Olares/cli/pkg/terminus/tasks.go

package terminus

import (
	"context"
	"crypto/tls"
	"encoding/json"
	"fmt"
	"io/fs"
	"io/ioutil"
	"net"
	"net/http"
	"net/url"
	"os"
	"os/exec"
	"path"
	"path/filepath"
	"strings"
	"sync"
	"time"

	"github.com/beclab/Olares/cli/version"

	"github.com/beclab/Olares/cli/pkg/storage"
	corev1 "k8s.io/api/core/v1"
	kerrors "k8s.io/apimachinery/pkg/api/errors"
	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
	"k8s.io/apimachinery/pkg/fields"
	"k8s.io/apimachinery/pkg/watch"
	"k8s.io/client-go/kubernetes"
	ctrl "sigs.k8s.io/controller-runtime"
	"sigs.k8s.io/controller-runtime/pkg/client"

	bootstraptpl "github.com/beclab/Olares/cli/pkg/bootstrap/os/templates"
	"github.com/beclab/Olares/cli/pkg/core/action"
	"github.com/beclab/Olares/cli/pkg/terminus/templates"

	"github.com/beclab/Olares/cli/pkg/common"
	cc "github.com/beclab/Olares/cli/pkg/core/common"
	"github.com/beclab/Olares/cli/pkg/core/connector"
	"github.com/beclab/Olares/cli/pkg/core/logger"
	"github.com/beclab/Olares/cli/pkg/core/util"
	"github.com/beclab/Olares/cli/pkg/files"
	"github.com/beclab/Olares/cli/pkg/utils"

	"github.com/pkg/errors"
)

type GetOlaresVersion struct {
}

func (t *GetOlaresVersion) Execute() (string, error) {
	var kubectlpath, err = util.GetCommand(common.CommandKubectl)
	if err != nil {
		return "", fmt.Errorf("kubectl not found, Olares might not be installed.")
	}

	var ctx, cancel = context.WithTimeout(context.Background(), 10*time.Second)
	defer cancel()

	cmd := exec.CommandContext(ctx, "/bin/sh", "-c", fmt.Sprintf("%s get terminus -o jsonpath='{.items[*].spec.version}'", kubectlpath))
	cmd.WaitDelay = 3 * time.Second
	output, err := cmd.CombinedOutput()
	if err != nil {
		return "", errors.Wrap(errors.WithStack(err), "get Olares version failed")
	}

	if version := string(output); version == "" {
		return "", fmt.Errorf("Olares might not be installed.")
	} else {
		return version, nil
	}
}

type CheckKeyPodsRunning struct {
	common.KubeAction
	Node string
}

func (t *CheckKeyPodsRunning) Execute(runtime connector.Runtime) error {
	kubeConfig, err := ctrl.GetConfig()
	if err != nil {
		return errors.Wrap(err, "failed to load kubeconfig")
	}
	kubeClient, err := kubernetes.NewForConfig(kubeConfig)
	if err != nil {
		return errors.Wrap(err, "failed to create kube client")
	}
	pods, err := kubeClient.CoreV1().Pods(corev1.NamespaceAll).List(context.Background(), metav1.ListOptions{})
	if err != nil {
		return errors.Wrap(err, "failed to list pods")
	}
	for _, pod := range pods.Items {
		if t.Node != "" && pod.Spec.NodeName != t.Node {
			logger.Debugf("skipping pod %s that's not on node %s", pod.Name, t.Node)
			continue
		}
		if !strings.HasPrefix(pod.Namespace, "user-") && !strings.HasPrefix(pod.Namespace, "os-") {
			continue
		}
		if err := utils.AssertPodReady(&pod); err != nil {
			return err
		}
	}
	return nil
}

type CheckPodsRunning struct {
	common.KubeAction
	labels map[string][]string
}

func (c *CheckPodsRunning) Execute(_ connector.Runtime) error {
	if c.labels == nil {
		return nil
	}

	var ctx, cancel = context.WithTimeout(context.Background(), 3*time.Minute)
	defer cancel()

	kubeConfig, err := ctrl.GetConfig()
	if err != nil {
		return errors.Wrap(err, "failed to load kubeconfig")
	}
	kubeClient, err := kubernetes.NewForConfig(kubeConfig)
	if err != nil {
		return errors.Wrap(err, "failed to create kube client")
	}

	for ns, labels := range c.labels {
		for _, label := range labels {
			podList, err := kubeClient.CoreV1().Pods(ns).List(ctx, metav1.ListOptions{LabelSelector: label})
			if err != nil {
				return fmt.Errorf("pod status invalid, namespace: %s, label: %s, waiting ...", ns, label)
			}

			if podList == nil || len(podList.Items) == 0 {
				return fmt.Errorf("no pod found, namespace: %s, label: %s, waiting ...", ns, label)
			}

			for i := range podList.Items {
				pod := &podList.Items[i]
				if err := utils.AssertPodReady(pod); err != nil {
					return err
				}
			}
		}
	}

	return nil
}

type Download struct {
	common.KubeAction
	Version     string
	BaseDir     string
	CDNService  string
	UrlOverride string
	ReleaseID   string
}

func (t *Download) Execute(runtime connector.Runtime) error {
	if t.UrlOverride == "" && t.KubeConf.Arg.OlaresVersion == "" {
		return errors.New("unknown version to download")
	}

	var stem string
	if t.ReleaseID != "" {
		stem = fmt.Sprintf("install-wizard-v%s.%s", t.Version, t.ReleaseID)
	} else {
		stem = fmt.Sprintf("install-wizard-v%s", t.Version)
	}
	wizard := &files.KubeBinary{
		ID:       "install-wizard",
		Type:     files.WIZARD,
		BaseDir:  filepath.Join(runtime.GetBaseDir(), "versions", fmt.Sprintf("v%s", t.Version)),
		FileName: fmt.Sprintf("%s.tar.gz", stem),
	}

	if t.UrlOverride == "" {
		md5URL, _ := url.JoinPath(wizard.GetDownloadMirrors(t.CDNService), version.VENDOR_REPO_PATH, fmt.Sprintf("%s.md5sum.txt", stem))
		var fetchMd5 = fmt.Sprintf("curl -sSfL %s |awk '{print $1}'", md5URL)
		md5sum, err := runtime.GetRunner().Cmd(fetchMd5, false, false)
		if err != nil {
			return errors.New("get md5sum failed")
		}
		wizard.Url, _ = url.JoinPath(wizard.GetDownloadMirrors(t.CDNService), version.VENDOR_REPO_PATH, fmt.Sprintf("%s.tar.gz", stem))
		wizard.CheckMd5Sum = true
		wizard.Md5sum = md5sum
	} else {
		wizard.CheckMd5Sum = false
		wizard.Url = t.UrlOverride
		util.RemoveFile(wizard.Path())
	}

	if err := wizard.CreateBaseDir(); err != nil {
		return errors.Wrapf(errors.WithStack(err), "create file %s base dir failed", wizard.FileName)
	}

	var exists = util.IsExist(wizard.Path())
	if exists {
		if err := wizard.Md5Check(); err == nil {
			// file exists, re-unpack
			return util.Untar(wizard.Path(), wizard.BaseDir)
		} else {
			logger.Error(err)
		}

		util.RemoveFile(wizard.Path())
	}

	logger.Infof("%s downloading %s %s ...", common.LocalHost, wizard.ID, wizard.Version)
	if err := wizard.Download(); err != nil {
		return fmt.Errorf("Failed to download %s binary: %s error: %w ", wizard.ID, wizard.Url, err)
	}

	return util.Untar(wizard.Path(), wizard.BaseDir)
}

func copyWizard(wizardPath string, np string, runtime connector.Runtime) {
	if util.IsExist(np) {
		util.RemoveDir(np)
	} else {
		// util.Mkdir(np)
	}
	_, err := runtime.GetRunner().Cmd(fmt.Sprintf("cp -a %s %s", wizardPath, np), false, false)
	if err != nil {
		logger.Errorf("copy -a %s to %s failed", wizardPath, np)
	}
}

type PrepareFinished struct {
	common.KubeAction
}

func (t *PrepareFinished) Execute(runtime connector.Runtime) error {
	var preparedFile = filepath.Join(runtime.GetBaseDir(), common.TerminusStateFilePrepared)
	return util.WriteFile(preparedFile, []byte(t.KubeConf.Arg.OlaresVersion), cc.FileMode0644)
	// if _, err := runtime.GetRunner().Cmd(fmt.Sprintf("touch %s", preparedFile), false, true); err != nil {
	// 	return err
	// }
	// return nil
}

type WriteReleaseFile struct {
	common.KubeAction
	WithoutName bool
}

func (t *WriteReleaseFile) Execute(runtime connector.Runtime) error {
	if util.IsExist(common.OlaresReleaseFile) {
		logger.Debugf("found existing release file: %s, overriding ...", common.OlaresReleaseFile)
	}
	return t.KubeConf.Arg.SaveReleaseInfo(t.WithoutName)
}

type RemoveReleaseFile struct {
	common.KubeAction
}

func (t *RemoveReleaseFile) Execute(runtime connector.Runtime) error {
	err := os.Remove(common.OlaresReleaseFile)
	if err != nil && !os.IsNotExist(err) {
		return err
	}
	return nil
}

type CheckPrepared struct {
	common.KubeAction
	Force bool
}

func (t *CheckPrepared) Execute(runtime connector.Runtime) error {
	var preparedPath = filepath.Join(runtime.GetBaseDir(), common.TerminusStateFilePrepared)

	if utils.IsExist(preparedPath) {
		t.PipelineCache.Set(common.CachePreparedState, true)
	} else if t.Force {
		return errors.New("Olares dependencies is not prepared, refuse to continue")
	}

	return nil
}

type CheckInstalled struct {
	common.KubeAction
	Force bool
}

func (t *CheckInstalled) Execute(runtime connector.Runtime) error {
	var installedPath = filepath.Join(runtime.GetBaseDir(), common.TerminusStateFileInstalled)

	if utils.IsExist(installedPath) {
		t.PipelineCache.Set(common.CacheInstalledState, true)
	} else if t.Force {
		return errors.New("Olares is not installed, refuse to continue")
	}

	return nil
}

type InstallFinished struct {
	common.KubeAction
}

func (t *InstallFinished) Execute(runtime connector.Runtime) error {
	var content = fmt.Sprintf("%s %s", t.KubeConf.Arg.OlaresVersion, t.KubeConf.Arg.Kubetype)
	var phaseState = path.Join(runtime.GetBaseDir(), common.TerminusStateFileInstalled)
	if err := util.WriteFile(phaseState, []byte(content), cc.FileMode0644); err != nil {
		return err
	}
	return nil
}

type DeleteUpgradeFiles struct {
	common.KubeAction
}

func (d *DeleteUpgradeFiles) Execute(runtime connector.Runtime) error {
	baseDir := runtime.GetBaseDir()

	files, err := os.ReadDir(baseDir)
	if err != nil {
		return errors.Wrapf(err, "failed to read directory %s", baseDir)
	}

	for _, file := range files {
		if strings.HasPrefix(file.Name(), "upgrade.") {
			filePath := path.Join(baseDir, file.Name())
			if err := os.RemoveAll(filePath); err != nil && !os.IsNotExist(err) {
				logger.Warnf("failed to delete %s: %v", filePath, err)
			}
		}
	}

	return nil
}

type DeleteWizardFiles struct {
	common.KubeAction
}

func (d *DeleteWizardFiles) Execute(runtime connector.Runtime) error {
	var locations = []string{
		path.Join(runtime.GetInstallerDir(), cc.BuildFilesCacheDir),
		runtime.GetWorkDir(),
		path.Join(runtime.GetInstallerDir(), cc.LogsDir, cc.InstallLogFile),
	}

	for _, location := range locations {
		if util.IsExist(location) {
			runtime.GetRunner().SudoCmd(fmt.Sprintf("rm -rf %s", location), false, true)
		}
	}
	return nil
}

type SystemctlCommand struct {
	common.KubeAction
	UnitNames           []string
	Command             string
	DaemonReloadPreExec bool
}

func (a *SystemctlCommand) Execute(runtime connector.Runtime) error {
	if a.DaemonReloadPreExec {
		if _, err := runtime.GetRunner().SudoCmd("systemctl daemon-reload", false, true); err != nil {
			return errors.Wrap(errors.WithStack(err), "systemctl reload failed")
		}
	}
	for _, unitName := range a.UnitNames {
		cmd := fmt.Sprintf("systemctl %s %s", a.Command, unitName)
		if _, err := runtime.GetRunner().SudoCmd(cmd, false, true); err != nil {
			return errors.Wrapf(err, "failed to execute command: %s", cmd)
		}
	}

	return nil
}

// PrepareFilesForETCDIPChange simply copies the current CA cert and key
// to the work directory of Installer,
// so that after the regeneration is done,
// all other certs and keys is replaced with a new one
// but issued by the same CA
type PrepareFilesForETCDIPChange struct {
	common.KubeAction
}

func (a *PrepareFilesForETCDIPChange) Execute(runtime connector.Runtime) error {
	srcCertsDir := "/etc/ssl/etcd/ssl"
	dstCertsDir := filepath.Join(runtime.GetWorkDir(), "/pki/etcd")
	if err := util.RemoveDir(dstCertsDir); err != nil {
		return errors.Wrap(err, "failed to clear work directory for etcd certs")
	}
	if err := util.Mkdir(dstCertsDir); err != nil {
		return errors.Wrap(err, "failed to create work directory for etcd certs")
	}
	return filepath.WalkDir(srcCertsDir, func(path string, d fs.DirEntry, err error) error {
		if err != nil {
			return err
		}
		if d.IsDir() {
			return nil
		}
		if strings.HasPrefix(filepath.Base(path), "ca") {
			if err := util.CopyFileToDir(path, dstCertsDir); err != nil {
				return err
			}
		}
		return nil
	})
}

type PrepareFilesForK8sIPChange struct {
	common.KubeAction
}

func (a *PrepareFilesForK8sIPChange) Execute(runtime connector.Runtime) error {
	k8sConfDir := "/etc/kubernetes"
	return filepath.WalkDir(k8sConfDir, func(path string, d fs.DirEntry, err error) error {
		if err != nil {
			return err
		}
		if d.IsDir() {
			return nil
		}
		if filepath.Ext(path) == ".conf" {
			logger.Debugf("removing %s", path)
			return util.RemoveFile(path)
		}
		if filepath.Base(path) == "kubeadm-config.yaml" {
			logger.Debugf("removing %s", path)
			return util.RemoveFile(path)
		}
		if strings.Contains(path, "pki") && !strings.HasPrefix(filepath.Base(path), "ca") {
			logger.Debugf("removing %s", path)
			return util.RemoveFile(path)
		}
		return nil
	})
}

type RegenerateFilesForK8s struct {
	common.KubeAction
}

func (a *RegenerateFilesForK8s) Execute(runtime connector.Runtime) error {
	initCmd := "/usr/local/bin/kubeadm init --config=/etc/kubernetes/kubeadm-config.yaml --skip-phases=preflight,mark-control-plane,bootstrap-token,addon,show-join-command"

	if _, err := runtime.GetRunner().SudoCmd(initCmd, false, false); err != nil {
		return err
	}
	return nil
}

type DeleteAllPods struct {
	common.KubeAction
	Node string
}

func (a *DeleteAllPods) Execute(runtime connector.Runtime) error {
	kubectlpath, err := util.GetCommand(common.CommandKubectl)
	if err != nil {
		return fmt.Errorf("kubectl not found")
	}
	var cmd = fmt.Sprintf("%s delete pod --all-namespaces -l tier!=control-plane", kubectlpath)
	if a.Node != "" {
		cmd += fmt.Sprintf(" --field-selector spec.nodeName=%s", a.Node)
	}
	if _, err := runtime.GetRunner().SudoCmd(cmd, false, true); err != nil {
		return err
	}

	return nil
}

type DeletePodsUsingHostIP struct {
	common.KubeAction
}

func (a *DeletePodsUsingHostIP) Execute(runtime connector.Runtime) error {
	kubeConfig, err := ctrl.GetConfig()
	if err != nil {
		return errors.Wrap(err, "failed to load kubeconfig")
	}
	kubeClient, err := kubernetes.NewForConfig(kubeConfig)
	if err != nil {
		return errors.Wrap(err, "failed to create kube client")
	}

	targetPods, err := getPodsUsingHostIP(kubeClient)
	if err != nil {
		return errors.Wrap(err, "failed to get pods using host IP")
	}
	var waitRecreationPodsCount int
	for _, pod := range targetPods {
		logger.Infof("restarting pod %s/%s that's using host IP", pod.Namespace, pod.Name)
		err = kubeClient.CoreV1().Pods(pod.Namespace).Delete(context.Background(), pod.Name, metav1.DeleteOptions{})
		if err != nil && !kerrors.IsNotFound(err) {
			return errors.Wrap(err, "failed to delete pod")
		}

		// pods not created by any owner resource
		// may not be recreated immediately and should not be waited
		if len(pod.OwnerReferences) > 0 {
			waitRecreationPodsCount++
		}
	}
	a.PipelineCache.Set(common.CacheCountPodsWaitForRecreation, waitRecreationPodsCount)

	// try our best to wait for the pods to be actually deleted
	// to avoid the next module getting the pods with a still running phase
	err = waitForPodsToBeGone(kubeClient, targetPods, 3*time.Minute)
	if err != nil {
		logger.Warnf("failed to wait for pods to be gone: %v, will delay and skip", err)
		time.Sleep(60 * time.Second)
		return nil
	}

	return nil
}

type WaitForPodsUsingHostIPRecreate struct {
	common.KubeAction
}

func (a *WaitForPodsUsingHostIPRecreate) Execute(runtime connector.Runtime) error {
	count, ok := a.PipelineCache.GetMustInt(common.CacheCountPodsWaitForRecreation)
	if !ok {
		return errors.New("failed to get the count of pods using host IP")
	}
	kubeConfig, err := ctrl.GetConfig()
	if err != nil {
		return errors.Wrap(err, "failed to load kubeconfig")
	}
	kubeClient, err := kubernetes.NewForConfig(kubeConfig)
	if err != nil {
		return errors.Wrap(err, "failed to create kube client")
	}

	targetPods, err := getPodsUsingHostIP(kubeClient)
	if err != nil {
		return errors.Wrap(err, "failed to get pods using host IP")
	}
	if len(targetPods) < count {
		return errors.New("waiting for pods using host IP to be recreated")
	}
	return nil
}

func getPodsUsingHostIP(kubeClient kubernetes.Interface) ([]*corev1.Pod, error) {
	pods, err := kubeClient.CoreV1().Pods(corev1.NamespaceAll).List(context.Background(), metav1.ListOptions{})
	if err != nil || len(pods.Items) == 0 {
		return nil, errors.Wrap(err, "failed to list pods")
	}
	var targetPods []*corev1.Pod
	for _, pod := range pods.Items {
		if podIsUsingHostIP(&pod) {
			targetPods = append(targetPods, &pod)
		}
	}
	return targetPods, nil
}

func podIsUsingHostIP(pod *corev1.Pod) bool {
	if pod == nil {
		return false
	}
	if pod.Spec.HostNetwork == true {
		return true
	}

	// coredns also counts as a pod using host ip
	// because the local network's gateway is often used as its upstream DNS server,
	// and it needs to be updated in case of a cidr change
	if pod.Namespace == "kube-system" && pod.Labels["k8s-app"] == "kube-dns" {
		return true
	}
	var allContainers []corev1.Container
	allContainers = append(allContainers, pod.Spec.Containers...)
	allContainers = append(allContainers, pod.Spec.InitContainers...)
	for _, container := range allContainers {
		for _, env := range container.Env {
			if env.ValueFrom != nil && env.ValueFrom.FieldRef != nil && env.ValueFrom.FieldRef.FieldPath == "status.hostIP" {
				return true
			}
		}
	}
	return false
}

func waitForPodsToBeGone(kubeClient *kubernetes.Clientset, pods []*corev1.Pod, timeout time.Duration) error {
	for _, pod := range pods {
		pod, err := kubeClient.CoreV1().Pods(pod.Namespace).Get(context.Background(), pod.Name, metav1.GetOptions{})
		if kerrors.IsNotFound(err) {
			continue
		}
		if err != nil {
			return errors.Wrap(err, "failed to check if pod exists")
		}
		watchOptions := metav1.ListOptions{
			FieldSelector:   fields.OneTermEqualSelector("metadata.name", pod.Name).String(),
			ResourceVersion: pod.GetResourceVersion(),
		}
		ctx, cancel := context.WithTimeout(context.Background(), timeout)
		defer cancel()
		watcher, err := kubeClient.CoreV1().Pods(pod.Namespace).Watch(ctx, watchOptions)
		if err != nil {
			return errors.Wrapf(err, "failed to watch pod %s for deletion", client.ObjectKeyFromObject(pod).String())
		}
		defer watcher.Stop()
		watchChan := watcher.ResultChan()
		for {
			event, ok := <-watchChan
			if !ok || event.Type == watch.Error {
				if !ok {
					err = errors.New("watch channel closed unexpectedly")
				} else {
					err = kerrors.FromObject(event.Object)
				}
				return errors.Wrapf(err, "failed to watch pod %s for deletion", client.ObjectKeyFromObject(pod).String())
			}
			if event.Type == watch.Deleted {
				break
			}
		}
	}
	return nil
}

type UpdateKubeKeyHosts struct {
	common.KubeAction
}

func (a *UpdateKubeKeyHosts) Execute(runtime connector.Runtime) error {

	scriptPath := filepath.Join(runtime.GetWorkDir(), "change-ip-scripts", "update-kubekey-hosts.sh")
	tplAction := &action.Template{
		Name:     "GenerateHostsUpdateScript",
		Template: templates.UpdateKKHostsScriptTmpl,
		Dst:      scriptPath,
		Data: util.Data{
			"Hosts": bootstraptpl.GenerateHosts(runtime, a.KubeConf),
		},
	}
	if err := tplAction.Execute(runtime); err != nil {
		return errors.Wrapf(err, fmt.Sprintf("failed to generate update hosts script: %s", scriptPath))
	}
	if _, err := runtime.GetRunner().SudoCmd(fmt.Sprintf("chmod +x %s", scriptPath), false, false); err != nil {
		return errors.Wrap(errors.WithStack(err), "failed to chmod +x update hosts script")
	}

	if _, err := runtime.GetRunner().SudoCmd(scriptPath, false, true); err != nil {
		return errors.Wrap(errors.WithStack(err), "failed to run update hosts script")
	}
	return nil
}

type CheckTerminusStateInHost struct {
	common.KubeAction
}

func (a *CheckTerminusStateInHost) Execute(runtime connector.Runtime) error {
	si := runtime.GetSystemInfo()
	var kubectlCMD string
	var kubectlCMDDefaultArgs []string
	var err error
	ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
	defer cancel()
	if si.IsDarwin() {
		minikube, err := util.GetCommand(common.CommandMinikube)
		if err != nil {
			return errors.Wrap(err, "failed to get minikube command")
		}
		_, _, err = util.Exec(ctx, fmt.Sprintf("%s update-context -p %s", minikube, a.KubeConf.Arg.MinikubeProfile), false, true)
		if err != nil {
			fmt.Printf("failed to update minikube context to %s, is it up and running?", a.KubeConf.Arg.MinikubeProfile)
			os.Exit(1)
		}
		kubectlCMD, err = util.GetCommand(common.CommandKubectl)
		if err != nil {
			return errors.Wrap(errors.WithStack(err), "kubectl not found")
		}
	} else if si.IsWindows() {
		kubectlCMD = "cmd"
		kubectlCMDDefaultArgs = []string{"/C", "wsl", "-d", a.KubeConf.Arg.WSLDistribution, "-u", "root", common.CommandKubectl}
	}

	getTerminusArgs := []string{"get", "terminus"}

	getTerminusCMD := exec.CommandContext(ctx, kubectlCMD, append(kubectlCMDDefaultArgs, getTerminusArgs...)...)
	getTerminusCMD.WaitDelay = 3 * time.Second
	output, err := getTerminusCMD.CombinedOutput()
	if err != nil {
		logger.Debugf("failed to run command %v, ouput: %s, err: %s", getTerminusCMD, output, err)
		fmt.Println("failed to check the existence of terminus, is it installed and running?")
		os.Exit(1)
	}

	return nil
}

type DetectPublicIPAddress struct {
	common.KubeAction
}

func (p *DetectPublicIPAddress) Execute(runtime connector.Runtime) error {
	if util.IsOnAWSEC2() {
		logger.Info("on AWS EC2 instance, will try to check if a public IP address is bound")
		awsPublicIP, err := util.GetPublicIPFromAWSIMDS()
		if err != nil {
			return errors.Wrap(err, "failed to get public IP from AWS")
		}
		if awsPublicIP != nil {
			logger.Info("retrieved public IP addresses from IMDS")
			p.KubeConf.Arg.NetworkSettings.CloudProviderPublicIP = awsPublicIP
			return nil
		}
	}

	if util.IsOnTencentCVM() {
		logger.Info("on Tencent CVM instance, will try to check if a public IP address is bound")
		tencentPublicIP, err := util.GetPublicIPFromTencentIMDS()
		if err != nil {
			return errors.Wrap(err, "failed to get public IP from Tencent")
		}
		if tencentPublicIP != nil {
			logger.Info("retrieved public IP addresses from IMDS")
			p.KubeConf.Arg.NetworkSettings.CloudProviderPublicIP = tencentPublicIP
			return nil
		}
	}

	if util.IsOnAliyunECS() {
		logger.Info("on Aliyun ECS instance, will try to check if a public IP address is bound")
		aliyunPublicIP, err := util.GetPublicIPFromAliyunIMDS()
		if err != nil {
			return errors.Wrap(err, "failed to get public IP from Aliyun")
		}
		if aliyunPublicIP != nil {
			logger.Info("retrieved public IP addresses from IMDS")
			p.KubeConf.Arg.NetworkSettings.CloudProviderPublicIP = aliyunPublicIP
			return nil
		}
	}

	osPublicIPs, err := util.GetPublicIPsFromOS()
	if err != nil {
		return errors.Wrap(err, "failed to get public IPs from OS")
	}
	if len(osPublicIPs) > 0 {
		logger.Info("detected public IP addresses on local network interface")
		p.KubeConf.Arg.NetworkSettings.OSPublicIPs = osPublicIPs
		return nil
	}

	if p.KubeConf.Arg.NetworkSettings.EnableReverseProxy != nil {
		if !*p.KubeConf.Arg.NetworkSettings.EnableReverseProxy {
			return nil
		}
		externalIP := getMyExternalIPAddr()
		if externalIP == nil {
			return errors.New("this installation is explicitly specified to disable reverse proxy but no valid public IP can be found")
		}
		p.KubeConf.Arg.NetworkSettings.ExternalPublicIP = externalIP
	}

	return nil

}

// getMyExternalIPAddr get my network outgoing ip address
func getMyExternalIPAddr() net.IP {
	sites := map[string]string{
		"httpbin":    "https://httpbin.org/ip",
		"ifconfigme": "https://ifconfig.me/all.json",
		"externalip": "https://myexternalip.com/json",
		"joinolares": "https://myip.joinolares.cn/ip",
	}

	type httpBin struct {
		Origin string `json:"origin"`
	}

	type ifconfigMe struct {
		IPAddr     string `json:"ip_addr"`
		RemoteHost string `json:"remote_host,omitempty"`
		UserAgent  string `json:"user_agent,omitempty"`
		Port       int    `json:"port,omitempty"`
		Method     string `json:"method,omitempty"`
		Encoding   string `json:"encoding,omitempty"`
		Via        string `json:"via,omitempty"`
		Forwarded  string `json:"forwarded,omitempty"`
	}

	type externalIP struct {
		IP string `json:"ip"`
	}

	var unmarshalFuncs = map[string]func(v []byte) string{
		"httpbin": func(v []byte) string {
			var hb httpBin
			if err := json.Unmarshal(v, &hb); err == nil && hb.Origin != "" {
				return hb.Origin
			}
			return ""
		},
		"ifconfigme": func(v []byte) string {
			var ifMe ifconfigMe
			if err := json.Unmarshal(v, &ifMe); err == nil && ifMe.IPAddr != "" {
				return ifMe.IPAddr
			}
			return ""
		},
		"externalip": func(v []byte) string {
			var extip externalIP
			if err := json.Unmarshal(v, &extip); err == nil && extip.IP != "" {
				return extip.IP
			}
			return ""
		},
		"joinolares": func(v []byte) string {
			return strings.TrimSpace(string(v))
		},
	}

	var mu sync.Mutex
	ch := make(chan any, len(sites))
	chSyncOp := func(f func()) {
		mu.Lock()
		defer mu.Unlock()
		if ch != nil {
			f()
		}
	}

	for site := range sites {
		go func(name string) {
			http.DefaultTransport.(*http.Transport).TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
			c := http.Client{Timeout: 5 * time.Second}
			resp, err := c.Get(sites[name])
			if err != nil {
				chSyncOp(func() { ch <- err })
				return
			}
			defer resp.Body.Close()
			respBytes, err := ioutil.ReadAll(resp.Body)
			if err != nil {
				chSyncOp(func() { ch <- err })
				return
			}

			ip := unmarshalFuncs[name](respBytes)
			//println(name, site, ip)
			chSyncOp(func() { ch <- ip })

		}(site)
	}

	tr := time.NewTimer(time.Duration(15*len(sites)+3) * time.Second)
	defer func() {
		tr.Stop()
		chSyncOp(func() {
			close(ch)
			ch = nil
		})
	}()

LOOP:
	for i := 0; i < len(sites); i++ {
		select {
		case r, ok := <-ch:
			if !ok {
				continue
			}

			switch v := r.(type) {
			case string:
				ip := net.ParseIP(v).To4()
				if ip.IsGlobalUnicast() && !ip.IsPrivate() {
					return ip
				}
			case error:
				logger.Debugf("got an error when reflecting public IP %v", v)
			}
		case <-tr.C:
			tr.Stop()
			logger.Debugf("timed out while fetching public IP")
			break LOOP
		}
	}

	return nil
}

type GetMasterInfo struct {
	common.KubeAction
	Print bool
}

type MasterInfo struct {
	JuiceFSEnabled      bool
	KubernetesInstalled bool
	OlaresInstalled     bool
	KubernetesType      string
	OlaresVersion       string
	MasterNodeName      string
	AllNodes            []string
}

func (t *GetMasterInfo) Execute(runtime connector.Runtime) (err error) {
	masterInfo := &MasterInfo{}
	defer func() {
		if err != nil {
			return
		}
		if t.Print {
			logger.Infof("Got master info:\nOlaresVersion: %s\nJuiceFSEnabled: %t\nKubernetesType: %s\nMasterNodeName: %s\nAllNodes: %s\n",
				masterInfo.OlaresVersion, masterInfo.JuiceFSEnabled, masterInfo.KubernetesType, masterInfo.MasterNodeName, strings.Join(masterInfo.AllNodes, ","))
		}

		t.PipelineCache.Set(common.MasterInfo, masterInfo)
	}()
	exist, err := runtime.GetRunner().FileExist(storage.JuiceFsServiceFile)
	if err != nil {
		return errors.Wrap(err, "failed to check whether JuiceFS service exists")
	}
	if !exist {
		masterInfo.JuiceFSEnabled = false
	} else {
		juiceFSCheckCMD := fmt.Sprintf("%s info %s", storage.JuiceFsFile, storage.OlaresJuiceFSRootDir)
		output, err := runtime.GetRunner().SudoCmd(juiceFSCheckCMD, false, false)
		if err != nil {
			return errors.Wrap(err, "failed to check JuiceFS status")
		}
		if !strings.Contains(output, "ERROR") {
			masterInfo.JuiceFSEnabled = true
		}
	}
	nodeList := &corev1.NodeList{}
	nodeCheckCMD := "kubectl get node -o json"
	output, err := runtime.GetRunner().SudoCmd(nodeCheckCMD, false, false)
	if err != nil {
		if strings.Contains(err.Error(), "command not found") {
			masterInfo.KubernetesInstalled = false
			return nil
		}
		return errors.Wrap(err, "failed to get Kubernetes node info")
	}
	masterInfo.KubernetesInstalled = true
	if err := json.Unmarshal([]byte(output), nodeList); err != nil {
		return errors.Wrap(err, "failed to parse Kubernetes node info")
	}
	for _, node := range nodeList.Items {
		masterInfo.AllNodes = append(masterInfo.AllNodes, node.Name)
		if _, ok := node.Labels["node-role.kubernetes.io/master"]; ok {
			masterInfo.MasterNodeName = node.Name
			if strings.Contains(node.Status.NodeInfo.KubeletVersion, common.K3s) {
				masterInfo.KubernetesType = common.K3s
			} else {
				masterInfo.KubernetesType = common.K8s
			}
		}
	}
	runtime.RemoteHost().SetName(masterInfo.MasterNodeName)
	t.KubeConf.Arg.MasterNodeName = masterInfo.MasterNodeName
	t.KubeConf.Arg.SetKubeVersion(masterInfo.KubernetesType)
	olaresVersionCMD := "kubectl get terminus -o jsonpath='{.items[*].spec.version}'"
	output, err = runtime.GetRunner().SudoCmd(olaresVersionCMD, false, false)
	if err != nil {
		if strings.Contains(err.Error(), "the server doesn't have a resource type") {
			masterInfo.OlaresInstalled = false
			return nil
		}
		return errors.Wrap(err, "failed to get Olares version (is it installed?)")
	}
	masterInfo.OlaresInstalled = true
	masterInfo.OlaresVersion = strings.TrimSpace(output)
	return nil

}

type AddNodePrecheck struct {
	common.KubeAction
}

func (a *AddNodePrecheck) Execute(runtime connector.Runtime) error {
	v, ok := a.PipelineCache.Get(common.MasterInfo)
	if !ok {
		return errors.New("failed to get master info")
	}
	masterInfo := v.(*MasterInfo)
	var errs []error
	defer func() {
		if len(errs) > 0 {
			var errStr string
			for _, err := range errs {
				errStr += err.Error() + "\n"
			}
			logger.Errorf("precheck failed, unable to add current node to the cluster:\n%s", errStr)
			os.Exit(1)
		}
	}()
	if !masterInfo.JuiceFSEnabled {
		errs = append(errs, errors.New("[JuiceFS] the master node has not enabled JuiceFS, which is required for multi nodes to share a same view of FileSystem"))
	}
	if !masterInfo.KubernetesInstalled {
		errs = append(errs, errors.New("[Kubernetes] the master node has not installed Kubernetes"))
	}
	if !masterInfo.OlaresInstalled {
		errs = append(errs, errors.New("[Olares] the master node has not installed Olares"))
	}
	for _, node := range masterInfo.AllNodes {
		if strings.EqualFold(node, runtime.GetSystemInfo().GetHostname()) {
			errs = append(errs, fmt.Errorf("[NodeName] the node name: \"%s\" has already been occupied by another node", node))
		}
	}
	return nil
}

type SaveMasterHostConfig struct {
	common.KubeAction
}

func (a *SaveMasterHostConfig) Execute(runtime connector.Runtime) error {
	if a.KubeConf.Arg.MasterHost == "" {
		logger.Info("master host is empty, skip saving to master config")
	}
	content, err := json.MarshalIndent(a.KubeConf.Arg.MasterHostConfig, "", "    ")
	if err != nil {
		return err
	}
	return os.WriteFile(filepath.Join(runtime.GetBaseDir(), common.MasterHostConfigFile), content, 0644)
}