I started my cluster with the  kubeadm init command and joined 2 nodes on it.
For testing purposes i wanted a nginx server inside a container. just to test 
the high availability of the cluster. 
I created a replication controller yaml file  that will create a pod with 3 
replica's and inside the pod a nginx container. With a service yaml file i 
created a service so that i need to surf to 1 ip address to access the nginx 
server. 

So far so good . i see the pods and which nodes they are running one with the 
kubectl get pods -o wide command. And i struggle with a high availability 
problem. Rught now i'am testing inside virtual machines until my raspberry pi's 
are arrived. But the goal is a high available raspberry pi cluster. When i kill 
one node (poweroff 1 vm ) it takes 5 minutes until kubernetes knows that a node 
is down. it reschedule it good but 5 minutes is too long , especially when in 
production on a raspberry pi cluster. 

Online(https://stackoverflow.com/a/47317892/6351302)  i found some answers to 
change the default lookup from 5 minutes to a time i can choose. with the 
following commands 


```

--node-monitor-period=2s (default 5s)
--node-monitor-grace-period=16s (default 40s)
--pod-eviction-timeout=30s (default 5m)
```

Because i try to do everything in yaml files to automate the entire process i 
searched for the kube controller manager yaml file and found it at 
/etc/kubernetes/manifest/kube-controller-manager.

When i add those parameters in "spec-containers-command" my entire cluster 
freezes. 
- i can not delete previous replication controllers 
- when i try to delete pods kubernetes won't create a new one. 
- when i create a new replication controller , the pods will not be created. 

I thought oké maybe it is because the new parameters are not set yet and i have 
to restart the kube controller manager service but there isn't any . I rebooted 
my vm and it still does not use the new parameters. 


With the command journalctl -xeu kubelet i got this error 
```
Feb 28 11:56:36 kubeMaster kubelet[448]: I0228 11:56:36.984643     448 
kuberuntime_manager.go:768] Back-off 5m0s restarting failed 
container=kube-controller-manager 
pod=kube-controller-manager-kubemaster_kube-system(e410a4a570860693452a4d5d29069eba)
Feb 28 11:56:36 kubeMaster kubelet[448]: E0228 11:56:36.984671     448 
pod_workers.go:186] Error syncing pod e410a4a570860693452a4d5d29069eba 
("kube-controller-manager-kubemaster_kube-system(e410a4a570860693452a4d5d29069eba)"),
 skipping: failed to "StartContainer
Feb 28 11:56:48 kubeMaster kubelet[448]: I0228 11:56:48.987153     448 
kuberuntime_manager.go:514] Container {Name:kube-controller-manager 
Image:gcr.io/google_containers/kube-controller-manager-amd64:v1.9.3 
Command:[kube-controller-manager --leader-elect=true --u

```

-- 
You received this message because you are subscribed to the Google Groups 
"Kubernetes user discussion and Q&A" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/kubernetes-users.
For more options, visit https://groups.google.com/d/optout.

Reply via email to