@harperfast/harper-pro 5.0.1 → 5.0.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/analytics/profile.ts +47 -0
- package/core/components/Application.ts +24 -9
- package/core/components/ApplicationScope.ts +2 -3
- package/core/components/componentLoader.ts +13 -2
- package/core/config/harperConfigEnvVars.ts +34 -0
- package/core/resources/Table.ts +12 -1
- package/core/resources/analytics/write.ts +7 -10
- package/core/resources/indexes/HierarchicalNavigableSmallWorld.ts +23 -7
- package/core/security/jsLoader.ts +68 -22
- package/core/security/user.ts +10 -8
- package/core/static/defaultConfig.yaml +1 -1
- package/dist/analytics/profile.js +47 -0
- package/dist/analytics/profile.js.map +1 -1
- package/dist/core/components/Application.js +15 -5
- package/dist/core/components/Application.js.map +1 -1
- package/dist/core/components/ApplicationScope.js +2 -3
- package/dist/core/components/ApplicationScope.js.map +1 -1
- package/dist/core/components/componentLoader.js +11 -2
- package/dist/core/components/componentLoader.js.map +1 -1
- package/dist/core/config/harperConfigEnvVars.js +33 -0
- package/dist/core/config/harperConfigEnvVars.js.map +1 -1
- package/dist/core/resources/Table.js +12 -1
- package/dist/core/resources/Table.js.map +1 -1
- package/dist/core/resources/analytics/write.js +7 -10
- package/dist/core/resources/analytics/write.js.map +1 -1
- package/dist/core/resources/indexes/HierarchicalNavigableSmallWorld.js +24 -7
- package/dist/core/resources/indexes/HierarchicalNavigableSmallWorld.js.map +1 -1
- package/dist/core/security/jsLoader.js +54 -21
- package/dist/core/security/jsLoader.js.map +1 -1
- package/dist/core/security/user.js +9 -8
- package/dist/core/security/user.js.map +1 -1
- package/dist/replication/setNode.js +5 -2
- package/dist/replication/setNode.js.map +1 -1
- package/dist/security/certificate.js +28 -6
- package/dist/security/certificate.js.map +1 -1
- package/npm-shrinkwrap.json +2 -2
- package/package.json +1 -1
- package/replication/setNode.ts +5 -2
- package/security/certificate.ts +28 -6
- package/static/defaultConfig.yaml +1 -1
- package/studio/web/assets/{index-C0iJWrnF.js → index-f5-e8ocl.js} +5 -5
- package/studio/web/assets/{index-C0iJWrnF.js.map → index-f5-e8ocl.js.map} +1 -1
- package/studio/web/assets/{index.lazy-C647wC7n.js → index.lazy-CCd1vMot.js} +2 -2
- package/studio/web/assets/{index.lazy-C647wC7n.js.map → index.lazy-CCd1vMot.js.map} +1 -1
- package/studio/web/assets/{profile-BTS_ZjxV.js → profile-gjpePJuu.js} +2 -2
- package/studio/web/assets/{profile-BTS_ZjxV.js.map → profile-gjpePJuu.js.map} +1 -1
- package/studio/web/assets/{status-Dc-S5M23.js → status-CmoVx0A5.js} +2 -2
- package/studio/web/assets/{status-Dc-S5M23.js.map → status-CmoVx0A5.js.map} +1 -1
- package/studio/web/index.html +1 -1
package/analytics/profile.ts
CHANGED
|
@@ -6,6 +6,10 @@ import { recordAction } from '../core/resources/analytics/write.ts';
|
|
|
6
6
|
import { getHdbBasePath } from '../core/utility/environment/environmentManager.js';
|
|
7
7
|
import { PACKAGE_ROOT } from '../core/utility/packageUtils.js';
|
|
8
8
|
import { realpathSync, readFileSync, readdirSync } from 'node:fs';
|
|
9
|
+
import { execFile } from 'node:child_process';
|
|
10
|
+
import { promisify } from 'node:util';
|
|
11
|
+
|
|
12
|
+
const execFileAsync = promisify(execFile);
|
|
9
13
|
import { time as timeProfiler } from '@datadog/pprof';
|
|
10
14
|
import { getWorkerIndex } from '../core/server/threads/manageThreads.js';
|
|
11
15
|
import * as log from '../core/utility/logging/harper_logger.js';
|
|
@@ -42,6 +46,7 @@ export function handleApplication({ options }: Scope) {
|
|
|
42
46
|
}, 1000); // wait for everything to load before we start the profiler
|
|
43
47
|
}
|
|
44
48
|
let lastChildCpuTime = 0;
|
|
49
|
+
let gpuAvailable = true;
|
|
45
50
|
|
|
46
51
|
export async function captureProfile(delayToNextCapture = (capturePeriod ?? 60) * 1000): Promise<void> {
|
|
47
52
|
clearTimeout(profilerTimer);
|
|
@@ -58,6 +63,8 @@ export async function captureProfile(delayToNextCapture = (capturePeriod ?? 60)
|
|
|
58
63
|
const samplesByLocationId = new Map<number, number>();
|
|
59
64
|
let totalUserCount = 0;
|
|
60
65
|
let totalHarperCount = 0;
|
|
66
|
+
// Start GPU measurement early so it runs in parallel with CPU profiling work
|
|
67
|
+
const gpuPromise = getWorkerIndex() === 0 && gpuAvailable ? getGpuUtilization() : null;
|
|
61
68
|
try {
|
|
62
69
|
const profile = timeProfiler.stop(true);
|
|
63
70
|
const strings = profile.stringTable.strings;
|
|
@@ -89,6 +96,11 @@ export async function captureProfile(delayToNextCapture = (capturePeriod ?? 60)
|
|
|
89
96
|
recordAction(childCpuTimeInInterval, 'cpu-usage', 'user', 'child-processes');
|
|
90
97
|
lastChildCpuTime = childCpuTime;
|
|
91
98
|
}
|
|
99
|
+
// Record GPU utilization for this process and child processes
|
|
100
|
+
const gpuSeconds = await gpuPromise;
|
|
101
|
+
if (gpuSeconds !== null) {
|
|
102
|
+
recordAction(gpuSeconds, 'gpu-usage', 'user');
|
|
103
|
+
}
|
|
92
104
|
}
|
|
93
105
|
} catch (error) {
|
|
94
106
|
log.error?.('analytics profiler error:', error);
|
|
@@ -177,6 +189,41 @@ function getChildProcessCpuTime(): number | null {
|
|
|
177
189
|
}
|
|
178
190
|
}
|
|
179
191
|
|
|
192
|
+
/**
|
|
193
|
+
* Get the total SM (shader/streaming multiprocessor) utilization percentage across all GPUs
|
|
194
|
+
* for this process and all descendant processes.
|
|
195
|
+
* Uses nvidia-smi pmon for per-process GPU utilization.
|
|
196
|
+
* Only works on Linux with NVIDIA GPUs. Returns null if unavailable.
|
|
197
|
+
*/
|
|
198
|
+
async function getGpuUtilization(): Promise<number | null> {
|
|
199
|
+
try {
|
|
200
|
+
const currentPid = process.pid;
|
|
201
|
+
const descendants = findAllDescendants(currentPid);
|
|
202
|
+
const pidsToMonitor = new Set([currentPid, ...descendants]);
|
|
203
|
+
|
|
204
|
+
const { stdout } = await execFileAsync('nvidia-smi', ['pmon', '-c', '1', '-s', 'u']);
|
|
205
|
+
|
|
206
|
+
let totalSmPercent = 0;
|
|
207
|
+
for (const line of stdout.split('\n')) {
|
|
208
|
+
if (line.startsWith('#') || !line.trim()) continue;
|
|
209
|
+
const parts = line.trim().split(/\s+/);
|
|
210
|
+
// pmon -s u format: gpu pid type fb sm enc dec jpg ofa command
|
|
211
|
+
if (parts.length < 5) continue;
|
|
212
|
+
const pid = parseInt(parts[1], 10);
|
|
213
|
+
if (isNaN(pid) || !pidsToMonitor.has(pid)) continue;
|
|
214
|
+
const sm = parseInt(parts[4], 10); // SM utilization %
|
|
215
|
+
if (!isNaN(sm)) totalSmPercent += sm;
|
|
216
|
+
}
|
|
217
|
+
|
|
218
|
+
// Convert SM utilization % to GPU-seconds over the capture period
|
|
219
|
+
// e.g. 50% utilization over 60s = 30 GPU-seconds
|
|
220
|
+
return (totalSmPercent / 100) * (capturePeriod / 1000);
|
|
221
|
+
} catch {
|
|
222
|
+
gpuAvailable = false;
|
|
223
|
+
return null;
|
|
224
|
+
}
|
|
225
|
+
}
|
|
226
|
+
|
|
180
227
|
/**
|
|
181
228
|
* Recursively find all descendant PIDs of the given parent PID.
|
|
182
229
|
*/
|
|
@@ -172,15 +172,30 @@ export async function extractApplication(application: Application) {
|
|
|
172
172
|
}
|
|
173
173
|
}
|
|
174
174
|
} else {
|
|
175
|
-
//
|
|
176
|
-
const {
|
|
177
|
-
|
|
178
|
-
|
|
179
|
-
|
|
180
|
-
|
|
181
|
-
|
|
182
|
-
|
|
183
|
-
|
|
175
|
+
// `npm pack --json` writes a JSON array describing the packed tarball(s).
|
|
176
|
+
const { stdout, code, stderr } = await nonInteractiveSpawn(
|
|
177
|
+
application.name,
|
|
178
|
+
'npm',
|
|
179
|
+
['pack', '--json', application.packageIdentifier],
|
|
180
|
+
parentDirPath
|
|
181
|
+
);
|
|
182
|
+
if (code !== 0) {
|
|
183
|
+
throw new Error(`Failed to download package ${application.packageIdentifier}: ${stderr}`);
|
|
184
|
+
}
|
|
185
|
+
|
|
186
|
+
let packResult: Array<{ filename: string }>;
|
|
187
|
+
try {
|
|
188
|
+
packResult = JSON.parse(stdout.slice(stdout.indexOf('[')));
|
|
189
|
+
} catch (err) {
|
|
190
|
+
throw new Error(
|
|
191
|
+
`Failed to parse npm pack output for ${application.packageIdentifier}: ${err.message}\nstdout: ${stdout}`
|
|
192
|
+
);
|
|
193
|
+
}
|
|
194
|
+
if (!Array.isArray(packResult) || typeof packResult[0]?.filename !== 'string') {
|
|
195
|
+
throw new Error(`Unexpected npm pack output for ${application.packageIdentifier}:\n${stdout}`);
|
|
196
|
+
}
|
|
197
|
+
|
|
198
|
+
tarballPath = join(parentDirPath, packResult[0].filename);
|
|
184
199
|
tarball = createReadStream(tarballPath);
|
|
185
200
|
}
|
|
186
201
|
}
|
|
@@ -21,10 +21,10 @@ export class ApplicationScope {
|
|
|
21
21
|
server: Server;
|
|
22
22
|
mode?: 'native' | 'vm' | 'vm-current-context' | 'compartment'; // option to set this from the scope
|
|
23
23
|
dependencyLoader?: 'native' | 'app' | 'auto'; // option to set this from the scope
|
|
24
|
-
|
|
24
|
+
allowedPath?: string;
|
|
25
25
|
config: any;
|
|
26
26
|
moduleCache: any; // used by the loader to retain a cache of modules, type is an internal detail of the loader
|
|
27
|
-
constructor(name: string, resources: Resources, server: Server, isInternal = false
|
|
27
|
+
constructor(name: string, resources: Resources, server: Server, isInternal = false) {
|
|
28
28
|
this.logger = forComponent(name, !isInternal);
|
|
29
29
|
|
|
30
30
|
this.resources = resources;
|
|
@@ -32,7 +32,6 @@ export class ApplicationScope {
|
|
|
32
32
|
|
|
33
33
|
this.mode = env.get(CONFIG_PARAMS.APPLICATIONS_MODULELOADER) ?? 'vm';
|
|
34
34
|
this.dependencyLoader = env.get(CONFIG_PARAMS.APPLICATIONS_DEPENDENCYLOADER);
|
|
35
|
-
this.verifyPath = verifyPath;
|
|
36
35
|
}
|
|
37
36
|
|
|
38
37
|
/**
|
|
@@ -274,7 +274,7 @@ export async function loadComponent(
|
|
|
274
274
|
autoReload,
|
|
275
275
|
appName,
|
|
276
276
|
} = options;
|
|
277
|
-
applicationScope.
|
|
277
|
+
applicationScope.allowedPath ??= realpathSync(componentDirectory);
|
|
278
278
|
if (providedLoadedComponents) loadedComponents = providedLoadedComponents;
|
|
279
279
|
try {
|
|
280
280
|
let config;
|
|
@@ -326,8 +326,13 @@ export async function loadComponent(
|
|
|
326
326
|
|
|
327
327
|
let extensionModule: any;
|
|
328
328
|
const pkg = componentConfig.package;
|
|
329
|
+
const loadComponentOption = componentConfig.loadComponent ?? 'always';
|
|
329
330
|
try {
|
|
330
331
|
if (pkg) {
|
|
332
|
+
if (loadComponentOption === 'dev-only' && !process.env.DEV_MODE) {
|
|
333
|
+
componentLifecycle.loaded(componentStatusName, `Component '${componentStatusName}' skipped (dev-only)`);
|
|
334
|
+
continue;
|
|
335
|
+
}
|
|
331
336
|
let componentPath: string | null = null;
|
|
332
337
|
if (isRoot) {
|
|
333
338
|
componentPath = join(componentDirectory, 'components', componentName);
|
|
@@ -344,7 +349,7 @@ export async function loadComponent(
|
|
|
344
349
|
}
|
|
345
350
|
}
|
|
346
351
|
if (componentPath) {
|
|
347
|
-
subApplicationScope.
|
|
352
|
+
subApplicationScope.allowedPath ??= realpathSync(componentPath);
|
|
348
353
|
if (!process.env.HARPER_SAFE_MODE) {
|
|
349
354
|
extensionModule = await loadComponent(componentPath, resources, origin, {
|
|
350
355
|
isRoot: false,
|
|
@@ -354,6 +359,12 @@ export async function loadComponent(
|
|
|
354
359
|
});
|
|
355
360
|
componentFunctionality[componentName] = true;
|
|
356
361
|
}
|
|
362
|
+
} else if (loadComponentOption === 'if-installed') {
|
|
363
|
+
componentLifecycle.loaded(
|
|
364
|
+
componentStatusName,
|
|
365
|
+
`Component '${componentStatusName}' skipped (not installed)`
|
|
366
|
+
);
|
|
367
|
+
continue;
|
|
357
368
|
} else {
|
|
358
369
|
throw new Error(`Unable to find package ${componentName}:${pkg}`);
|
|
359
370
|
}
|
|
@@ -15,6 +15,7 @@ import type { Logger } from '../utility/logging/logger.ts';
|
|
|
15
15
|
import * as fs from 'fs-extra';
|
|
16
16
|
import * as path from 'node:path';
|
|
17
17
|
import * as crypto from 'node:crypto';
|
|
18
|
+
import { cloneDeep } from 'lodash';
|
|
18
19
|
import { getBackupDirPath } from './configHelpers.ts';
|
|
19
20
|
|
|
20
21
|
const STATE_FILE_NAME = '.harper-config-state.json';
|
|
@@ -590,6 +591,39 @@ function cleanupRemovedEnvVar(
|
|
|
590
591
|
logger.debug?.(`${envVarName} removed, cleaned up values`);
|
|
591
592
|
}
|
|
592
593
|
|
|
594
|
+
/**
|
|
595
|
+
* Compose a merged config from HARPER_DEFAULT_CONFIG and HARPER_SET_CONFIG
|
|
596
|
+
* layered with an optional base. Later layers win:
|
|
597
|
+
* HARPER_DEFAULT_CONFIG < base < HARPER_SET_CONFIG
|
|
598
|
+
*
|
|
599
|
+
* HARPER_DEFAULT_CONFIG provides scaffolding defaults, the base (e.g., the
|
|
600
|
+
* user's existing config file) is layered on top, and HARPER_SET_CONFIG
|
|
601
|
+
* force-overrides everything. This matches the precedence applied by the
|
|
602
|
+
* runtime pipeline in applyRuntimeEnvConfig.
|
|
603
|
+
*
|
|
604
|
+
* Unlike applyRuntimeEnvConfig, this does NOT read or write the config state
|
|
605
|
+
* file and does NOT track sources — it returns a fresh object. Use when you
|
|
606
|
+
* need the effective value of a config key before the state/file wiring is in
|
|
607
|
+
* place (e.g., during clone / pre-install).
|
|
608
|
+
*/
|
|
609
|
+
export function composeConfigFromEnv(base: ConfigObject = {}): ConfigObject {
|
|
610
|
+
const result: ConfigObject = {};
|
|
611
|
+
const layers: (ConfigObject | null)[] = [
|
|
612
|
+
parseConfigEnvVar(process.env.HARPER_DEFAULT_CONFIG, 'HARPER_DEFAULT_CONFIG'),
|
|
613
|
+
cloneDeep(base),
|
|
614
|
+
parseConfigEnvVar(process.env.HARPER_SET_CONFIG, 'HARPER_SET_CONFIG'),
|
|
615
|
+
];
|
|
616
|
+
|
|
617
|
+
for (const layer of layers) {
|
|
618
|
+
if (!layer) continue;
|
|
619
|
+
for (const [p, value] of Object.entries(flattenObject(layer))) {
|
|
620
|
+
setNestedValue(result, p, value);
|
|
621
|
+
}
|
|
622
|
+
}
|
|
623
|
+
|
|
624
|
+
return result;
|
|
625
|
+
}
|
|
626
|
+
|
|
593
627
|
/**
|
|
594
628
|
* Apply HARPER_DEFAULT_CONFIG and HARPER_SET_CONFIG
|
|
595
629
|
* Can be used for both install-time and runtime
|
package/core/resources/Table.ts
CHANGED
|
@@ -2262,7 +2262,9 @@ export function makeTable(options) {
|
|
|
2262
2262
|
return (entryA, entryB) => {
|
|
2263
2263
|
const a = getAttributeValue(entryA, order.attribute, context);
|
|
2264
2264
|
const b = getAttributeValue(entryB, order.attribute, context);
|
|
2265
|
-
const diff = descending
|
|
2265
|
+
const diff = descending
|
|
2266
|
+
? compareKeys(convertToComparableKeys(b), convertToComparableKeys(a))
|
|
2267
|
+
: compareKeys(convertToComparableKeys(a), convertToComparableKeys(b));
|
|
2266
2268
|
if (diff === 0) return nextComparator?.(entryA, entryB) || 0;
|
|
2267
2269
|
return diff;
|
|
2268
2270
|
};
|
|
@@ -4572,3 +4574,12 @@ function hasOtherProcesses(store) {
|
|
|
4572
4574
|
return +line.match(/\d+/)?.[0] != pid;
|
|
4573
4575
|
});
|
|
4574
4576
|
}
|
|
4577
|
+
function convertToComparableKeys(a) {
|
|
4578
|
+
if (a instanceof Date) {
|
|
4579
|
+
return a.getTime();
|
|
4580
|
+
}
|
|
4581
|
+
if (Array.isArray(a)) {
|
|
4582
|
+
return a.map(convertToComparableKeys);
|
|
4583
|
+
}
|
|
4584
|
+
return a;
|
|
4585
|
+
}
|
|
@@ -219,15 +219,12 @@ export async function recordHostname() {
|
|
|
219
219
|
const nodeId = stableNodeId(hostname);
|
|
220
220
|
log.trace?.('recordHostname nodeId:', nodeId);
|
|
221
221
|
const hostnamesTable = getAnalyticsHostnameTable();
|
|
222
|
-
const
|
|
223
|
-
|
|
224
|
-
|
|
225
|
-
|
|
226
|
-
|
|
227
|
-
|
|
228
|
-
log.trace?.(`recordHostname storing hostname: ${JSON.stringify(hostnameRecord)}`);
|
|
229
|
-
await hostnamesTable.put(hostnameRecord.id, hostnameRecord);
|
|
230
|
-
}
|
|
222
|
+
const hostnameRecord = {
|
|
223
|
+
id: nodeId,
|
|
224
|
+
hostname,
|
|
225
|
+
};
|
|
226
|
+
log.trace?.(`recordHostname storing hostname: ${JSON.stringify(hostnameRecord)}`);
|
|
227
|
+
await hostnamesTable.put(hostnameRecord.id, hostnameRecord);
|
|
231
228
|
}
|
|
232
229
|
|
|
233
230
|
export interface Metric {
|
|
@@ -514,7 +511,7 @@ async function aggregation(fromPeriod, toPeriod = 60000) {
|
|
|
514
511
|
await rest();
|
|
515
512
|
}
|
|
516
513
|
for (const entry of threadsToAverage) {
|
|
517
|
-
// eslint-disable-next-line @typescript-eslint/no-unused-vars
|
|
514
|
+
// eslint-disable-next-line @typescript-eslint/no-unused-vars
|
|
518
515
|
let { path, method, type, metric, count, total, distribution, threads, ...measures } = entry;
|
|
519
516
|
threads = threads.filter((thread) => thread);
|
|
520
517
|
for (const measureName in measures) {
|
|
@@ -130,8 +130,8 @@ export class HierarchicalNavigableSmallWorld {
|
|
|
130
130
|
// Generate random level for this new element
|
|
131
131
|
const level = oldNode.level ?? Math.min(Math.floor(-Math.log(Math.random()) * this.mL), MAX_LEVEL);
|
|
132
132
|
let currentLevel = entryPoint.level;
|
|
133
|
-
if (level
|
|
134
|
-
// if we are at
|
|
133
|
+
if (level > currentLevel) {
|
|
134
|
+
// if we are at a higher level, make this the new entry point
|
|
135
135
|
if (typeof nodeId !== 'number') {
|
|
136
136
|
throw new Error('Invalid nodeId: ' + nodeId);
|
|
137
137
|
}
|
|
@@ -232,6 +232,19 @@ export class HierarchicalNavigableSmallWorld {
|
|
|
232
232
|
oldNode[l] = oldConnections;
|
|
233
233
|
}
|
|
234
234
|
oldConnections.splice(oldPosition, 1);
|
|
235
|
+
// update the distance in the reverse connection if the vector changed
|
|
236
|
+
if (oldConnection.distance !== distance) {
|
|
237
|
+
const neighborNode = updateNode(id, node);
|
|
238
|
+
if (neighborNode[l]) {
|
|
239
|
+
if (Object.isFrozen(neighborNode[l])) {
|
|
240
|
+
neighborNode[l] = neighborNode[l].slice();
|
|
241
|
+
}
|
|
242
|
+
const reverseIdx = neighborNode[l].findIndex(({ id: nid }) => nid === nodeId);
|
|
243
|
+
if (reverseIdx >= 0) {
|
|
244
|
+
neighborNode[l][reverseIdx] = { id: nodeId, distance };
|
|
245
|
+
}
|
|
246
|
+
}
|
|
247
|
+
}
|
|
235
248
|
} else {
|
|
236
249
|
// add new connection since this is truly a new connection now
|
|
237
250
|
this.addConnection(id, updateNode(id, node), nodeId, l, distance, updateNode, options);
|
|
@@ -360,7 +373,7 @@ export class HierarchicalNavigableSmallWorld {
|
|
|
360
373
|
const candidates = [
|
|
361
374
|
{
|
|
362
375
|
id: entryPointId,
|
|
363
|
-
distance:
|
|
376
|
+
distance: distanceFunction(queryVector, entryPoint.vector),
|
|
364
377
|
node: entryPoint,
|
|
365
378
|
},
|
|
366
379
|
];
|
|
@@ -531,10 +544,13 @@ export class HierarchicalNavigableSmallWorld {
|
|
|
531
544
|
if (removedNode) {
|
|
532
545
|
// Remove the reverse connection if it exists
|
|
533
546
|
if (removedNode[level]) {
|
|
534
|
-
|
|
535
|
-
|
|
536
|
-
|
|
537
|
-
logger.info?.('
|
|
547
|
+
const filtered = removedNode[level].filter(({ id }) => id !== fromId);
|
|
548
|
+
if (level === 0 && filtered.length === 0) {
|
|
549
|
+
// don't remove the last connection at level 0 — it would orphan this node
|
|
550
|
+
logger.info?.('skipping removal of last connection', fromId, toId);
|
|
551
|
+
} else {
|
|
552
|
+
removedNode = updateNode(removed.id, removedNode);
|
|
553
|
+
removedNode[level] = filtered;
|
|
538
554
|
}
|
|
539
555
|
}
|
|
540
556
|
}
|
|
@@ -14,7 +14,16 @@ import * as child_process from 'node:child_process';
|
|
|
14
14
|
import { CONFIG_PARAMS } from '../utility/hdbTerms.ts';
|
|
15
15
|
import { contentTypes } from '../server/serverHelpers/contentTypes.ts';
|
|
16
16
|
import type { CompartmentOptions } from 'ses';
|
|
17
|
-
import {
|
|
17
|
+
import {
|
|
18
|
+
mkdirSync,
|
|
19
|
+
readFileSync,
|
|
20
|
+
writeFileSync,
|
|
21
|
+
unlinkSync,
|
|
22
|
+
openSync,
|
|
23
|
+
closeSync,
|
|
24
|
+
statSync,
|
|
25
|
+
realpathSync,
|
|
26
|
+
} from 'node:fs';
|
|
18
27
|
import { join } from 'node:path';
|
|
19
28
|
import { EventEmitter } from 'node:events';
|
|
20
29
|
import { whenComponentsLoaded } from '../server/threads/threadServer.js';
|
|
@@ -495,13 +504,13 @@ async function loadModuleWithVM(moduleUrl: string, scope: ApplicationScope, useC
|
|
|
495
504
|
}
|
|
496
505
|
|
|
497
506
|
if (url.startsWith('file://') && usePrivateGlobal) {
|
|
498
|
-
checkAllowedModulePath(url, scope.
|
|
507
|
+
checkAllowedModulePath(url, scope.allowedPath);
|
|
499
508
|
const source = readFileSync(new URL(url), { encoding: 'utf-8' });
|
|
500
509
|
return createModuleFromSource(url, source, usePrivateGlobal);
|
|
501
510
|
}
|
|
502
511
|
|
|
503
512
|
// For Node.js built-in modules (node:) and npm packages without application loader for dependency
|
|
504
|
-
const replacedModule = checkAllowedModulePath(url, scope.
|
|
513
|
+
const replacedModule = checkAllowedModulePath(url, scope.allowedPath);
|
|
505
514
|
if (replacedModule) {
|
|
506
515
|
return createSyntheticModule(url, normalizeImportedModule(replacedModule));
|
|
507
516
|
}
|
|
@@ -570,7 +579,7 @@ async function getCompartment(scope: ApplicationScope, globals) {
|
|
|
570
579
|
}
|
|
571
580
|
return new StaticModuleRecord(moduleText, moduleSpecifier);
|
|
572
581
|
} else {
|
|
573
|
-
checkAllowedModulePath(moduleSpecifier, scope.
|
|
582
|
+
checkAllowedModulePath(moduleSpecifier, scope.allowedPath);
|
|
574
583
|
const moduleExports = await import(moduleSpecifier);
|
|
575
584
|
return {
|
|
576
585
|
imports: [],
|
|
@@ -761,23 +770,54 @@ function isProcessRunning(pid: number): boolean {
|
|
|
761
770
|
* Acquires an exclusive lock using the PID file itself (synchronously with busy-wait)
|
|
762
771
|
* Returns 0 if lock was acquired (need to spawn new process), or the existing PID if process is running
|
|
763
772
|
*/
|
|
764
|
-
function
|
|
773
|
+
function parsePidFile(content: string): { pid: number; version: number } {
|
|
774
|
+
const lines = content.trim().split('\n');
|
|
775
|
+
const pid = Number.parseInt(lines[0], 10);
|
|
776
|
+
const version = lines.length > 1 ? parseInt(lines[1], 10) : 0;
|
|
777
|
+
return { pid, version };
|
|
778
|
+
}
|
|
779
|
+
|
|
780
|
+
function acquirePidFileLock(
|
|
781
|
+
pidFilePath: string,
|
|
782
|
+
requestedVersion?: number,
|
|
783
|
+
maxRetries = 100,
|
|
784
|
+
retryDelay = 5
|
|
785
|
+
): { pid: number; version: number } {
|
|
765
786
|
for (let attempt = 0; attempt < maxRetries; attempt++) {
|
|
766
787
|
try {
|
|
767
788
|
// Try to open exclusively - 'wx' fails if file exists
|
|
768
789
|
const fd = openSync(pidFilePath, 'wx');
|
|
769
790
|
closeSync(fd);
|
|
770
|
-
return 0; // Successfully acquired lock (file created), caller should spawn process
|
|
791
|
+
return { pid: 0, version: 0 }; // Successfully acquired lock (file created), caller should spawn process
|
|
771
792
|
} catch (err) {
|
|
772
793
|
if (err.code === 'EEXIST') {
|
|
773
794
|
// File exists - check if it contains a valid running process
|
|
774
795
|
try {
|
|
775
796
|
const pidContent = readFileSync(pidFilePath, 'utf-8');
|
|
776
|
-
const existingPid =
|
|
797
|
+
const { pid: existingPid, version: existingVersion } = parsePidFile(pidContent);
|
|
777
798
|
|
|
778
799
|
if (!isNaN(existingPid) && isProcessRunning(existingPid)) {
|
|
779
|
-
//
|
|
780
|
-
|
|
800
|
+
// If the version isn't the one we want, kill the existing process and re-acquire
|
|
801
|
+
if (requestedVersion != null && requestedVersion !== existingVersion) {
|
|
802
|
+
try {
|
|
803
|
+
process.kill(existingPid);
|
|
804
|
+
} catch {
|
|
805
|
+
// Process may have already exited
|
|
806
|
+
}
|
|
807
|
+
try {
|
|
808
|
+
unlinkSync(pidFilePath);
|
|
809
|
+
} catch {
|
|
810
|
+
// Another thread may have removed it
|
|
811
|
+
}
|
|
812
|
+
// Retry to acquire the lock for the new version
|
|
813
|
+
const start = Date.now();
|
|
814
|
+
while (Date.now() - start < retryDelay) {
|
|
815
|
+
// Busy wait for process cleanup
|
|
816
|
+
}
|
|
817
|
+
continue;
|
|
818
|
+
}
|
|
819
|
+
// Valid process is running at same or higher version, return its PID
|
|
820
|
+
return { pid: existingPid, version: existingVersion };
|
|
781
821
|
}
|
|
782
822
|
|
|
783
823
|
// Invalid/empty PID - check file age to determine if it's stale or being written
|
|
@@ -824,6 +864,7 @@ function createSpawn(spawnFunction: (...args: any) => child_process.ChildProcess
|
|
|
824
864
|
throw new Error(
|
|
825
865
|
`Calling ${spawnFunction.name} in Harper must have a process "name" in the options to ensure that a single process is started and reused`
|
|
826
866
|
);
|
|
867
|
+
const requestedVersion = options?.version;
|
|
827
868
|
|
|
828
869
|
// Ensure PID directory exists
|
|
829
870
|
const pidDir = join(basePath, 'pids');
|
|
@@ -831,20 +872,22 @@ function createSpawn(spawnFunction: (...args: any) => child_process.ChildProcess
|
|
|
831
872
|
|
|
832
873
|
const pidFilePath = join(pidDir, `${processName}.pid`);
|
|
833
874
|
|
|
834
|
-
// Try to acquire lock - returns 0 if acquired, or existing PID
|
|
835
|
-
const
|
|
875
|
+
// Try to acquire lock - returns pid: 0 if acquired, or existing PID/version
|
|
876
|
+
const existing = acquirePidFileLock(pidFilePath, requestedVersion);
|
|
836
877
|
|
|
837
|
-
if (
|
|
878
|
+
if (existing.pid !== 0) {
|
|
838
879
|
// Existing process is running, return wrapper
|
|
839
|
-
return new ExistingProcessWrapper(
|
|
880
|
+
return new ExistingProcessWrapper(existing.pid);
|
|
840
881
|
}
|
|
841
882
|
|
|
842
883
|
// We acquired the lock (file was created), spawn new process
|
|
843
884
|
const childProcess = spawnFunction(command, args, options, callback);
|
|
844
885
|
|
|
845
|
-
// Write PID to the file we just created
|
|
886
|
+
// Write PID (and version if provided) to the file we just created
|
|
887
|
+
const pidFileContent =
|
|
888
|
+
requestedVersion != null ? `${childProcess.pid}\n${requestedVersion}` : childProcess.pid.toString();
|
|
846
889
|
try {
|
|
847
|
-
writeFileSync(pidFilePath,
|
|
890
|
+
writeFileSync(pidFilePath, pidFileContent, 'utf-8');
|
|
848
891
|
} catch (err) {
|
|
849
892
|
// Failed to write PID, clean up
|
|
850
893
|
try {
|
|
@@ -869,21 +912,24 @@ function createSpawn(spawnFunction: (...args: any) => child_process.ChildProcess
|
|
|
869
912
|
|
|
870
913
|
/**
|
|
871
914
|
* Validates whether a module can be loaded based on security restrictions and returns the module path or replacement.
|
|
872
|
-
* For file URLs, ensures the module is within the
|
|
915
|
+
* For file URLs, ensures the module is within the allowed path.
|
|
873
916
|
* For node built-in modules, checks against an allowlist and returns any replacements.
|
|
874
917
|
*
|
|
875
918
|
* @param {string} moduleUrl - The URL or identifier of the module to be loaded, which may be a file: URL, node: URL, or bare module specifier.
|
|
876
|
-
* @param {string}
|
|
919
|
+
* @param {string} allowedPath - The absolute path that the module is allowed to load from.
|
|
877
920
|
* @return {any} Returns undefined for allowed file paths, or a replacement module identifier for allowed node built-in modules.
|
|
878
|
-
* @throws {Error} Throws an error if the module is outside the
|
|
921
|
+
* @throws {Error} Throws an error if the module is outside the allowed path or if the module is not in the allowed list.
|
|
879
922
|
*/
|
|
880
|
-
function checkAllowedModulePath(moduleUrl: string,
|
|
923
|
+
function checkAllowedModulePath(moduleUrl: string, allowedPath?: string): boolean {
|
|
881
924
|
if (moduleUrl.startsWith('file:')) {
|
|
882
|
-
|
|
883
|
-
|
|
925
|
+
let path = moduleUrl.slice(7);
|
|
926
|
+
try {
|
|
927
|
+
path = realpathSync(path);
|
|
928
|
+
} catch {}
|
|
929
|
+
if (!allowedPath || path.startsWith(allowedPath)) {
|
|
884
930
|
return;
|
|
885
931
|
}
|
|
886
|
-
throw new Error(`Can not load module outside of
|
|
932
|
+
throw new Error(`Can not load module outside of allowed path`);
|
|
887
933
|
}
|
|
888
934
|
let simpleName = moduleUrl.startsWith('node:') ? moduleUrl.slice(5) : moduleUrl;
|
|
889
935
|
simpleName = simpleName.split('/')[0];
|
package/core/security/user.ts
CHANGED
|
@@ -258,14 +258,16 @@ async function userInfo(body): Promise<string | User> {
|
|
|
258
258
|
}
|
|
259
259
|
|
|
260
260
|
let user = _.cloneDeep(body.hdb_user);
|
|
261
|
-
let roleData =
|
|
262
|
-
|
|
263
|
-
|
|
264
|
-
|
|
265
|
-
|
|
266
|
-
|
|
261
|
+
let roleData =
|
|
262
|
+
user.role &&
|
|
263
|
+
(await search.searchByHash({
|
|
264
|
+
schema: 'system',
|
|
265
|
+
table: 'hdb_role',
|
|
266
|
+
hash_values: [user.role.id],
|
|
267
|
+
get_attributes: ['*'],
|
|
268
|
+
}));
|
|
267
269
|
|
|
268
|
-
user.role = roleData[0];
|
|
270
|
+
user.role = roleData?.[0];
|
|
269
271
|
delete user.password;
|
|
270
272
|
delete user.refresh_token;
|
|
271
273
|
delete user.hash;
|
|
@@ -429,7 +431,7 @@ async function getSuperUser(): Promise<User | undefined> {
|
|
|
429
431
|
await setUsersWithRolesCache();
|
|
430
432
|
}
|
|
431
433
|
for (let [, user] of usersWithRolesMap) {
|
|
432
|
-
if (user.role
|
|
434
|
+
if (user.role?.role === 'super_user') return user;
|
|
433
435
|
}
|
|
434
436
|
}
|
|
435
437
|
|
|
@@ -44,6 +44,9 @@ const write_ts_1 = require("../core/resources/analytics/write.js");
|
|
|
44
44
|
const environmentManager_js_1 = require("../core/utility/environment/environmentManager.js");
|
|
45
45
|
const packageUtils_js_1 = require("../core/utility/packageUtils.js");
|
|
46
46
|
const node_fs_1 = require("node:fs");
|
|
47
|
+
const node_child_process_1 = require("node:child_process");
|
|
48
|
+
const node_util_1 = require("node:util");
|
|
49
|
+
const execFileAsync = (0, node_util_1.promisify)(node_child_process_1.execFile);
|
|
47
50
|
const pprof_1 = require("@datadog/pprof");
|
|
48
51
|
const manageThreads_js_1 = require("../core/server/threads/manageThreads.js");
|
|
49
52
|
const log = __importStar(require("../core/utility/logging/harper_logger.js"));
|
|
@@ -77,6 +80,7 @@ function handleApplication({ options }) {
|
|
|
77
80
|
}, 1000); // wait for everything to load before we start the profiler
|
|
78
81
|
}
|
|
79
82
|
let lastChildCpuTime = 0;
|
|
83
|
+
let gpuAvailable = true;
|
|
80
84
|
async function captureProfile(delayToNextCapture = (capturePeriod ?? 60) * 1000) {
|
|
81
85
|
clearTimeout(profilerTimer);
|
|
82
86
|
if (!profilerStarted) {
|
|
@@ -92,6 +96,8 @@ async function captureProfile(delayToNextCapture = (capturePeriod ?? 60) * 1000)
|
|
|
92
96
|
const samplesByLocationId = new Map();
|
|
93
97
|
let totalUserCount = 0;
|
|
94
98
|
let totalHarperCount = 0;
|
|
99
|
+
// Start GPU measurement early so it runs in parallel with CPU profiling work
|
|
100
|
+
const gpuPromise = (0, manageThreads_js_1.getWorkerIndex)() === 0 && gpuAvailable ? getGpuUtilization() : null;
|
|
95
101
|
try {
|
|
96
102
|
const profile = pprof_1.time.stop(true);
|
|
97
103
|
const strings = profile.stringTable.strings;
|
|
@@ -122,6 +128,11 @@ async function captureProfile(delayToNextCapture = (capturePeriod ?? 60) * 1000)
|
|
|
122
128
|
(0, write_ts_1.recordAction)(childCpuTimeInInterval, 'cpu-usage', 'user', 'child-processes');
|
|
123
129
|
lastChildCpuTime = childCpuTime;
|
|
124
130
|
}
|
|
131
|
+
// Record GPU utilization for this process and child processes
|
|
132
|
+
const gpuSeconds = await gpuPromise;
|
|
133
|
+
if (gpuSeconds !== null) {
|
|
134
|
+
(0, write_ts_1.recordAction)(gpuSeconds, 'gpu-usage', 'user');
|
|
135
|
+
}
|
|
125
136
|
}
|
|
126
137
|
}
|
|
127
138
|
catch (error) {
|
|
@@ -211,6 +222,42 @@ function getChildProcessCpuTime() {
|
|
|
211
222
|
return null;
|
|
212
223
|
}
|
|
213
224
|
}
|
|
225
|
+
/**
|
|
226
|
+
* Get the total SM (shader/streaming multiprocessor) utilization percentage across all GPUs
|
|
227
|
+
* for this process and all descendant processes.
|
|
228
|
+
* Uses nvidia-smi pmon for per-process GPU utilization.
|
|
229
|
+
* Only works on Linux with NVIDIA GPUs. Returns null if unavailable.
|
|
230
|
+
*/
|
|
231
|
+
async function getGpuUtilization() {
|
|
232
|
+
try {
|
|
233
|
+
const currentPid = process.pid;
|
|
234
|
+
const descendants = findAllDescendants(currentPid);
|
|
235
|
+
const pidsToMonitor = new Set([currentPid, ...descendants]);
|
|
236
|
+
const { stdout } = await execFileAsync('nvidia-smi', ['pmon', '-c', '1', '-s', 'u']);
|
|
237
|
+
let totalSmPercent = 0;
|
|
238
|
+
for (const line of stdout.split('\n')) {
|
|
239
|
+
if (line.startsWith('#') || !line.trim())
|
|
240
|
+
continue;
|
|
241
|
+
const parts = line.trim().split(/\s+/);
|
|
242
|
+
// pmon -s u format: gpu pid type fb sm enc dec jpg ofa command
|
|
243
|
+
if (parts.length < 5)
|
|
244
|
+
continue;
|
|
245
|
+
const pid = parseInt(parts[1], 10);
|
|
246
|
+
if (isNaN(pid) || !pidsToMonitor.has(pid))
|
|
247
|
+
continue;
|
|
248
|
+
const sm = parseInt(parts[4], 10); // SM utilization %
|
|
249
|
+
if (!isNaN(sm))
|
|
250
|
+
totalSmPercent += sm;
|
|
251
|
+
}
|
|
252
|
+
// Convert SM utilization % to GPU-seconds over the capture period
|
|
253
|
+
// e.g. 50% utilization over 60s = 30 GPU-seconds
|
|
254
|
+
return (totalSmPercent / 100) * (capturePeriod / 1000);
|
|
255
|
+
}
|
|
256
|
+
catch {
|
|
257
|
+
gpuAvailable = false;
|
|
258
|
+
return null;
|
|
259
|
+
}
|
|
260
|
+
}
|
|
214
261
|
/**
|
|
215
262
|
* Recursively find all descendant PIDs of the given parent PID.
|
|
216
263
|
*/
|