qwen-code CI: cold Docker sandbox builds blow the 60-minute cap, cancelling half of E2E runs
yiliang114 · ghdev · 2026-11-03
A P1 issue on QwenLM/qwen-code dissects why half of main E2E runs end up cancelled: the sandbox:docker leg always hits its 60-minute cap, and the time goes into the sandbox image build, not the tests.
- On a cold build, fetching just the apt index took 13m33s at 11.4 kB/s; vitest never even started before the timeout. Other cancelled legs the same day ran at 13–20 kB/s.
- Comparison data shows the leg only passes when the apt-get install layer is cached: 35–57 minutes with cache vs >59 minutes cold.
- The legacy builder can't reuse the apt layer on cold starts with slow pool-host networks, making results dependent on which host the job lands on.
Includes a per-host build-time table — a practical case study in Docker CI caching and self-hosted runner network debugging.
More from coding & agent
- Dev proposes LLM-generated ultra-rare content to build anime-style MMORPGs — rickasaurus · 2026-11-04
- Dev builds Chrome extension to bring back the classic Twitter bird on X — joonasvirtanen · 2026-11-04
- Developer lets his app record 'wishes' and auto-spawn OpenCode background agents to fulfill them — SubtleGradient · 2026-11-04
- Post-train your model on your own agent harness with OpenHands, Harbor, SkyRL and Snowflake — StasBekman · 2026-11-04
- Google Antigravity redesign leaked with context usage and quota indicators — clmt · 2026-11-04
- Microsoft's 4B Mage-VL-Cowork Hits 66% on OSWorld-Verified, Near Human Level — microsoft · 2026-11-04