67 lines
2.8 KiB
Diff
67 lines
2.8 KiB
Diff
From 1cbe2e6c0865f11fa264c24378bb0180cce6d414 Mon Sep 17 00:00:00 2001
|
|
From: chenjiankun <chenjiankun1@huawei.com>
|
|
Date: Wed, 22 Sep 2021 16:09:44 +0800
|
|
Subject: [PATCH] docker:fix bug where failed kills didnt fallback to unix kill
|
|
|
|
if killPossiblyDeadProcess fails, we expect to execute killProcessDirectly to
|
|
direct kill the process. But container.Wait return err when the timeout deadline
|
|
exceeded, and not execute the killProcessDirectly fucntion. Then docker stop will
|
|
hang.
|
|
---
|
|
components/engine/daemon/kill.go | 14 +++++++++-----
|
|
components/engine/daemon/stop.go | 6 ++++--
|
|
2 files changed, 13 insertions(+), 7 deletions(-)
|
|
|
|
diff --git a/components/engine/daemon/kill.go b/components/engine/daemon/kill.go
|
|
index 4c8ccf93d..593275cf8 100644
|
|
--- a/components/engine/daemon/kill.go
|
|
+++ b/components/engine/daemon/kill.go
|
|
@@ -153,8 +153,8 @@ func (daemon *Daemon) Kill(container *containerpkg.Container) error {
|
|
ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
|
|
defer cancel()
|
|
|
|
- if status := <-container.Wait(ctx, containerpkg.WaitConditionNotRunning); status.Err() != nil {
|
|
- return err
|
|
+ if status := <-container.Wait(ctx, containerpkg.WaitConditionNotRunning); status.Err() == nil {
|
|
+ return nil
|
|
}
|
|
}
|
|
|
|
@@ -166,9 +166,13 @@ func (daemon *Daemon) Kill(container *containerpkg.Container) error {
|
|
return err
|
|
}
|
|
|
|
- // Wait for exit with no timeout.
|
|
- // Ignore returned status.
|
|
- <-container.Wait3(context.Background(), containerpkg.WaitConditionNotRunning, waitStop)
|
|
+ // wait for container to exit one last time, if it doesn't then kill didnt work, so return error
|
|
+ ctx2, cancel2 := context.WithTimeout(context.Background(), 2*time.Second)
|
|
+ defer cancel2()
|
|
+
|
|
+ if status := <-container.Wait3(ctx2, containerpkg.WaitConditionNotRunning, waitStop); status.Err() != nil {
|
|
+ return errors.New("tried to kill container, but did not receive an exit event")
|
|
+ }
|
|
|
|
return nil
|
|
}
|
|
diff --git a/components/engine/daemon/stop.go b/components/engine/daemon/stop.go
|
|
index 40bc36dfd..741f5d5dd 100644
|
|
--- a/components/engine/daemon/stop.go
|
|
+++ b/components/engine/daemon/stop.go
|
|
@@ -82,8 +82,10 @@ func (daemon *Daemon) containerStop(container *containerpkg.Container, seconds i
|
|
logrus.Infof("Container %v failed to exit within %d seconds of signal %d - using the force", container.ID, seconds, stopSignal)
|
|
// 3. If it doesn't, then send SIGKILL
|
|
if err := daemon.Kill(container); err != nil {
|
|
- // Wait without a timeout, ignore result.
|
|
- <-container.Wait(context.Background(), containerpkg.WaitConditionNotRunning)
|
|
+ ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
|
|
+ defer cancel()
|
|
+
|
|
+ <-container.Wait(ctx, containerpkg.WaitConditionNotRunning)
|
|
logrus.Warn(err) // Don't return error because we only care that container is stopped, not what function stopped it
|
|
}
|
|
}
|
|
--
|
|
2.27.0
|
|
|