katalyst-core/pkg/util/service-discovery/discovery.go

/*
Copyright 2022 The Katalyst Authors.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/

package service_discovery

import (
	"context"
	"fmt"
	"sync"
	"time"

	v1 "k8s.io/api/core/v1"
	"k8s.io/apimachinery/pkg/api/errors"
	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
	"k8s.io/apimachinery/pkg/labels"
	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
	"k8s.io/apimachinery/pkg/util/wait"
	"k8s.io/client-go/informers"
	"k8s.io/client-go/kubernetes"
	corelisters "k8s.io/client-go/listers/core/v1"
	"k8s.io/client-go/tools/cache"
	"k8s.io/klog/v2"

	"github.com/kubewharf/katalyst-core/pkg/util/native"
)

const (
	defaultReSyncPeriod = time.Hour * 24
	defaultSyncPeriod   = time.Second * 3
)

// ServiceDiscoveryManager is used to discover all available endpoints.
type ServiceDiscoveryManager interface {
	// GetEndpoints get all endpoints list in the format `host:port`
	GetEndpoints() ([]string, error)

	// Run starts the service discovery manager
	Run() error
}

type DummyServiceDiscoveryManager struct{}

var _ ServiceDiscoveryManager = DummyServiceDiscoveryManager{}

func (d DummyServiceDiscoveryManager) GetEndpoints() ([]string, error) {
	return []string{}, nil
}

func (d DummyServiceDiscoveryManager) Run() error {
	return nil
}

type podInformerServiceDiscoveryManager struct {
	sync.RWMutex
	endpoints map[string]string

	portName   string
	ctx        context.Context
	podLister  corelisters.PodLister
	syncedFunc cache.InformerSynced
}

func NewPodInformerServiceDiscoveryManager(ctx context.Context, client kubernetes.Interface,
	podSelector labels.Selector, portName string) ServiceDiscoveryManager {
	klog.Infof("service discovery manager enabled with pod selector: %v", podSelector.String())
	podFactory := informers.NewSharedInformerFactoryWithOptions(client, defaultReSyncPeriod,
		informers.WithTweakListOptions(func(options *metav1.ListOptions) {
			options.LabelSelector = podSelector.String()
		}))
	podInformer := podFactory.Core().V1().Pods()

	m := &podInformerServiceDiscoveryManager{
		portName:   portName,
		ctx:        ctx,
		endpoints:  make(map[string]string),
		podLister:  podInformer.Lister(),
		syncedFunc: podInformer.Informer().HasSynced,
	}

	podInformer.Informer().AddEventHandler(cache.FilteringResourceEventHandler{
		FilterFunc: func(obj interface{}) bool {
			switch t := obj.(type) {
			case *v1.Pod:
				return native.PodIsReady(t)
			case cache.DeletedFinalStateUnknown:
				if pod, ok := t.Obj.(*v1.Pod); ok {
					return native.PodIsReady(pod)
				}
				utilruntime.HandleError(fmt.Errorf("unable to convert object %T to *v1.Pod", obj))
				return false
			default:
				utilruntime.HandleError(fmt.Errorf("unable to handle object: %T", obj))
				return false
			}
		},
		Handler: cache.ResourceEventHandlerFuncs{
			AddFunc:    m.addPod,
			UpdateFunc: m.updatePod,
			DeleteFunc: m.deletePod,
		},
	})

	podFactory.Start(ctx.Done())

	return m
}

// GetEndpoints get current all endpoints
func (m *podInformerServiceDiscoveryManager) GetEndpoints() ([]string, error) {
	m.RLock()
	defer m.RUnlock()

	endpoints := make([]string, 0, len(m.endpoints))
	for _, ep := range m.endpoints {
		endpoints = append(endpoints, ep)
	}

	return endpoints, nil
}

func (m *podInformerServiceDiscoveryManager) Run() error {
	if !cache.WaitForCacheSync(m.ctx.Done(), m.syncedFunc) {
		return fmt.Errorf("unable to sync caches for podInformerServiceDiscoveryManager")
	}

	go wait.Until(m.sync, defaultSyncPeriod, m.ctx.Done())
	return nil
}

func (m *podInformerServiceDiscoveryManager) addPod(obj interface{}) {
	pod, ok := obj.(*v1.Pod)
	if !ok {
		klog.ErrorS(nil, "Cannot convert to *v1.Pod", "obj", obj)
		return
	}

	klog.V(6).Infof("add pod %v", pod.Name)
	m.addEndpoint(pod)
}

func (m *podInformerServiceDiscoveryManager) updatePod(_, newObj interface{}) {
	newPod, ok := newObj.(*v1.Pod)
	if !ok {
		klog.ErrorS(nil, "Cannot convert to *v1.Pod", "obj", newObj)
		return
	}

	klog.V(6).Infof("update pod %v", newPod.Name)
	m.addEndpoint(newPod)
}

func (m *podInformerServiceDiscoveryManager) deletePod(obj interface{}) {
	pod, ok := obj.(*v1.Pod)
	if !ok {
		klog.ErrorS(nil, "Cannot convert to *v1.Pod", "obj", obj)
		return
	}

	klog.V(6).Infof("delete pod %v", pod.Name)
	m.removeEndpoint(pod)
}

func (m *podInformerServiceDiscoveryManager) removeEndpoint(pod *v1.Pod) {
	m.Lock()
	defer m.Unlock()

	key, err := cache.DeletionHandlingMetaNamespaceKeyFunc(pod)
	if err != nil {
		klog.Errorf("couldn't get key for pod %#v: %v", pod, err)
		return
	}

	delete(m.endpoints, key)
}

func (m *podInformerServiceDiscoveryManager) addEndpoint(pod *v1.Pod) {
	m.Lock()
	defer m.Unlock()

	key, err := cache.DeletionHandlingMetaNamespaceKeyFunc(pod)
	if err != nil {
		klog.Errorf("couldn't get key for pod %#v: %v", pod, err)
		return
	}

	endpoint, err := m.getPodEndpoint(pod)
	if err != nil {
		klog.ErrorS(err, "get new endpoint failed", "pod", pod)
		return
	}

	if originEndpoint, exist := m.endpoints[key]; exist && originEndpoint == endpoint {
		return
	}

	klog.Infof("add endpoint %s for pod %v", endpoint, key)

	m.endpoints[key] = endpoint
}

func (m *podInformerServiceDiscoveryManager) getPodEndpoint(pod *v1.Pod) (string, error) {
	ports := native.ParseHostPortsForPod(pod, m.portName)
	if len(ports) != 1 {
		return "", fmt.Errorf("pod has invalid amount of valid ports: %v", ports)
	}
	port := ports[0]

	hostIP, err := native.GetPodHostIP(pod)
	if err != nil {
		return "", fmt.Errorf("get pod hostIP failed: %v", err)
	}

	return fmt.Sprintf("%s:%d", hostIP, port), nil
}

func (m *podInformerServiceDiscoveryManager) sync() {
	m.Lock()
	defer m.Unlock()

	for key := range m.endpoints {
		namespace, name, err := cache.SplitMetaNamespaceKey(key)
		if err != nil {
			klog.Errorf("failed to split namespace and name from key %s", key)
			continue
		}

		if _, err := m.podLister.Pods(namespace).Get(name); err != nil {
			if errors.IsNotFound(err) {
				delete(m.endpoints, key)
			} else {
				klog.Errorf("failed to get pod %s: %s", key, err)
			}
		}
	}
}