1: 删除NODE_ID主机变量，改成使用ansible的run_once运行单节点任务

2：在keepalived模板中添加24位子网掩码 3：添加等待master节点服务启动完成检测任务 4：添加kubectl create之前的检测任务，避免重复操作时报错
chen0031 · Mar 14, 2018 · 182b082 · 182b082
1 parent 92afc3b
commit 182b082
Show file tree

Hide file tree

Showing 9 changed files with 48 additions and 26 deletions.
diff --git a/example/hosts.allinone.example b/example/hosts.allinone.example
@@ -10,15 +10,15 @@
 192.168.1.1 NODE_IP="192.168.1.1"
 
 [kube-node]
-192.168.1.1 NODE_ID=node1 NODE_IP="192.168.1.1"
+192.168.1.1 NODE_IP="192.168.1.1"
 
 # 如果启用harbor，请配置后面harbor相关参数
 [harbor]
 #192.168.1.8 NODE_IP="192.168.1.8"
 
 # 预留组，后续添加node节点使用
 [new-node]
-#192.168.1.xx NODE_ID=node6 NODE_IP="192.168.1.xx"
+#192.168.1.xx NODE_IP="192.168.1.xx"
 
 [all:vars]
 # ---------集群主要参数---------------

diff --git a/example/hosts.m-masters.example b/example/hosts.m-masters.example
@@ -24,18 +24,18 @@ ROUTER_ID=57                    # 取值在0-255之间，区分多个instance的
 MASTER_PORT="8443"		# 设置 api-server VIP地址的服务端口
 
 [kube-node]
-192.168.1.2 NODE_ID=node1 NODE_IP="192.168.1.2"
-192.168.1.3 NODE_ID=node2 NODE_IP="192.168.1.3"
-192.168.1.4 NODE_ID=node3 NODE_IP="192.168.1.4"
+192.168.1.2 NODE_IP="192.168.1.2"
+192.168.1.3 NODE_IP="192.168.1.3"
+192.168.1.4 NODE_IP="192.168.1.4"
 
 # 如果启用harbor，请配置后面harbor相关参数
 [harbor]
 #192.168.1.8 NODE_IP="192.168.1.8"
 
 # 预留组，后续添加node节点使用
 [new-node]
-#192.168.1.xx NODE_ID=node6 NODE_IP="192.168.1.xx"
-#192.168.1.xx NODE_ID=node7 NODE_IP="192.168.1.xx"
+#192.168.1.xx NODE_IP="192.168.1.xx"
+#192.168.1.xx NODE_IP="192.168.1.xx"
 
 [all:vars]
 # ---------集群主要参数---------------

diff --git a/example/hosts.s-master.example b/example/hosts.s-master.example
@@ -12,17 +12,17 @@
 192.168.1.1 NODE_IP="192.168.1.1"
 
 [kube-node]
-192.168.1.1 NODE_ID=node1 NODE_IP="192.168.1.1"
-192.168.1.2 NODE_ID=node2 NODE_IP="192.168.1.2"
-192.168.1.3 NODE_ID=node3 NODE_IP="192.168.1.3"
+192.168.1.1 NODE_IP="192.168.1.1"
+192.168.1.2 NODE_IP="192.168.1.2"
+192.168.1.3 NODE_IP="192.168.1.3"
 
 # 如果启用harbor，请配置后面harbor相关参数
 [harbor]
 #192.168.1.8 NODE_IP="192.168.1.8"
 
 # 预留组，后续添加node节点使用
 [new-node]
-#192.168.1.xx NODE_ID=node6 NODE_IP="192.168.1.xx"
+#192.168.1.xx NODE_IP="192.168.1.xx"
 
 [all:vars]
 # ---------集群主要参数---------------

diff --git a/roles/calico/tasks/main.yml b/roles/calico/tasks/main.yml
@@ -27,11 +27,16 @@
 - name: 准备 calico rbac文件 
   template: src=calico-rbac.yaml.j2 dest=/root/local/kube-system/calico/calico-rbac.yaml
 
-# 只需单节点执行一次，重复执行的报错可以忽略
+- name: 获取所有已经创建的POD信息
+  command: "kubectl get pod --all-namespaces"
+  register: pod_info
+  run_once: true
+
+# 只需单节点执行一次
 - name: 运行 calico网络
   shell: "{{ bin_dir }}/kubectl create -f /root/local/kube-system/calico/ && sleep 15"
-  when: NODE_ID is defined and NODE_ID == "node1"
-  ignore_errors: true
+  run_once: true
+  when: '"calico" not in pod_info.stdout'
 
 # 删除原有cni配置
 - name: 删除默认cni配置

diff --git a/roles/flannel/tasks/main.yml b/roles/flannel/tasks/main.yml
@@ -16,11 +16,16 @@
 - name: 准备 flannel DaemonSet yaml文件
   template: src=kube-flannel.yaml.j2 dest=/root/local/kube-system/flannel/kube-flannel.yaml
 
-# 只需单节点执行一次，重复执行的报错可以忽略
+- name: 获取所有已经创建的POD信息
+  command: "kubectl get pod --all-namespaces"
+  register: pod_info
+  run_once: true
+
+# 只需单节点执行一次
 - name: 运行 flannel网络
   shell: "{{ bin_dir }}/kubectl create -f /root/local/kube-system/flannel/ && sleep 15"
-  when: NODE_ID is defined and NODE_ID == "node1"
-  ignore_errors: true
+  run_once: true
+  when: '"flannel" not in pod_info.stdout'
 
 # 删除原有cni配置
 - name: 删除默认cni配置 

diff --git a/roles/kube-master/tasks/main.yml b/roles/kube-master/tasks/main.yml
@@ -63,3 +63,11 @@
 
 - name: start-kube-scheduler
   shell: systemctl restart kube-scheduler
+
+- name: 以轮询的方式等待master服务启动完成
+  command: "kubectl get node"
+  register: result
+  until:    result.rc == 0
+  retries:  5
+  delay: 6
+  run_once: True
diff --git a/roles/kube-node/tasks/main.yml b/roles/kube-node/tasks/main.yml
@@ -14,16 +14,20 @@
   - bridge
   - host-local
   - loopback
+
+- name: get clusterrolebinding info
+  command: "kubectl get clusterrolebinding --all-namespaces"
+  register: clusterrolebinding_info
+  run_once: true
 
 ##----------kubelet 配置部分--------------
 # kubelet 启动时向 kube-apiserver 发送 TLS bootstrapping 请求，需要绑定该角色
-# 只需单节点执行一次，重复执行的报错可以忽略
-# 增加15s等待kube-apiserver正常工作
+# 只需单节点执行一次
 - name: kubelet-bootstrap-setting
-  shell: "sleep 15 && {{ bin_dir }}/kubectl create clusterrolebinding kubelet-bootstrap \
+  shell: "{{ bin_dir }}/kubectl create clusterrolebinding kubelet-bootstrap \
         --clusterrole=system:node-bootstrapper --user=kubelet-bootstrap"
-  when: NODE_ID is defined and NODE_ID == "node1"
-  ignore_errors: true
+  run_once: True
+  when: '"kubelet-bootstrap" not in clusterrolebinding_info.stdout'
 
 #创建bootstrap.kubeconfig配置文件
 - name: 设置集群参数
@@ -59,8 +63,8 @@
   tags: kubelet 
 
 - name: approve-kubelet-csr
-  shell: "sleep 15 && {{ bin_dir }}/kubectl get csr|grep 'Pending' | awk 'NR>0{print $1}'| xargs {{ bin_dir }}/kubectl certificate approve"
-  when: NODE_ID is defined and NODE_ID == "node1"
+  shell: "{{ bin_dir }}/kubectl get csr|grep 'Pending' | awk 'NR>0{print $1}'| xargs {{ bin_dir }}/kubectl certificate approve"
+  run_once: true
   ignore_errors: true
 
 ##-------kube-proxy部分----------------

diff --git a/roles/lb/templates/keepalived-backup.conf.j2 b/roles/lb/templates/keepalived-backup.conf.j2
@@ -10,6 +10,6 @@ vrrp_instance VI-kube-master {
     virtual_router_id {{ ROUTER_ID }} 
     advert_int 3
     virtual_ipaddress {
-        {{ MASTER_IP }}
+        {{ MASTER_IP }}/24
     }
 }
diff --git a/roles/lb/templates/keepalived-master.conf.j2 b/roles/lb/templates/keepalived-master.conf.j2
@@ -19,6 +19,6 @@ vrrp_instance VI-kube-master {
         check-haproxy
     }
     virtual_ipaddress {
-        {{ MASTER_IP }}
+        {{ MASTER_IP }}/24
     }
 }