Type of Change

• New Feature
• Bug Fix
• Documentation
• Performance Improvement
• Test/CI
• Refactor
• Other

Related Issues

• The speed slowed down after shutting down one of the 4 nodes. #806

Summary of Changes

• set_disk.rs：新增 750 ms TTL 的磁盘健康缓存，只在缓存过期时对远端盘做 is_online() 探测，并在在线磁盘少于仲裁时立即返回 ErasureWriteQuorum，避免反复触发慢速路径。
  set_disk.rs: Added a 750 ms disk-health cache so writes probe remote disks only when the cache expires, and now fail fast with ErasureWriteQuorum when healthy disks are below quorum.
• set_disk.rs：补充 DiskHealthEntry 单测，验证缓存命中与过期语义。
  Added unit tests for DiskHealthEntry to ensure cache hit/expiry semantics.
• net.rs：测试 DNS 解析器改为 RAII guard（MockResolverGuard），并按需导入 MutexGuard，防止并发测试互相污染全局解析器。
  Updated the test DNS resolver to use an RAII MockResolverGuard and gated MutexGuard under #[cfg(test)] to eliminate test races.

Checklist

• I have read and followed the CONTRIBUTING.md guidelines
• Passed make pre-commit
• Added/updated necessary tests（新增 DiskHealthEntry 单测，net 测试仍可运行）
• Documentation updated (not needed)
• CI/CD passed (needs rerun with network)

Impact

• Breaking change (compatibility)
• Requires doc/config/deployment update
• Other impact: 写入仲裁不足时更早返回 ErasureWriteQuorum
  Other: Writes now fail fast with ErasureWriteQuorum when quorum isn’t met.