douyong1285 2019-06-17 09:06
浏览 177
已采纳

如何转储正在运行的kubelet的goroutines堆栈竞赛

kubernetes is complicated, kubelet run into deadlocks after long running in some scenarios.

Is there a way to dump goroutine stack trace of the running kubelet?

The expected output like following which is very helpful to debug deadlock kind issues of kubelet.

goroutine 386 [chan send, 1140 minutes]:
k8s.io/kubernetes/pkg/kubelet/pleg.(*GenericPLEG).relist(0xc42069ea20)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/kubelet/pleg/generic.go:261 +0x74e
k8s.io/kubernetes/pkg/kubelet/pleg.(*GenericPLEG).(k8s.io/kubernetes/pkg/kubelet/pleg.relist)-fm()
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/kubelet/pleg/generic.go:130 +0x2a
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.JitterUntil.func1(0xc4212ee520)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:133 +0x54
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.JitterUntil(0xc4212ee520, 0x3b9aca00, 0x0, 0x1, 0xc420056540)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:134 +0xbd
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.Until(0xc4212ee520, 0x3b9aca00, 0xc420056540)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:88 +0x4d
created by k8s.io/kubernetes/pkg/kubelet/pleg.(*GenericPLEG).Start
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/kubelet/pleg/generic.go:130 +0x88
...
goroutine 309 [sleep]:
time.Sleep(0x12a05f200)
    /usr/local/go/src/runtime/time.go:102 +0x166
k8s.io/kubernetes/pkg/kubelet.(*Kubelet).syncLoop(0xc4205e3b00, 0xc420ff2780, 0x3e56a60, 0xc4205e3b00)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/kubelet/kubelet.go:1777 +0x1e7
k8s.io/kubernetes/pkg/kubelet.(*Kubelet).Run(0xc4205e3b00, 0xc420ff2780)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/kubelet/kubelet.go:1396 +0x27f
k8s.io/kubernetes/cmd/kubelet/app.startKubelet.func1()
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/cmd/kubelet/app/server.go:998 +0x67
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.JitterUntil.func1(0xc42105dfb0)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:133 +0x54
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.JitterUntil(0xc42105dfb0, 0x0, 0x0, 0x1, 0xc420056540)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:134 +0xbd
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait.Until(0xc42105dfb0, 0x0, 0xc420056540)
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/wait/wait.go:88 +0x4d
created by k8s.io/kubernetes/cmd/kubelet/app.startKubelet
    /workspace/anago-v1.11.5-beta.0.24+753b2dbc622f5c/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/cmd/kubelet/app/server.go:996 +0xea
...

I appreciate that anyone could share the experience about how to dump goroutines stack race of kubelet that something like what docker provided[1]

$ pkill -SIGUSR1 dockerd

[1]. https://success.docker.com/article/how-to-dump-goroutines-stacktraces

  • 写回答

1条回答 默认 最新

  • duannuci4008 2019-06-18 12:10
    关注
    1. pprof, it will keep kubelet running

    2. send signal to kubelet which caused kubelet to exit with a stack dump

      kill -SIGABRT

    本回答被题主选为最佳回答 , 对您是否有帮助呢?
    评论

报告相同问题?

悬赏问题

  • ¥15 安卓adb backup备份应用数据失败
  • ¥15 eclipse运行项目时遇到的问题
  • ¥15 关于#c##的问题:最近需要用CAT工具Trados进行一些开发
  • ¥15 南大pa1 小游戏没有界面,并且报了如下错误,尝试过换显卡驱动,但是好像不行
  • ¥15 没有证书,nginx怎么反向代理到只能接受https的公网网站
  • ¥50 成都蓉城足球俱乐部小程序抢票
  • ¥15 yolov7训练自己的数据集
  • ¥15 esp8266与51单片机连接问题(标签-单片机|关键词-串口)(相关搜索:51单片机|单片机|测试代码)
  • ¥15 电力市场出清matlab yalmip kkt 双层优化问题
  • ¥30 ros小车路径规划实现不了,如何解决?(操作系统-ubuntu)