From 245052243d23c8de21fcc95bbf47fb1dbc731ab4 Mon Sep 17 00:00:00 2001
From: Michael Crosby <crosbymichael@gmail.com>
Date: Thu, 20 Jun 2019 16:13:51 -0400
Subject: [PATCH] Add timeout for I/O waitgroups

Closes #3286

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.

Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
---
 runtime/v1/linux/proc/exec.go  |  2 +-
 runtime/v1/linux/proc/init.go  |  2 +-
 runtime/v1/linux/proc/utils.go | 19 +++++++++++++++++++
 3 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/runtime/v1/linux/proc/exec.go b/runtime/v1/linux/proc/exec.go
index 5ab232ae7..1779725d6 100644
--- a/runtime/v1/linux/proc/exec.go
+++ b/runtime/v1/linux/proc/exec.go
@@ -106,7 +106,7 @@ func (e *execProcess) Delete(ctx context.Context) error {
 }
 
 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 c6a09575a..36e6f1674 100644
--- a/runtime/v1/linux/proc/init.go
+++ b/runtime/v1/linux/proc/init.go
@@ -284,7 +284,7 @@ func (p *Init) Delete(ctx context.Context) error {
 }
 
 func (p *Init) delete(ctx context.Context) error {
-	p.wg.Wait()
+	waitTimeout(ctx, &p.wg, 2*time.Second)
 	err := p.runtime.Delete(ctx, 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 2d085e62f..a2b236f65 100644
--- a/runtime/v1/linux/proc/utils.go
+++ b/runtime/v1/linux/proc/utils.go
@@ -19,6 +19,7 @@
 package proc
 
 import (
+	"context"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -143,3 +144,21 @@ func (p *pidFile) Path() string {
 func (p *pidFile) Read() (int, error) {
 	return runc.ReadPidFile(p.path)
 }
+
+// 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()
+	}
+}