woodpecker/pipeline/backend/kubernetes/kubernetes.go

package kubernetes

import (
	"context"
	"fmt"
	"io"
	"os"
	"strings"
	"time"

	"github.com/rs/zerolog/log"
	"github.com/woodpecker-ci/woodpecker/pipeline/backend/types"
	"gopkg.in/yaml.v3"

	"github.com/urfave/cli/v2"
	v1 "k8s.io/api/core/v1"
	"k8s.io/apimachinery/pkg/api/errors"
	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
	"k8s.io/client-go/informers"
	"k8s.io/client-go/kubernetes"
	"k8s.io/client-go/kubernetes/scheme"
	"k8s.io/client-go/rest"
	"k8s.io/client-go/tools/cache"

	// To authenticate to GCP K8s clusters
	_ "k8s.io/client-go/plugin/pkg/client/auth/gcp"
)

var noContext = context.Background()

type kube struct {
	ctx    context.Context
	client kubernetes.Interface
	config *Config
}

type Config struct {
	Namespace      string
	StorageClass   string
	VolumeSize     string
	StorageRwx     bool
	PodLabels      map[string]string
	PodAnnotations map[string]string
}

func configFromCliContext(ctx context.Context) (*Config, error) {
	if ctx != nil {
		if c, ok := ctx.Value(types.CliContext).(*cli.Context); ok {
			config := Config{
				Namespace:      c.String("backend-k8s-namespace"),
				StorageClass:   c.String("backend-k8s-storage-class"),
				VolumeSize:     c.String("backend-k8s-volume-size"),
				StorageRwx:     c.Bool("backend-k8s-storage-rwx"),
				PodLabels:      make(map[string]string), // just init empty map to prevent nil panic
				PodAnnotations: make(map[string]string), // just init empty map to prevent nil panic
			}
			// Unmarshal label and annotation settings here to ensure they're valid on startup
			if labels := c.String("backend-k8s-pod-labels"); labels != "" {
				if err := yaml.Unmarshal([]byte(labels), &config.PodLabels); err != nil {
					log.Error().Msgf("could not unmarshal pod labels '%s': %s", c.String("backend-k8s-pod-labels"), err)
					return nil, err
				}
			}
			if annotations := c.String("backend-k8s-pod-annotations"); annotations != "" {
				if err := yaml.Unmarshal([]byte(c.String("backend-k8s-pod-annotations")), &config.PodAnnotations); err != nil {
					log.Error().Msgf("could not unmarshal pod annotations '%s': %s", c.String("backend-k8s-pod-annotations"), err)
					return nil, err
				}
			}
			return &config, nil
		}
	}

	return nil, types.ErrNoCliContextFound
}

// New returns a new Kubernetes Engine.
func New(ctx context.Context) types.Engine {
	return &kube{
		ctx: ctx,
	}
}

func (e *kube) Name() string {
	return "kubernetes"
}

func (e *kube) IsAvailable(context.Context) bool {
	host := os.Getenv("KUBERNETES_SERVICE_HOST")
	return len(host) > 0
}

func (e *kube) Load(context.Context) error {
	config, err := configFromCliContext(e.ctx)
	if err != nil {
		return err
	}
	e.config = config

	var kubeClient kubernetes.Interface
	_, err = rest.InClusterConfig()
	if err != nil {
		kubeClient, err = getClientOutOfCluster()
	} else {
		kubeClient, err = getClientInsideOfCluster()
	}

	if err != nil {
		return err
	}

	e.client = kubeClient

	return nil
}

// Setup the pipeline environment.
func (e *kube) Setup(ctx context.Context, conf *types.Config) error {
	log.Trace().Msgf("Setting up Kubernetes primitives")

	for _, vol := range conf.Volumes {
		pvc, err := PersistentVolumeClaim(e.config.Namespace, vol.Name, e.config.StorageClass, e.config.VolumeSize, e.config.StorageRwx)
		if err != nil {
			return err
		}

		_, err = e.client.CoreV1().PersistentVolumeClaims(e.config.Namespace).Create(ctx, pvc, metav1.CreateOptions{})
		if err != nil {
			return err
		}
	}

	extraHosts := []string{}

	for _, stage := range conf.Stages {
		if stage.Alias == "services" {
			for _, step := range stage.Steps {
				stepName, err := dnsName(step.Name)
				if err != nil {
					return err
				}
				log.Trace().Str("pod-name", stepName).Msgf("Creating service: %s", step.Name)
				// TODO: support ports setting
				// svc, err := Service(e.config.Namespace, step.Name, stepName, step.Ports)
				svc, err := Service(e.config.Namespace, step.Name, stepName, []string{})
				if err != nil {
					return err
				}

				svc, err = e.client.CoreV1().Services(e.config.Namespace).Create(ctx, svc, metav1.CreateOptions{})
				if err != nil {
					return err
				}

				extraHosts = append(extraHosts, step.Networks[0].Aliases[0]+":"+svc.Spec.ClusterIP)
			}
		}
	}

	log.Trace().Msgf("Adding extra hosts: %s", strings.Join(extraHosts, ", "))
	for _, stage := range conf.Stages {
		for _, step := range stage.Steps {
			step.ExtraHosts = extraHosts
		}
	}

	return nil
}

// Start the pipeline step.
func (e *kube) Exec(ctx context.Context, step *types.Step) error {
	pod, err := Pod(e.config.Namespace, step, e.config.PodLabels, e.config.PodAnnotations)
	if err != nil {
		return err
	}

	log.Trace().Msgf("Creating pod: %s", pod.Name)
	_, err = e.client.CoreV1().Pods(e.config.Namespace).Create(ctx, pod, metav1.CreateOptions{})
	return err
}

// Wait for the pipeline step to complete and returns
// the completion results.
func (e *kube) Wait(ctx context.Context, step *types.Step) (*types.State, error) {
	podName, err := dnsName(step.Name)
	if err != nil {
		return nil, err
	}

	finished := make(chan bool)

	podUpdated := func(old, new interface{}) {
		pod := new.(*v1.Pod)
		if pod.Name == podName {
			if isImagePullBackOffState(pod) {
				finished <- true
			}

			switch pod.Status.Phase {
			case v1.PodSucceeded, v1.PodFailed, v1.PodUnknown:
				finished <- true
			}
		}
	}

	// TODO 5 seconds is against best practice, k3s didn't work otherwise
	si := informers.NewSharedInformerFactoryWithOptions(e.client, 5*time.Second, informers.WithNamespace(e.config.Namespace))
	if _, err := si.Core().V1().Pods().Informer().AddEventHandler(
		cache.ResourceEventHandlerFuncs{
			UpdateFunc: podUpdated,
		},
	); err != nil {
		return nil, err
	}

	stop := make(chan struct{})
	si.Start(stop)
	defer close(stop)

	// TODO Cancel on ctx.Done
	<-finished

	pod, err := e.client.CoreV1().Pods(e.config.Namespace).Get(ctx, podName, metav1.GetOptions{})
	if err != nil {
		return nil, err
	}

	if isImagePullBackOffState(pod) {
		return nil, fmt.Errorf("Could not pull image for pod %s", pod.Name)
	}

	bs := &types.State{
		ExitCode:  int(pod.Status.ContainerStatuses[0].State.Terminated.ExitCode),
		Exited:    true,
		OOMKilled: false,
	}

	return bs, nil
}

// Tail the pipeline step logs.
func (e *kube) Tail(ctx context.Context, step *types.Step) (io.ReadCloser, error) {
	podName, err := dnsName(step.Name)
	if err != nil {
		return nil, err
	}

	up := make(chan bool)

	podUpdated := func(old, new interface{}) {
		pod := new.(*v1.Pod)
		if pod.Name == podName {
			switch pod.Status.Phase {
			case v1.PodRunning, v1.PodSucceeded, v1.PodFailed:
				up <- true
			}
		}
	}

	// TODO 5 seconds is against best practice, k3s didn't work otherwise
	si := informers.NewSharedInformerFactoryWithOptions(e.client, 5*time.Second, informers.WithNamespace(e.config.Namespace))
	if _, err := si.Core().V1().Pods().Informer().AddEventHandler(
		cache.ResourceEventHandlerFuncs{
			UpdateFunc: podUpdated,
		},
	); err != nil {
		return nil, err
	}

	stop := make(chan struct{})
	si.Start(stop)
	defer close(stop)

	<-up

	opts := &v1.PodLogOptions{
		Follow:    true,
		Container: podName,
	}

	logs, err := e.client.CoreV1().RESTClient().Get().
		Namespace(e.config.Namespace).
		Name(podName).
		Resource("pods").
		SubResource("log").
		VersionedParams(opts, scheme.ParameterCodec).
		Stream(ctx)
	if err != nil {
		return nil, err
	}
	rc, wc := io.Pipe()

	go func() {
		defer logs.Close()
		defer wc.Close()
		defer rc.Close()

		_, err = io.Copy(wc, logs)
		if err != nil {
			return
		}
	}()
	return rc, nil

	// rc := io.NopCloser(bytes.NewReader(e.logs.Bytes()))
	// e.logs.Reset()
	// return rc, nil
}

// Destroy the pipeline environment.
func (e *kube) Destroy(_ context.Context, conf *types.Config) error {
	gracePeriodSeconds := int64(0) // immediately
	dpb := metav1.DeletePropagationBackground

	deleteOpts := metav1.DeleteOptions{
		GracePeriodSeconds: &gracePeriodSeconds,
		PropagationPolicy:  &dpb,
	}

	// Use noContext because the ctx sent to this function will be canceled/done in case of error or canceled by user.
	// Don't abort on 404 errors from k8s, they most likely mean that the pod hasn't been created yet, usually because pipeline was canceled before running all steps.
	// Trace log them in case the info could be useful when troubleshooting.

	for _, stage := range conf.Stages {
		for _, step := range stage.Steps {
			stepName, err := dnsName(step.Name)
			if err != nil {
				return err
			}
			log.Trace().Msgf("Deleting pod: %s", stepName)
			if err := e.client.CoreV1().Pods(e.config.Namespace).Delete(noContext, stepName, deleteOpts); err != nil {
				if errors.IsNotFound(err) {
					log.Trace().Err(err).Msgf("Unable to delete pod %s", stepName)
				} else {
					return err
				}
			}
		}
	}

	for _, stage := range conf.Stages {
		if stage.Alias == "services" {
			for _, step := range stage.Steps {
				log.Trace().Msgf("Deleting service: %s", step.Name)
				// TODO: support ports setting
				// svc, err := Service(e.config.Namespace, step.Name, step.Alias, step.Ports)
				svc, err := Service(e.config.Namespace, step.Name, step.Alias, []string{})
				if err != nil {
					return err
				}
				if err := e.client.CoreV1().Services(e.config.Namespace).Delete(noContext, svc.Name, deleteOpts); err != nil {
					if errors.IsNotFound(err) {
						log.Trace().Err(err).Msgf("Unable to delete service %s", svc.Name)
					} else {
						return err
					}
				}
			}
		}
	}

	for _, vol := range conf.Volumes {
		pvc, err := PersistentVolumeClaim(e.config.Namespace, vol.Name, e.config.StorageClass, e.config.VolumeSize, e.config.StorageRwx)
		if err != nil {
			return err
		}
		err = e.client.CoreV1().PersistentVolumeClaims(e.config.Namespace).Delete(noContext, pvc.Name, deleteOpts)
		if err != nil {
			if errors.IsNotFound(err) {
				log.Trace().Err(err).Msgf("Unable to delete pvc %s", pvc.Name)
			} else {
				return err
			}
		}
	}

	return nil
}