Openshift集群升级3.6至3.7

更新方法与策略：自动升级方法

更新openshift-ansible的脚本代码

git pull 
git checkout -b release-3.7

安装需要的包

ansible all -m yum -a 'name=pyOpenSSL state=present'

关闭Service Catalog和Template Service Broker。
3.7 以上版本默认会启用 Service Catalog 和 Template Service Broker

# /etc/ansible/hosts
...
openshift_enable_service_catalog=false
template_service_broker_install=false
...

关闭etcd目录的selinux检查
关闭任务Check selinux label of '{{ etcd_data_dir }}'与Make sure the '{{ etcd_data_dir }}' has the proper label

# roles/etcd/tasks/backup/backup.yml
...
- name: Detecting Atomic Host Operating System
  stat:
    path: /run/ostree-booted
  register: l_ostree_booted

#- name: Check selinux label of '{{ etcd_data_dir }}'
#  command: >
#    stat -c '%C' {{ etcd_data_dir }}
#  register: l_etcd_selinux_labels
#
#- debug:
#    msg: "{{ l_etcd_selinux_labels }}"
#
#- name: Make sure the '{{ etcd_data_dir }}' has the proper label
#  command: >
#    chcon -t svirt_sandbox_file_t  "{{ etcd_data_dir }}"
#  when:
#  - l_etcd_selinux_labels.rc == 0
#  - "'svirt_sandbox_file_t' not in l_etcd_selinux_labels.stdout"

- name: Generate etcd backup
  command: >
    {{ r_etcd_common_etcdctl_command }} backup --data-dir={{ l_etcd_incontainer_data_dir }}
    --backup-dir={{ l_etcd_incontainer_backup_dir }}
...

执行更新

ansible-playbook -i inventory-hosts playbooks/byo/openshift-cluster/upgrades/v3_7/upgrade.yml

查看当前oc版本及各node版本

oc version
oc get node

升级后发现的问题

RBD的storageclass异常

# 解决方法，storageclass.parameters中添加 imageFormat: "1"
apiVersion: storage.k8s.io/v1
metadata:
   name: ceph-rbd-sc
Parameters:
  adminId: admin
  adminSecretName: ceph-secret
  adminSecretNamespace: kube-system
  fsType: ext4
  imageFormat: "1"
  monitors: 192.168.1.3:6789
  pool: rbd
  userId: admin
  userSecretName: ceph-secret
Provisioner: kubernetes.io/rbd