[CI] Fix nightly test error and add container cleanup in build_rl

[EmmonsCurse]

Motivation

• Fix failures observed in nightly tests, primarily related to container runtime and system-level limits.
• Address issues caused by insufficient kernel parameters (e.g., shared memory / IPC limits), which can lead to unstable behavior or test crashes.
• Improve CI security and stability by removing unnecessary --privileged usage.

Modifications

• Fix nightly test errors by stabilizing the container runtime environment.
• Add required sysctl parameters to ensure sufficient system resource limits for model inference and multiprocessing workloads.
• Remove --privileged from Docker execution:
  • Reduce security risks by avoiding full host privilege exposure.
  • Replace with minimal required capabilities / configurations.
  • Ensure compatibility with CI environments where --privileged may be restricted.
• Add container cleanup logic in build_rl to avoid leftover containers affecting subsequent jobs.

Usage or Command

N/A

Accuracy Tests

N/A

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[EmmonsCurse]

/skip-ci all

[paddle-bot]

Thanks for your contribution!

[fastdeploy-bot]

🤖 AI Code Review | 2026-04-11

📋 Review 摘要

PR 概述：修复 nightly 测试错误，移除 --privileged 增强安全性，新增容器清理逻辑

变更范围：CI workflow 配置（.github/workflows/）

影响面 Tag：[CI]

PR 规范检查

PR 标题和描述符合规范：

• ✅ 标题包含有效标签 [CI]
• ✅ 描述完整填写了 Motivation 和 Modifications 章节

发现的问题

级别	文件	概述
🟡 建议	.github/workflows/_build_linux_rl.yml:142	--shm-size=64G 大小写与 _accuracy_test.yml 中的 64g 不统一

总体评价

PR 变更整体合理：

1. 移除 --privileged 减少安全风险，保留必要的 --cap-add=SYS_PTRACE
2. 新增 --sysctl 参数增加 IPC 消息队列容量，与测试 workflow 保持一致
3. 新增容器清理逻辑与项目现有实践一致

唯一的小问题是 --shm-size 单位大小写不统一（64G vs 64g），建议保持一致性。

[fastdeploy-bot]

🟡 建议 --shm-size=64G 与本 PR 中 _accuracy_test.yml 的 64g 大小写不一致。

建议统一使用小写单位（64g）以保持一致性。