mirror of
https://github.com/garrytan/gbrain.git
synced 2026-07-28 14:59:47 +00:00
* fix(cycle): budget the patterns subagent from remaining job time, not a fixed constant (#2781) The autopilot-cycle job gets an interval-derived timeout stamped at submit, but the patterns phase submits its subagent with a fixed 30-min job timeout and waits up to 35 min — one phase's worst case exceeds ANY interval-derived budget <= 35 min, so the parent job dead-letters mid-patterns and the tail phases (consolidate -> schema-suggest) starve for days (#2781, the deeper half left open by the #2852 dispatch-floor fix). - MinionJobContext.deadlineAtMs: absolute deadline from the claim-time timeout_at stamp (the DB ground truth handleTimeouts() sweeps against; re-stamped on every claim so retries get a fresh budget). Null when the job has no per-job timeout. - worker: the per-job abort timer now derives its delay from timeout_at when present, so the in-process timer, the DB sweeper, and the handler-visible deadline agree on ONE absolute instant. - autopilot-cycle + autopilot-global-maintenance handlers thread deadlineAtMs into runCycle; CycleOpts carries it to the patterns phase. - patterns: clampSubagentBudgets() derives BOTH the child job timeout and the wait timeout from the same child deadline (parent deadline minus a 60s stop-margin reserve — enough for the wait poll + force-evict grace + cleanup, deliberately NOT a promise that tail phases complete). Under a 2-min minimum the phase skips honestly (insufficient_cycle_budget) instead of submitting a guaranteed-kill LLM call; the next cycle retries with a fresh budget. - Direct callers (gbrain dream) pass no deadline and keep the configured timeouts unchanged. Follow-up (separate PR): synthesize has the same shape plus sequential per-child waits that accumulate N x subagent_wait_timeout_ms past any parent budget; it needs per-wait remaining-time recomputation. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RaNDjecPPvhC7LRqkxkhGF * fix(cycle): address review — cancel timed-out patterns child; thread deadline through phase-wrapper handlers - P1: the child's timeout_ms clock starts at ITS claim, so a queued child could outlive the parent deadline the wait was clamped to. On wait timeout, cancelJob strips it (waiting -> cancelled; active -> lock stripped, worker abort fires next renew tick). - P2: makePhaseHandler (standalone patterns/synthesize/... minion jobs) now threads job.deadlineAtMs into runCycle like the autopilot handlers. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RaNDjecPPvhC7LRqkxkhGF --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
145 lines
5.2 KiB
TypeScript
145 lines
5.2 KiB
TypeScript
/**
|
|
* Tests for src/core/minions/handlers/embed-backfill.ts (v0.40 D2, D6).
|
|
*
|
|
* Validates the handler-side contract:
|
|
* - Happy path: embeds, returns 'success' with chunk + spend counts
|
|
* - D2 lock: second concurrent handler call returns 'already_in_progress'
|
|
* - D15.1 finally: lock ALWAYS releases (try/finally even on abort)
|
|
*
|
|
* Hermetic — uses injected embedFn via the underlying embedStaleForSource
|
|
* test seam? No — the handler doesn't expose embedFn passthrough. Instead
|
|
* we exercise the handler against a brain with zero stale chunks so no
|
|
* actual embed call lands. That gives us a deterministic test of the lock
|
|
* + budget + status branches without needing a fake gateway.
|
|
*
|
|
* The kill-resume contract is covered by test/embed-stale.test.ts at the
|
|
* helper layer; the handler just routes through.
|
|
*/
|
|
import { describe, test, expect, beforeAll, afterAll, beforeEach } from 'bun:test';
|
|
import { PGLiteEngine } from '../src/core/pglite-engine.ts';
|
|
import { makeEmbedBackfillHandler } from '../src/core/minions/handlers/embed-backfill.ts';
|
|
import { tryAcquireDbLock } from '../src/core/db-lock.ts';
|
|
import type { MinionJobContext } from '../src/core/minions/types.ts';
|
|
|
|
let engine: PGLiteEngine;
|
|
|
|
beforeAll(async () => {
|
|
engine = new PGLiteEngine();
|
|
await engine.connect({});
|
|
await engine.initSchema();
|
|
}, 30000);
|
|
|
|
afterAll(async () => {
|
|
await engine.disconnect();
|
|
});
|
|
|
|
beforeEach(async () => {
|
|
// Clean minion_jobs + lock rows. Preserve config (schema version + flags).
|
|
await engine.executeRaw('DELETE FROM minion_jobs');
|
|
await engine.executeRaw(`DELETE FROM gbrain_cycle_locks WHERE id LIKE 'gbrain-embed-backfill:%'`);
|
|
});
|
|
|
|
/** Build a minimal MinionJobContext for testing. */
|
|
function fakeJob(data: Record<string, unknown>): MinionJobContext {
|
|
const controller = new AbortController();
|
|
return {
|
|
id: 1,
|
|
name: 'embed-backfill',
|
|
data,
|
|
attempts_made: 0,
|
|
signal: controller.signal,
|
|
deadlineAtMs: null,
|
|
shutdownSignal: controller.signal,
|
|
updateProgress: async () => {},
|
|
updateTokens: async () => {},
|
|
log: async () => {},
|
|
isActive: async () => true,
|
|
readInbox: async () => [],
|
|
};
|
|
}
|
|
|
|
describe('embed-backfill handler — happy path', () => {
|
|
test('zero stale chunks → success with embedded=0', async () => {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
const result = await handler(fakeJob({ sourceId: 'default' }));
|
|
expect(result).toMatchObject({
|
|
status: 'success',
|
|
sourceId: 'default',
|
|
embedded: 0,
|
|
chunksProcessed: 0,
|
|
pagesProcessed: 0,
|
|
});
|
|
});
|
|
|
|
test('throws when sourceId missing', async () => {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
await expect(handler(fakeJob({}))).rejects.toThrow(/sourceId is required/);
|
|
});
|
|
|
|
test('throws when sourceId is empty string', async () => {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
await expect(handler(fakeJob({ sourceId: '' }))).rejects.toThrow(/sourceId is required/);
|
|
});
|
|
});
|
|
|
|
describe('embed-backfill handler — D2 lock contract', () => {
|
|
test('IRON-RULE: second call returns already_in_progress when lock is held', async () => {
|
|
// Hold the per-source lock externally
|
|
const lock = await tryAcquireDbLock(engine, 'gbrain-embed-backfill:default', 60);
|
|
expect(lock).not.toBeNull();
|
|
|
|
try {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
const result = await handler(fakeJob({ sourceId: 'default' }));
|
|
expect(result).toMatchObject({
|
|
status: 'already_in_progress',
|
|
sourceId: 'default',
|
|
embedded: 0,
|
|
spentUsd: 0,
|
|
});
|
|
} finally {
|
|
await lock?.release();
|
|
}
|
|
});
|
|
|
|
test('different sources do not contend on each other locks', async () => {
|
|
await engine.executeRaw(
|
|
`INSERT INTO sources (id, name, config) VALUES ('other-src', 'other-src', '{"federated":true}') ON CONFLICT (id) DO NOTHING`,
|
|
);
|
|
const lockA = await tryAcquireDbLock(engine, 'gbrain-embed-backfill:default', 60);
|
|
expect(lockA).not.toBeNull();
|
|
try {
|
|
// 'other-src' should still succeed
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
const result = await handler(fakeJob({ sourceId: 'other-src' }));
|
|
expect(result.status).toBe('success');
|
|
} finally {
|
|
await lockA?.release();
|
|
}
|
|
});
|
|
|
|
test('IRON-RULE: lock is released after handler completes (try/finally)', async () => {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
await handler(fakeJob({ sourceId: 'default' }));
|
|
|
|
// After handler returns, the lock row should NOT block a fresh acquire.
|
|
const lock = await tryAcquireDbLock(engine, 'gbrain-embed-backfill:default', 60);
|
|
expect(lock).not.toBeNull();
|
|
await lock?.release();
|
|
});
|
|
|
|
test('IRON-RULE: lock released on throw (sourceId-missing path)', async () => {
|
|
const handler = makeEmbedBackfillHandler(engine);
|
|
try {
|
|
await handler(fakeJob({})); // throws before lock is acquired
|
|
} catch {
|
|
// expected
|
|
}
|
|
// Lock was never acquired (throw happened in parseParams pre-lock),
|
|
// so the row should be cleanly absent. Verify a fresh acquire works.
|
|
const lock = await tryAcquireDbLock(engine, 'gbrain-embed-backfill:default', 60);
|
|
expect(lock).not.toBeNull();
|
|
await lock?.release();
|
|
});
|
|
});
|