离线安装calico V3.28.1

1. 去github上面下载自己所需的calico离线包，项目地址：
   https://github.com/projectcalico/calico （下载3.28.1）

2. 假设要安装最新版本v3.28.1，首先可以下载这个版本的calico.yaml，具体命令是
   curl -O -L https://raw.githubusercontent.com/projectcalico/calico/v3.28.1/manifests/calico.yaml

3. 下载完成之后可以通过calico.yaml查看需要安装哪些离线包，命令是：
   $ cat calico.yaml | grep image
   image: docker.io/calico/cni:v3.28.1
   imagePullPolicy: IfNotPresent
   image: docker.io/calico/cni:v3.28.1
   imagePullPolicy: IfNotPresent
   image: docker.io/calico/node:v3.28.1
   imagePullPolicy: IfNotPresent
   image: docker.io/calico/node:v3.28.1
   imagePullPolicy: IfNotPresent
   image: docker.io/calico/kube-controllers:v3.28.1
   imagePullPolicy: IfNotPresent

4. 通过上述命令，查看到需要安装calico-cni.tar， calico-kube-controllers.tar 和 calico-node.tar三个包，然后需要将这三个包导入到k8s的命名空间中。

查询命名空间的命令是

$ sudo ctr namespace ls
NAME LABELS
default
k8s.io
moby
k8s的命名空间是k8s.io

5、使用导入命令将这三个包导入到k8s的命名空间中：
sudo ctr -n k8s.io images import calico-cni.tar
sudo ctr -n k8s.io images import calico-kube-controllers.tar
sudo ctr -n k8s.io images import calico-node.tar

6. 导入之后就可以apply calico.yaml 文件了
   kubectl apply -f calico.yaml

7. 导入之后查看calico的pod，发现calico和coredns已经起来了
   $ kubectl get pods -n kube-system
   NAME READY STATUS RESTARTS AGE
   calico-kube-controllers-8d76c5f9b-brv86 1/1 Running 0 22h
   calico-node-hxks2 1/1 Running 0 22h
   coredns-66f779496c-9hqsx 1/1 Running 0 23h
   coredns-66f779496c-rcc74 1/1 Running 0 23h
   etcd-kevin-pc 1/1 Running 4 (28m ago) 23h
   kube-apiserver-kevin-pc 1/1 Running 4 (28m ago) 23h
   kube-controller-manager-kevin-pc 1/1 Running 4 (28m ago) 23h
   kube-proxy-gglh4 1/1 Running 1 (28m ago) 23h
   kube-scheduler-kevin-pc 1/1 Running 4 (28m ago) 23h
   至此安装完成。

另：如果是node节点加入集群，先在master1上打印加入集群的命令 kubeadm token create --print-join-command，然后在node1上执行，执行后node节点会出现找不到pause3.9和cni插件镜像，手动导入即可。另外如果日志报错node节点使用IP地址冲突（root@master1:~# kubectl logs -n kube-system calico-node-xj9k2 -c calico-node 2025-06-10 03:04:45.341 [INFO][9] startup/startup.go 431: Early log level set to info 2025-06-10 03:04:45.342 [INFO][9] startup/utils.go 126: Using NODENAME environment for node name node1 2025-06-10 03:04:45.342 [INFO][9] startup/utils.go 138: Determined node name: node1 2025-06-10 03:04:45.342 [INFO][9] startup/startup.go 95: Starting node node1 with version v3.28.1 2025-06-10 03:04:45.342 [INFO][9] startup/startup.go 436: Checking datastore connection 2025-06-10 03:04:45.348 [INFO][9] startup/startup.go 460: Datastore connection verified 2025-06-10 03:04:45.349 [INFO][9] startup/startup.go 105: Datastore is ready 2025-06-10 03:04:45.353 [WARNING][9] startup/winutils.go 150: Neither --kubeconfig nor --master was specified. Using the inClusterConfig. This might not work. 2025-06-10 03:04:45.358 [INFO][9] startup/startup.go 489: Initialize BGP data 2025-06-10 03:04:45.358 [INFO][9] startup/autodetection_methods.go 103: Using autodetected IPv4 address on interface nerdctl0: 10.4.0.1/24 2025-06-10 03:04:45.358 [INFO][9] startup/startup.go 565: Node IPv4 changed, will check for conflicts 2025-06-10 03:04:45.360 [WARNING][9] startup/startup.go 1002: Calico node 'master1' is already using the IPv4 address 10.4.0.1. 2025-06-10 03:04:45.360 [INFO][9] startup/startup.go 395: Clearing out-of-date IPv4 address from this node IP="10.4.0.1/24" 2025-06-10 03:04:45.365 [WARNING][9] startup/utils.go 48: Terminating Calico node failed to start root@master1:~#） 因node1 的物理网卡 eth0 的 IP 地址是 10.0.12.4/22，而 Calico 错误地选择了 10.4.0.1/24（来自 nerdctl0 接口）作为节点 IP，导致与 master1 冲突。

解决办法：在 master1 上编辑 Calico DaemonSet，明确指定使用 eth0 接口：
1、kubectl edit daemonset -n kube-system calico-node
2、添加以下内容
name: IP_AUTODETECTION_METHOD
value: "interface=eth0"

2、pod会自动重建，重建后查看日志显示
master1无法与node1的IP建立BGP连接

node1无法与master1的IP建立BGP连接，

3、在腾讯云后台，分别开放两台机器的179，4789端口，Pod ready!!!

查看node calico版本
1、kubectl get pods -n kube-system | grep calico
2、kubectl exec -it -n kube-system -- /bin/sh
3、在powershell里运行 calico-node -v