CalvinBackup/gstack
1
0
Fork 0
mirror of https://github.com/garrytan/gstack.git synced 2026-09-09 14:38:59 +02:00
Code Issues Packages Projects Releases Wiki Activity

feat: paid-runner defaults 4x4 -> 8x2 — halve the local gate worst case

Browse Source 
39 of 75 skill-e2e files hold exactly ONE test, so within-shard
concurrency was dead weight for most shards: 4 jobs x 4 concurrency
yielded only ~4-6 real in-flight sessions and a 13-wave local gate worst
case (~6.5h). 8 jobs x 2 gives ~10-13 in-flight — under the
documented-safe ~15 — and ~7 waves (~3.3h worst case). CI lanes keep
their explicit EVALS_JOBS env (2 per slice; 4 for gate-census); this
changes local defaults. Rollback trigger: sustained 429 storms in the WS1
telemetry across 2 PR cycles. test/eval-detach-timeout-floor.test.ts
recomputed green (the raise LOWERS the worst-case floor).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Garry Tan
2026-08-31 05:02:08 +00:00
co-authored by Claude Fable 5
parent 3ec20bc33f
commit be30f0eb52
1 changed files with 10 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/test-paid-shards.ts
+10 -5
View File
@@ -85,11 +85,16 @@ export type PaidTier = 'gate' | 'periodic';
export const DEFAULT_TIER: PaidTier = 'gate';
export const DEFAULT_SHARD_TIMEOUT_MS = 30 * 60_000;
export const DEFAULT_MAX_FILES_PER_SHARD = 1;
export const DEFAULT_JOBS = 4;
// Within one shard's bun process. 4 jobs × 4 ≈ the legacy single-process
// default of 15, keeping total in-flight `claude` sessions inside known-safe
// API rate headroom.
export const DEFAULT_WITHIN_SHARD_CONCURRENCY = 4;
// 8 jobs × 2 within-shard ≈ 10-13 real in-flight sessions (39 of 75
// skill-e2e files hold exactly ONE test, so within-shard concurrency is
// dead weight for most shards) — under the documented-safe ~15 the legacy
// 40-way runner established. The old 4×4 yielded only ~4-6 in-flight and a
// 13-wave local gate worst case (~6.5h); 8×2 halves it. Watch the WS1
// flake telemetry for sustained 429 storms across 2 PR cycles — that is
// the rollback trigger. Prerequisite (landed): per-shard TMPDIR/
// CHROMIUM_PROFILE isolation in runPaidShard.
export const DEFAULT_JOBS = 8;
export const DEFAULT_WITHIN_SHARD_CONCURRENCY = 2;
export function collectPaidTestFiles(rootDir = ROOT): string[] {
const testDir = path.join(rootDir, 'test');