Files
twenty/packages/twenty-server/test/integration/metadata/suites/application/logic-function-install-performance.integration-spec.ts
T
Charles Bochet fcaf2b4d9b chore(twenty-server): temporary instrumentation for app-install 504 (#21365)
## Why

App installs on cloud intermittently fail with a 504, surfacing in
Sentry as `Migration action 'update' for 'logicFunction' failed` +
`Failed to rollback transaction: Query runner already released`. This is
**temporary instrumentation** to pin down where the time goes — to be
reverted once the bottleneck is fixed. Everything is greppable via
`[install-perf]` and marked `// TODO(install-perf)`.

## What the local repro already told us

I instrumented the manifest-sync/migration path and ran a local harness
(new skipped spec) installing **1 / 8 / 30 logic functions**, for both
create and the checksum-bump **update** (the incident path):

| stage (N=30, update) | ms |
|---|---|
| flat-maps recompute | ~1 |
| build migration | ~11 |
| transaction (all actions + commit) | ~79 |
| post-commit cache invalidate | ~6 |
| **full sync** | **~135** |

Nothing approached 1s, let alone 10s; no slow queries logged. So the
migration/cache code is **not** the algorithmic cause. Given the
in-transaction `UPDATE ... WHERE id=?` is intrinsically fast, a >10s in
prod almost certainly means it was **blocked on a lock**, and the 10s
node-pg `query_timeout` (`core.datasource.ts`) then killed the
connection → the observed errors + 504. Local can't reproduce prod lock
contention / table sizes, hence this instrumentation.

## What this adds (all `TODO`-marked)

- **hrtime per-stage timing** — flat-maps recompute, build vs run,
per-action (`>50ms`), transaction summary, post-commit cache
invalidation. Uses `process.hrtime` because the integration harness
enables fake timers (so `Date.now()` is useless there).
- **`maxQueryExecutionTime`** slow-query logging on the core datasource
(logs the offending SQL).
- **Scoped `SET LOCAL lock_timeout = '8s'`** on the migration
transaction (below the 10s `query_timeout`) → a blocked action fails
fast with a clear *"canceling statement due to lock timeout"* instead of
the opaque connection kill.
- **Best-effort `pg_stat_activity` snapshot on failure** (on a fresh
pooled connection) to identify the blocking session, plus a **guarded
rollback** so a released connection stops masking the real error.
- **Skipped local perf harness**
(`logic-function-install-performance.integration-spec.ts`) — run
manually with `nx test:integration:with-db-reset -- --testPathPattern
"logic-function-install-performance"`.

## How we'll use it

Deploy, reproduce the failing install, and read the `[install-perf]`
logs: the per-action timing names the action, the `lock_timeout` message
+ `pg_stat_activity` snapshot name the **blocking** query/PID. Then
revert this PR and fix the actual contention.

Typecheck (`nx typecheck twenty-server`) is clean.
2026-06-09 15:57:25 +02:00

152 lines
4.8 KiB
TypeScript

import { buildBaseManifest } from 'test/integration/metadata/suites/application/utils/build-base-manifest.util';
import { cleanupApplicationAndAppRegistration } from 'test/integration/metadata/suites/application/utils/cleanup-application-and-app-registration.util';
import { setupApplicationForSync } from 'test/integration/metadata/suites/application/utils/setup-application-for-sync.util';
import { syncApplication } from 'test/integration/metadata/suites/application/utils/sync-application.util';
import {
type LogicFunctionManifest,
type Manifest,
} from 'twenty-shared/application';
import { v4 as uuidv4 } from 'uuid';
/**
* Performance harness for installing / updating many logic functions through
* application manifest sync. The goal is to find what could take >10s in prod
* (which trips the node-postgres `query_timeout` in core.datasource.ts and
* produces "Migration action 'update' for 'logicFunction' failed" + 504).
*
* IMPORTANT: the integration harness boots the NestJS app in-process with
* `fakeTimers.enableGlobally: true`. We call `jest.useRealTimers()` for the whole
* suite so timing (`performance.now()`) is real and cache-lock retry delays etc.
* do not hang.
*/
// Real timers for the whole suite — see note above.
jest.useRealTimers();
jest.setTimeout(120000);
const FN_COUNTS = [1, 8, 30];
// Stable universalIdentifiers across versions so the second sync exercises
// UPDATE (the actual incident), not CREATE+DELETE.
const buildManifest = ({
appId,
roleId,
universalIdentifiers,
checksumVersion,
}: {
appId: string;
roleId: string;
universalIdentifiers: string[];
checksumVersion: 'v1' | 'v2';
}): Manifest => {
const logicFunctions: LogicFunctionManifest[] = universalIdentifiers.map(
(universalIdentifier, i) => ({
universalIdentifier,
name: `PerfFn${i}`,
description: `Perf logic function ${i}`,
handlerName: 'handler',
sourceHandlerPath: `src/fn-${i}.ts`,
builtHandlerPath: `dist/fn-${i}.mjs`,
builtHandlerChecksum: `checksum-${i}-${checksumVersion}`,
httpRouteTriggerSettings: {
path: `/fn-${i}`,
httpMethod: 'GET',
isAuthRequired: true,
},
}),
);
return buildBaseManifest({
appId,
roleId,
overrides: { logicFunctions },
});
};
const timeSync = async (
label: string,
manifest: Manifest,
): Promise<number> => {
const start = performance.now();
await syncApplication({ manifest, expectToFail: false });
const ms = performance.now() - start;
// oxlint-disable-next-line no-console
console.log(`[install-perf][test] ${label} took ${ms.toFixed(1)}ms`);
return ms;
};
// TODO(install-perf): temporary manual perf harness, remove. Skipped in CI.
describe.skip('Logic function install performance', () => {
it.each(FN_COUNTS)(
'create + update sync with %i logic functions',
async (count) => {
const appId = uuidv4();
const roleId = uuidv4();
const universalIdentifiers = Array.from({ length: count }, () =>
uuidv4(),
);
await setupApplicationForSync({
applicationUniversalIdentifier: appId,
name: `Perf App ${count}`,
description: `Perf app with ${count} logic functions`,
sourcePath: `perf-app-${count}`,
});
jest.useRealTimers();
try {
// No built-handler file upload is needed: the migration create/update
// handlers never read the built file for LIVE functions (prebuilt
// install is skipped), and uploading N files would trip the file-upload
// rate limiter (30 per 30s). We only measure migration + cache cost.
// oxlint-disable-next-line no-console
console.log(
`[install-perf][test] ===== N=${count} : FIRST SYNC (create ${count} functions) =====`,
);
const createMs = await timeSync(
`N=${count} create sync`,
buildManifest({
appId,
roleId,
universalIdentifiers,
checksumVersion: 'v1',
}),
);
// oxlint-disable-next-line no-console
console.log(
`[install-perf][test] ===== N=${count} : SECOND SYNC (update ${count} functions, checksum v2) =====`,
);
const updateMs = await timeSync(
`N=${count} update sync`,
buildManifest({
appId,
roleId,
universalIdentifiers,
checksumVersion: 'v2',
}),
);
// oxlint-disable-next-line no-console
console.log(
`[install-perf][test] ===== N=${count} SUMMARY: create=${createMs.toFixed(1)}ms update=${updateMs.toFixed(1)}ms =====`,
);
} finally {
await cleanupApplicationAndAppRegistration({
applicationUniversalIdentifier: appId,
});
}
},
120000,
);
});