Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

render: docker engine discards the network-removal error, so crossplane-render-* networks leak silently

未关闭
  • 哪里不对？
#398 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 3 天内回复

@jcogilvie 已经在做这个了。

开始于 2026年10月1日。

  • #405 来自 @jcogilvie —— 未关闭

评估

难度
4/5
预计耗时
3-5 天
新手友好度
48/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
docker, go
领域
infrastructure

调研方向

Start with the cleanup returned by Engine.Setup in cmd/crossplane/render/engine_docker.go, then trace its deferred use in xr/cmd.go and op/cmd.go around StopFunctionRuntimes. Reproduce the start-failure or slow-stop cases from #396 and #397, and verify that a failed network removal becomes visible to the caller or command instead of being discarded.

由索引模型根据 Issue 内容生成。

描述

bug
What happened?

The temporary crossplane-render-* network the docker render engine creates is left behind whenever a
container is still attached to it at cleanup time, and nothing reports the failure.

The cleanup returned by the docker engine's Setup discards its error
(engine_docker.go L106-L108):

cleanup := func() { //nolint:contextcheck // Detached context for cleanup.
	_ = removeRenderNetwork(context.Background(), networkID)
}

In xr/cmd.go and op/cmd.go this cleanup is deferred before StopFunctionRuntimes, so it runs last,
which is the right order. But if any container was leaked (#396, #397), it is still attached to the
network, NetworkRemove fails because the network has active endpoints, and the error is dropped. The
cleanup func signature (func()) means callers can't learn about it either.

Expected: a network-removal failure is reported (returned to, or at least logged visibly for, the caller),
so it doesn't vanish silently.

This affects library consumers that call Engine.Setup as well as the render commands.

How can we reproduce it?

Run either the start-failure reproduction from #396 or the slow-stop reproduction from #397. The run's network is still
there afterwards:

$ docker network ls --filter name=crossplane-render- --format '{{.CreatedAt}} {{.Name}}'
2026-09-30 18:55:29 -0400 EDT crossplane-render-b2rvl4jh

Once the leaked containers are removed by hand, docker network rm on the same network succeeds.

In our testing, every run that leaked a container also leaked its network (the start-failure run, all 7
slow-stop trials, and both interrupted runs from #400); each network's creation time matched its run.
Runs where every container was removed left no network. That points to the attached containers as what
makes network removal fail.

Possible fixes:

  • Return or log the removeRenderNetwork error instead of discarding it. Changing the Setup cleanup's type
    from func() to func() error would let callers handle it, though that changes the Engine interface.
  • Optionally, before removing the network, disconnect or force-remove any render-owned containers still
    attached to it. That would be much safer once render-created resources are labelled (#401), because a
    user-supplied --crossplane-docker-network may have unrelated containers on it.
What environment did it happen in?
  • Crossplane CLI version: built from main @ 29316fea54f2ede9d2c039d9c54f0c29cbad4b65
  • Platform (e.g., linux/amd64): darwin/arm64, Docker Engine 29.5.3 (Rancher Desktop)
  • Crossplane version (if applicable): render engine image xpkg.crossplane.io/crossplane/crossplane:stable
主要语言
Go
星标
19
派生
33
平均合并
3 天 10 小时
30 天内合并 PR
24

环境准备

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

crossplane/cli 的其他 Issue

查看 crossplane/cli 的全部 Issue

相似的 Issue

更多 Go Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。