From d886f6c03cca051b45fd77cc77d0cc870aed1aed Mon Sep 17 00:00:00 2001 From: build Date: Wed, 4 Sep 2019 05:21:06 -0400 Subject: [PATCH] containerd: add timeout for I/O waitgroups reason: This and a combination of a couple Docker changes are needed to fully resolve the issue on the Docker side. However, this ensures that after processes exit, we still leave some time for the I/O to fully flush before closing. Without this timeout, the delete methods would block forever. Cherry-pick from upstream 245052243d Reference from https://github.com/containerd/containerd/pull/3361 Signed-off-by: Michael Crosby --- runtime/v1/linux/proc/exec.go | 2 +- runtime/v1/linux/proc/init.go | 2 +- runtime/v1/linux/proc/utils.go | 20 ++++++++++++++++++++ 3 files changed, 22 insertions(+), 2 deletions(-) diff --git a/runtime/v1/linux/proc/exec.go b/runtime/v1/linux/proc/exec.go index 715a977..08c581f 100644 --- a/runtime/v1/linux/proc/exec.go +++ b/runtime/v1/linux/proc/exec.go @@ -94,7 +94,7 @@ func (e *execProcess) setExited(status int) { } func (e *execProcess) delete(ctx context.Context) error { - e.wg.Wait() + waitTimeout(ctx, &e.wg, 2*time.Second) if e.io != nil { for _, c := range e.closers { c.Close() diff --git a/runtime/v1/linux/proc/init.go b/runtime/v1/linux/proc/init.go index 44d3f58..49fa8ec 100644 --- a/runtime/v1/linux/proc/init.go +++ b/runtime/v1/linux/proc/init.go @@ -263,7 +263,7 @@ func (p *Init) setExited(status int) { } func (p *Init) delete(context context.Context) error { - p.wg.Wait() + waitTimeout(context, &p.wg, 2*time.Second) err := p.runtime.Delete(context, p.id, nil) // ignore errors if a runtime has already deleted the process // but we still hold metadata and pipes diff --git a/runtime/v1/linux/proc/utils.go b/runtime/v1/linux/proc/utils.go index ab9f5fa..d6f047c 100644 --- a/runtime/v1/linux/proc/utils.go +++ b/runtime/v1/linux/proc/utils.go @@ -19,10 +19,12 @@ package proc import ( + "context" "encoding/json" "io" "os" "strings" + "sync" "time" "github.com/containerd/containerd/errdefs" @@ -103,3 +105,21 @@ func checkKillError(err error) error { func hasNoIO(r *CreateConfig) bool { return r.Stdin == "" && r.Stdout == "" && r.Stderr == "" } + +// waitTimeout handles waiting on a waitgroup with a specified timeout. +// this is commonly used for waiting on IO to finish after a process has exited +func waitTimeout(ctx context.Context, wg *sync.WaitGroup, timeout time.Duration) error { + ctx, cancel := context.WithTimeout(ctx, timeout) + defer cancel() + done := make(chan struct{}, 1) + go func() { + wg.Wait() + close(done) + }() + select { + case <-done: + return nil + case <-ctx.Done(): + return ctx.Err() + } +} -- 2.20.1