CentOS7 AltArch aarch64环境kubeasz flannel : 轮询等待flannel 运行，视下载镜像速度而定

亦凉 2022-12-03 01:32 140阅读 0赞

故障

TASK [flannel : 检查是否已下载离线flannel镜像] ******************************************************************************************************************
    changed: [192.168.111.193]
    
    TASK [flannel : 尝试推送离线docker 镜像（若执行失败，可忽略）] **********************************************************************************************************
    ok: [192.168.111.193] => (item=pause.tar)
    ok: [192.168.111.193] => (item=flannel_v0.12.0.tar)
    
    TASK [flannel : 获取flannel离线镜像推送情况] *******************************************************************************************************************
    changed: [192.168.111.193]
    
    TASK [flannel : 导入 flannel的离线镜像（若执行失败，可忽略）] **********************************************************************************************************
    changed: [192.168.111.193] => (item=pause.tar)
    changed: [192.168.111.193] => (item=flannel_v0.12.0.tar)
    
    TASK [flannel : 导入 flannel的离线镜像（若执行失败，可忽略）] **********************************************************************************************************
    
    TASK [flannel : 运行 flannel网络] ************************************************************************************************************************
    changed: [192.168.111.193]
    
    TASK [flannel : 删除默认cni配置] ***************************************************************************************************************************
    changed: [192.168.111.193]
    
    TASK [flannel : 轮询等待flannel 运行，视下载镜像速度而定] ************************************************************************************************************
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (15 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (14 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (13 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (12 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (11 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (10 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (9 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (8 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (7 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (6 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (5 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (4 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (3 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (2 retries left).
    FAILED - RETRYING: 轮询等待flannel 运行，视下载镜像速度而定 (1 retries left).
    fatal: [192.168.111.193]: FAILED! => {"attempts": 15, "changed": true, "cmd": "/opt/kube/bin/kubectl get pod -n kube-system -o wide|grep 'flannel'|grep ' 192.168.111.193 '|awk '{print $3}'", "delta": "0:00:00.175250", "end": "2020-09-01 09:59:32.723445", "rc": 0, "start": "2020-09-01 09:59:32.548195", "stderr": "No resources found in kube-system namespace.", "stderr_lines": ["No resources found in kube-system namespace."], "stdout": "", "stdout_lines": []}
    ...ignoring

对应playbook内容

# 【可选】推送离线docker 镜像，可以忽略执行错误
    - block:
        - name: 检查是否已下载离线flannel镜像
          command: "ls {
       { base_dir }}/down"
          register: download_info
          connection: local
          run_once: true
    
        - name: 尝试推送离线docker 镜像（若执行失败，可忽略）
          copy: src={
       { base_dir }}/down/{
       { item }} dest=/opt/kube/images/{
       { item }}
          when: 'item in download_info.stdout'
          with_items:
          - "pause.tar"
          - "{
       { flannel_offline }}"
          ignore_errors: true
    
        - name: 获取flannel离线镜像推送情况
          command: "ls /opt/kube/images"
          register: image_info
    
        # 如果目录下有离线镜像，就把它导入到node节点上
        - name: 导入 flannel的离线镜像（若执行失败，可忽略）
          shell: "{
       { bin_dir }}/docker load -i /opt/kube/images/{
       { item }}"
          with_items:
          - "pause.tar"
          - "{
       { flannel_offline }}"
          ignore_errors: true
          when: "item in image_info.stdout and CONTAINER_RUNTIME == 'docker'"
    
        - name: 导入 flannel的离线镜像（若执行失败，可忽略）
          shell: "{
       { bin_dir }}/ctr -n=k8s.io images import /opt/kube/images/{
       { item }}"
          with_items:
          - "pause.tar"
          - "{
       { flannel_offline }}"
          ignore_errors: true
          when: "item in image_info.stdout and CONTAINER_RUNTIME == 'containerd'"
    
    # 只需单节点执行一次
    - name: 运行 flannel网络
      shell: "{
       { bin_dir }}/kubectl apply -f /opt/kube/kube-system/flannel.yaml"
      run_once: true
    
    # 删除原有cni配置
    - name: 删除默认cni配置
      file: path=/etc/cni/net.d/10-default.conf state=absent
    
    # 等待网络插件部署成功，视下载镜像速度而定
    - name: 轮询等待flannel 运行，视下载镜像速度而定
      shell: "{
       { bin_dir }}/kubectl get pod -n kube-system -o wide|grep 'flannel'|grep ' {
       { inventory_hostname }} '|awk '{print $3}'"
      register: pod_status
      until: pod_status.stdout == "Running"
      retries: 15
      delay: 8
      ignore_errors: true

解决：

由于当前处于内核aarch64架构下，修改

/etc/ansible/roles/flannel/templates/kube-flannel.yaml.j2

beta.kubernetes.io/arch 为arm64

spec:
          affinity:
            nodeAffinity:
              requiredDuringSchedulingIgnoredDuringExecution:
                nodeSelectorTerms:
                  - matchExpressions:
                      - key: beta.kubernetes.io/os
                        operator: In
                        values:
                          - linux
                      - key: beta.kubernetes.io/arch
                        operator: In
                        values:
                          - arm64

重新部署