Files
OmniRoute/tests/unit/probe-testall-isolation.test.ts
Diego Rodrigues de Sa e Souza 93265eede3 test(infra): retry recursive temp-dir removal on main (main twin of #11968) (#12246)
* test(infra): retry recursive temp-dir removal on main (main twin of #11968)

`main` has been red since b342c1a361 on the vitest and integration gates:

  ✖ tests/unit/autoCombo/provider-family-combos.test.ts > auto/<family>
  ✖ chat pipeline applies Codex OAuth fingerprint and priority tier inside combos

Both call resetStorage() from beforeEach, which does an fs.rmSync(TEST_DATA_DIR,
{recursive: true, force: true}) with no retry, and intermittently loses the race
with a not-yet-released SQLite handle (ENOTEMPTY).

release/v3.8.51 fixed this in #11968 with a mechanical codemod adding
maxRetries/retryDelay to every recursive rm/rmSync/rmdirSync under tests/, but
that PR landed only on the release branch. Because main only receives work at
the release squash, it stayed broken for the whole cycle — and repo-wide gates
then turn every open PR into main red on checks unrelated to their diff.

This is the --base main twin: re-runs the same codemod that already shipped on
the release branch (scripts/ad-hoc/codemod-rm-maxretries.mjs), so the two
branches converge on identical test-teardown semantics. Test-only; no product
logic is touched.

The remaining three failures reported on #12133 (unit full suite exceeding its
4800s ceiling, package-artifact exceeding 1200s, and the boot-smoke that is
skipped as a consequence) are runner-contention timeouts, not code defects —
validate-release-green.mjs runs those heavy gates concurrently on one shared
hosted runner. There is no fix to port for those.

* chore(scripts): carry the rm-maxretries codemod onto main alongside its output

The codemod that generated the previous commit lives in the repo on
release/v3.8.51 (added by #11968) but was never on main. Bringing it over keeps
the tool next to the change it produced, so the transformation stays
reproducible and auditable from either branch.
2026-09-01 01:48:00 -03:00

177 lines
6.2 KiB
TypeScript

import { test } from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-probe-testall-"));
process.env.DATA_DIR = TEST_DATA_DIR;
const core = await import("../../src/lib/db/core.ts");
const { createProviderConnection } = await import("../../src/lib/db/providers.ts");
const { runSingleModelTest } = await import("../../src/lib/api/modelTestRunner.ts");
const { resetAllCircuitBreakers } = await import("../../src/shared/utils/circuitBreaker.ts");
const { invalidateDbCache } = await import("../../src/lib/db/readCache.ts");
const { refreshConnectionRateLimits, enableRateLimitProtection } =
await import("@omniroute/open-sse/services/rateLimitManager.ts");
const originalFetch = globalThis.fetch;
// A test-all 403 can also open the provider circuit breaker and stale the
// 5s connections read cache (rawConnectionsCache) — either would
// short-circuit the NEXT tests before chatCore, a false positive for the
// isolation asserts. Reset both before every test.
test.beforeEach(() => {
resetAllCircuitBreakers();
invalidateDbCache("connections");
});
test.after(() => {
globalThis.fetch = originalFetch;
core.resetDbInstance();
fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 });
});
function readConnectionRow(connId: string) {
const db = core.getDbInstance() as unknown as {
prepare: (sql: string) => {
get: (id: string) =>
| {
is_active: unknown;
test_status: unknown;
rate_limited_until: unknown;
last_error: unknown;
}
| undefined;
};
};
return db
.prepare(
"SELECT is_active, test_status, rate_limited_until, last_error FROM provider_connections WHERE id = ?"
)
.get(connId);
}
async function createConnection(): Promise<string> {
const conn = await createProviderConnection({
provider: "openai",
authType: "apikey",
name: "probe-testall",
apiKey: "sk-probe-testall", // pragma: allowlist secret
isActive: true,
testStatus: "active",
});
return String((conn as { id: string }).id);
}
// Warm up the chat-completions pipeline (SSE translators, compression
// settings, etc. lazy-init on the first real request in a process) with a
// fast success mock, mirroring model-test-runner.test.ts.
async function warmUp(connId: string): Promise<void> {
globalThis.fetch = async () =>
new Response(JSON.stringify({ choices: [{ message: { role: "assistant", content: "OK" } }] }), {
headers: { "content-type": "application/json" },
});
await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
}
async function mockUpstream(status: number, message: string): Promise<void> {
globalThis.fetch = async () =>
new Response(JSON.stringify({ error: { message } }), {
status,
headers: { "content-type": "application/json" },
});
}
const ASSERT_ISOLATED = (connId: string) => {
const row = readConnectionRow(connId);
assert.equal(row?.is_active, 1, "connection stays active after a probe failure");
assert.notEqual(row?.test_status, "banned", "no terminal banned status from a probe");
assert.notEqual(row?.test_status, "deactivated", "no deactivated status from a probe");
assert.equal(row?.rate_limited_until, null, "no cooldown persisted by a probe");
assert.ok(row?.last_error, "probe failure is recorded for visibility");
};
test("test-all FORBIDDEN failure (Sentinel) does not deactivate the connection", async () => {
const connId = await createConnection();
await warmUp(connId);
await mockUpstream(403, "SENTINEL_BLOCKED");
const result = await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
assert.equal(result.status, "error");
ASSERT_ISOLATED(connId);
});
test("test-all ACCOUNT_DEACTIVATED failure does not deactivate the connection", async () => {
const connId = await createConnection();
await warmUp(connId);
await mockUpstream(403, "this account is deactivated");
const result = await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
assert.equal(result.status, "error");
ASSERT_ISOLATED(connId);
});
test("queued (rate-limited) test-all failure stays isolated", async () => {
const connId = await createConnection();
await warmUp(connId);
// Rate-limit protection is OFF by default for test connections (empty
// enabledConnections — withRateLimit:537-540 would short-circuit directly,
// a false positive on the inner wrapper). Enable it so the calls really
// go through the Bottleneck limiter.
enableRateLimitProtection(connId);
// minTime 200 forces the 2nd job to wait behind the 1st — the queued job
// must run inside the probe context (inner wrapper in withRateLimit);
// without it, ASSERT_ISOLATED goes red.
refreshConnectionRateLimits(connId, { minTime: 200 });
await mockUpstream(403, "SENTINEL_BLOCKED");
await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
ASSERT_ISOLATED(connId);
});
test("RESTORE: probe with opt-in probeCanDisable=true deactivates like real traffic", async () => {
const settingsDb = await import("../../src/lib/db/settings.ts");
await settingsDb.updateSettings({ probeCanDisable: true });
try {
const connId = await createConnection();
await warmUp(connId);
await mockUpstream(403, "SENTINEL_BLOCKED");
const result = await runSingleModelTest({
providerId: "openai",
modelId: "gpt-4o",
connectionId: connId,
timeoutMs: 10_000,
});
assert.equal(result.status, "error");
const row = readConnectionRow(connId);
assert.equal(row?.is_active, 0, "opt-in restores historical behavior: probe deactivates");
assert.equal(row?.test_status, "banned", "terminal banned status restored for probe");
} finally {
await settingsDb.updateSettings({ probeCanDisable: false });
}
});