chore(ckpt): release v0.4.3

[yummypeng]

Summary

• ws-ckpt v0.4.3 release: k8s sidecar containerized deployment support, daemon memory leak fix, loop device direct-IO optimization, and other stability improvements
• Bumps version across Cargo.toml, Cargo.lock, component.toml, plugin.yaml, package.json, and spec

Changes since v0.4.2

Features

• Added k8s sidecar containerized deployment support (fix(ckpt): handle external mounts in k8s sidecar scenario #2034)

Bug Fixes

• Fixed daemon memory leak by reaping finished tasks in accept loop (fix(ckpt): reap finished tasks in accept loop #2554)
• Fixed loop device direct-IO mode to eliminate double page-cache bottleneck (fix(ckpt): open loop device in direct-io mode #2523)
• Fixed bootstrap failure cleanup and error reporting (fix(ckpt): improve bootstrap failure handling and error reporting #1956)
• Fixed init to reject mounted workspace root with a clear error (fix(ckpt): reject mounted workspace root and document interop #1798)
• Added RPM component identity declaration (fix(ckpt): add RPM component identity #2568)

Test plan

• cargo fmt --check && cargo clippy -D warnings && cargo test pass in src/ws-ckpt/src/
• RPM build via rpmbuild -bb ws-ckpt.spec succeeds
• Version strings consistent across all manifests (0.4.3)

🤖 Generated with Claude Code

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

Yuanhong Peng seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7420683dba

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[kongche-jbw]

Review baseline: c26b6fd20390...7420683dbaf3

[P1] 按规范重写 0.4.3 发布说明

src/ws-ckpt/CHANGELOG.md:7 新增的 Features / Bug Fixes 与仓库要求的
Keep a Changelog Added / Fixed 标题不一致；第 11–15 行又主要描述
accept loop、direct-IO 和 bootstrap 等实现细节，没有说明用户能观察到的改善。
发布后用户无法据此判断内存增长、性能退化和启动失败等问题在何种场景下已解决，
src/ws-ckpt/CHANGELOG_zh.md:7 的对应条目也存在同样问题。
Possible direction: 改用规范章节名，将中英文条目都改为触发场景及用户可观察结果，
并补充组件 CHANGELOG 的格式回归检查。

[yummypeng]

@kongche-jbw 感谢 review，两点都已修复：

1. 标题格式：所有版本条目（含历史）已从 Features / Bug Fixes / Breaking Changes 统一为 Keep a Changelog 的 Added / Fixed / Changed，中英文各 11 处。
2. 用户视角描述：0.4.3 条目已重写，去掉 accept loop、direct-IO、page cache 等实现细节，改为描述用户可观察的效果（如"持续负载下内存无限增长最终 OOM"、"并发 IO 下 p99 延迟降低至 1/5"）。

[kongche-jbw]

Review baseline: c26b6fd20390...21bd181defde

[P1] 不要把整体延迟改善写成 p99 改善

src/ws-ckpt/CHANGELOG.md:12 和 CHANGELOG_zh.md:14 把 #2523 写成 p99
降低最多 5 倍，但该 PR 的约 5 倍数据是 fio 压力下 checkpoint 延迟从
305.10 ms 降至 62.55 ms。它只列出修复后的 p99 为 237 ms，与 btrfs-base
相同，没有提供修复前 p99，因此当前发布说明承诺了未经测量的尾延迟改善。
Possible direction: 删除 p99，按 #2523 的实际指标同步修正中英文和 RPM
changelog；若保留量化结论，请补充对应 benchmark 证据，并在 release 检查中
核对量化指标与引用 PR 的数据。

[kongche-jbw]

Review baseline: 702b26bb680e...91bbc901f49c

未发现阻塞问题。

Remaining risks:

• Test ws-ckpt 当前在未改动的 dispatcher.rs:444 因 Clippy 告警失败，
  后续测试及 coverage gates 未运行。
• license/cla 仍为 pending。

Verification scope:

• 复查了完整九文件 diff、全部评论与 review threads，以及六个关联 PR。
• 版本一致性检查、Docs/PR lint 与 git diff --check 通过。
• 最新表述与 #2523 的 305.10 ms → 62.55 ms benchmark 一致。

[kongche-jbw]

lgtm thx