CCE provides the container management services based on the native kubernetes. To facilitate the users’ better use of CCE, we summarize the checklist of some best practices from three aspects of cluster, application and troubleshooting. We strongly recommend you to view the checklist before you use CCE or the services are online. To view the checklist can help you migrate the services to CCE, and reduce the risks of application abnormality or cluster re-construction due to improper use.

Cluster

Application

Troubleshooting

Does the container fail to start?

Generally, you can view the error messages by the following two methods:

1.kubectl describe podName 2.kubectl logs podName

If no obvious errors are found by the above methods, you can modify the start command of the container in YAML, for example, setting the start command as sleep 3600:

apiVersion: extensions/v1beta1 
kind: Deployment 
metadata: 
  name: nginx-deployment 
  labels: 
    app: nginx 
spec: 
  replicas: 1 
  selector: 
    matchLabels: 
      app: nginx 
  template: 
    metadata: 
      labels: 
        app: nginx 
    spec: 
      containers: 
      - name: nginx 
        image: hub.baidubce.com/cce/nginx-alpine-go:latest 
        command: ["/bin/sh", "-c", "sleep 3600"] 

After starting the service, enter the container by kubectl exec -it podName/bin/sh, and manually execute the start command to view the error information of services.

Does the creation of LoadBalancer Service fail ?

You can view the Events and troubleshoot the failure reasons by kubectl describe service serviceName. Generally, the reason is that the quotas of EIP and BLB exceed the limit, and you can send a work order to apply for increase of the quotas.

Note: The number of EIP instances which can be purchased by the users <=current number of existing BCC instances + current number of existing BLB instances+2

Does the container network access fail?

The container network access fails in the following conditions:

• Serivce EIP cannot access;
• ServiceName cannot access in the container;
• Service ClusterIP cannot access in the cluster;
• PodIP cannot access in the cluster;
• ...

The container network problems are different access problems of Service which are generally caused by PodIP blockage. First, check whether the ping can be unblocked for PodIP in the nodes and Pods respectively. If not, check two places:

• View the VPC routing table, and confirm whether any routing rules conflict with CCE;
• View the VPC security group and ensure whether any policies intercept the requests.

If none is confirmed, you can send a work order to contact the administrator for troubleshooting.

Note: If the ping can be directly unblocked for Service clusterIP, access should be made by ip:port; the ping can be unblocked for PodIP.