    "@earendil-works/pi-durable": "~1.0.4"
+    "@earendil-works/chord": ">=1.0.4 <1.4.0",
+    "@earendil-works/pi-durable": ">=1.0.4 <1.4.0"
   },
   "dependencies": {
     "disk": "1.7.0"
   },
   "devDependencies": {
-    "@earendil-works/chord": "1.0.4",
-    "@earendil-works/pi-ai": "1.0.4",
-    "@earendil-works/pi-durable": "1.0.4"
+    "@earendil-works/chord": "1.1.0",
+    "@earendil-works/pi-ai": "1.1.0",
+    "@earendil-works/pi-durable": "1.1.0"
   },
   "repository": {
     "type": "git",