当我尝试遵循Kubernetes Configure Multiple Schedulers guide时,无法成功运行my-scheduler pod。
我成功构建了镜像并将其推送到注册表,该镜像显示在我的docker和Google Could Platform中。
但是,当我应用部署文件设置Pod时,状态为CrashLoopBackOff
。
my-scheduler-64576b7897-8b549 0/1 CrashLoopBackOff 14 84m
这是
kuberctl describe pod my-scheduler-64576b7897-8b549 -n kube-system
的输出:Name: my-scheduler-64576b7897-8b549
Namespace: kube-system
Priority: 0
PriorityClassName: <none>
Node: minikube/10.0.2.15
Start Time: Tue, 07 May 2019 15:06:59 -0500
Labels: component=scheduler
pod-template-hash=64576b7897
tier=control-plane
version=second
Annotations: <none>
Status: Running
IP: 172.17.0.15
Controlled By: ReplicaSet/my-scheduler-64576b7897
Containers:
kube-second-scheduler:
Container ID: docker://a0cadf03dd5ffd11cbb6f1f8e669fc04cd6aa00aa6f361550f06f32785a4d63c
Image: gcr.io/my-gcp-project-239900/my-kube-scheduler:1.0
Image ID: docker-pullable://gcr.io/my-gcp-project-239900/my-kube-scheduler@sha256:0895687b7646d159438f086e8f8b23b253332930e169cf29d8b9951ebf3ba653
Port: <none>
Host Port: <none>
Command:
/usr/local/bin/kube-scheduler
--address=1.1.1.1
--leader-elect=false
--scheduler-name=my-scheduler
State: Terminated
Reason: Error
Exit Code: 1
Started: Tue, 07 May 2019 15:07:45 -0500
Finished: Tue, 07 May 2019 15:07:45 -0500
Last State: Terminated
Reason: Error
Exit Code: 1
Started: Tue, 07 May 2019 15:07:24 -0500
Finished: Tue, 07 May 2019 15:07:24 -0500
Ready: False
Restart Count: 3
Requests:
cpu: 100m
Liveness: http-get http://:10251/healthz delay=15s timeout=1s period=10s #success=1 #failure=3
Readiness: http-get http://:10251/healthz delay=0s timeout=1s period=10s #success=1 #failure=3
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from my-scheduler-token-lzwvk (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
my-scheduler-token-lzwvk:
Type: Secret (a volume populated by a Secret)
SecretName: my-scheduler-token-lzwvk
Optional: false
QoS Class: Burstable
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 48s default-scheduler Successfully assigned kube-system/my-scheduler-64576b7897-8b549 to minikube
Normal Pulled 2s (x4 over 47s) kubelet, minikube Container image "gcr.io/my-gcp-project-239900/my-kube-scheduler:1.0" already present on machine
Normal Created 2s (x4 over 47s) kubelet, minikube Created container kube-second-scheduler
Normal Started 2s (x4 over 47s) kubelet, minikube Started container kube-second-scheduler
Warning BackOff 1s (x6 over 45s) kubelet, minikube Back-off restarting failed container
当我尝试查看此Pod的日志时,出现此错误消息:
standard_init_linux.go:207: exec user process caused "exec format error"
我的
Dockerfile
与指南有点不同,它看起来像:FROM busybox
ADD ./_output/local/bin/darwin/amd64/kube-scheduler /usr/local/bin/kube-scheduler
我变了
ADD ./_output/dockerized/bin/linux/amd64/kube-scheduler /usr/local/bin/kube-scheduler
进入
ADD ./_output/local/bin/darwin/amd64/kube-scheduler /usr/local/bin/kube-scheduler
因为
./_output/local/bin/darwin/amd64/kube-scheduler
是调度程序文件的正确位置。如果有人能告诉我为什么会这样,我也将不胜感激。一些更新
文件位置不同
按照指南,我在shell中键入了这些命令
git clone https://github.com/kubernetes/kubernetes.git
cd kubernetes
make
通过执行
make
,Makefile
文件将在特定路径(包括我们想要的kube-scheduler)中生成kubernetes exec文件。生成路径与您的系统有关。由于我使用 minikube 来运行kubernetes和docker,因此它在 darwin(amd64)上运行。我认为这就是我的路线与指南不同的原因。“exec格式错误”
对于这个错误
standard_init_linux.go:207: exec user process caused "exec format error"
我发现this issue建议在尝试在容器中运行命令之前添加
#!/bin/bash
。这告诉系统将命令视为bash命令。在busybox中,它应该是/bin/sh
。 (我对 shell 语言机制了解不多,如果我错了,请纠正我)所以我在
my-sheduler.yaml
中更改了这些行:- command:
- /usr/local/bin/kube-scheduler
- --address=0.0.0.0
- --leader-elect=false
- --scheduler-name=my-scheduler
进入
- command:
- /bin/sh
- /usr/local/bin/kube-scheduler
- --address=0.0.0.0
- --leader-elect=false
- --scheduler-name=my-schedule
现在,
my-sheduler
容器仍然失败,并出现CrashLoopBackOff
错误。好消息是执行
kubectl logs
时现在有了日志。坏消息是这样的日志不可读
/usr/local/bin/kube-scheduler: line 1: ����: not found
/usr/local/bin/kube-scheduler: line 2: syntax error: unexpected ")"
我认为这是由于docker镜像中发生了某些事情而发生的。但是由于我是从kubernetes github page直接使用脚本,所以应该没有任何错误。
当我运行
docker run -it {my-image-id} /bin/sh
并进入镜像并运行exec kube-scheduler文件时,将显示相同的日志。我无法弄清楚GO脚本中的内容会导致此问题。
另一种选择
我最终尝试了另一种构建自定义调度程序的方法。
我遵循了本教程:https://www.youtube.com/watch?v=IYcL0Un1io0&t=1807s
您可以在此处找到源代码:https://github.com/kelseyhightower/scheduler
与官方指南相比,这是更清晰的信息(在官方指南中,您必须构建整个程序才能获取调度程序文件)。
如果有人遇到相同的问题,希望这对您有所帮助。
最佳答案
这完全看起来像是您使用darwin
kube-scheduler
go二进制文件在Linux container中运行。
您可能在Mac上运行了此命令:
git clone https://github.com/kubernetes/kubernetes.git
cd kubernetes
make
它创建了
darwin
二进制文件,而不是Linux二进制文件。我会尝试从Linux机器上运行它。或将make目标更改为使用 GOOS=linux
and GOARCH=amd64
。请注意,在撰写本文时,K8主要支持Linux,仅支持Windows节点。
关于go - 遵循 “Configure Multiple Schedulers” Kubernetes指南时发生错误,我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/56030098/