Installation and upgrade

Review frequently encountered installation and upgrade errors.

• Installation hangs or fails
  
• The etcd component fails to start
  
• Failure to connect through ssh
  
• Failure to create containers
  
• Kubelet container fails to start
  
• The nginx ingress controller does not start
  
• Pods fail to initialize
  
• Installation fails when firewalld is enabled
  
• Ingress controller reported: epoll_create() failed
  
• Pods fail with CrashLoopBackOff
  
• manifest-tool command not found
  
• Disable custom-metrics-adapter service when disabling monitoring service during IBM Cloud Private installation
  
• Out-of-memory errors on Power systems with many vCPUs
  
• etcd maintenance
  
• Large-scale clusters (1000 worker node)
  
• Installation failed waiting for Tiller to start
  
• NVIDIA GPU driver fails to initialize
  
• Log flooding when using systemd as cgroup driver
  
• Transferring roles from the master node
  
• Installation errors with SELinux enabled
  
• Docker upgrade does not support rollback
  
• Docker options are missing after upgrade
  
• Pods that use the host network fail to start
  
• Kubernetes CLI cannot uncordon the master node
  
• Fix pack cannot roll back when GlusterFS is enabled
  
• Logging pods show CrashLoopBackOff status in Linux on Power (ppc64le) clusters