deployment.go

/*
Copyright 2015 The Kubernetes 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.

This file was copied and modified from the kubernetes/kubernetes project
https://github.com/kubernetes/kubernetes/release-1.8/pkg/controller/deployment/deployment_controller.go

Modifications Copyright (c) 2017 SAP SE or an SAP affiliate company. All rights reserved.
*/

// Package controller is used to provide the core functionalities of machine-controller-manager
package controller

import (
	"fmt"
	"reflect"
	"sync"
	"time"

	"github.com/golang/glog"

	"k8s.io/api/core/v1"
	"k8s.io/apimachinery/pkg/api/errors"
	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
	"k8s.io/apimachinery/pkg/labels"
	"k8s.io/apimachinery/pkg/types"
	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
	"k8s.io/apimachinery/pkg/util/sets"
	"k8s.io/client-go/tools/cache"

	machine "github.com/gardener/machine-controller-manager/pkg/apis/cluster"
	common "github.com/gardener/machine-controller-manager/pkg/apis/cluster/common"
	"github.com/gardener/machine-controller-manager/pkg/apis/cluster/v1alpha1"
)

// controllerKind contains the schema.GroupVersionKind for this controller type.
var controllerKind = v1alpha1.SchemeGroupVersion.WithKind("MachineDeployment")

// GroupVersionKind is the version kind used to identify objects managed by machine-controller-manager
var GroupVersionKind = "machine.sapcloud.io/v1alpha1"

func (dc *controller) addMachineDeployment(obj interface{}) {
	d := obj.(*v1alpha1.MachineDeployment)
	glog.V(4).Infof("Adding machine deployment %s", d.Name)
	dc.enqueueMachineDeployment(d)
}

func (dc *controller) updateMachineDeployment(old, cur interface{}) {
	oldD := old.(*v1alpha1.MachineDeployment)
	curD := cur.(*v1alpha1.MachineDeployment)
	glog.V(4).Infof("Updating machine deployment %s", oldD.Name)
	dc.enqueueMachineDeployment(curD)
}

func (dc *controller) deleteMachineDeployment(obj interface{}) {
	d, ok := obj.(*v1alpha1.MachineDeployment)
	if !ok {
		tombstone, ok := obj.(cache.DeletedFinalStateUnknown)
		if !ok {
			//utilruntime.HandleError(fmt.Errorf("Couldn't get object from tombstone %#v", obj))
			return
		}
		d, ok = tombstone.Obj.(*v1alpha1.MachineDeployment)
		if !ok {
			utilruntime.HandleError(fmt.Errorf("Tombstone contained object that is not a Machine Deployment %#v", obj))
			return
		}
	}
	glog.V(4).Infof("Deleting machine deployment %s", d.Name)
	dc.enqueueMachineDeployment(d)
}

// addMachineSet enqueues the deployment that manages a MachineSet when the MachineSet is created.
func (dc *controller) addMachineSetToDeployment(obj interface{}) {
	is := obj.(*v1alpha1.MachineSet)

	if is.DeletionTimestamp != nil {
		// On a restart of the controller manager, it's possible for an object to
		// show up in a state that is already pending deletion.
		dc.deleteMachineSetToDeployment(is)
		return
	}

	// If it has a ControllerRef, that's all that matters.
	if controllerRef := metav1.GetControllerOf(is); controllerRef != nil {
		d := dc.resolveDeploymentControllerRef(is.Namespace, controllerRef)
		if d == nil {
			return
		}
		glog.V(4).Infof("MachineSet %s added.", is.Name)
		dc.enqueueMachineDeployment(d)
		return
	}

	// Otherwise, it's an orphan. Get a list of all matching Deployments and sync
	// them to see if anyone wants to adopt it.
	ds := dc.getMachineDeploymentsForMachineSet(is)
	if len(ds) == 0 {
		return
	}
	glog.V(4).Infof("Orphan MachineSet %s added.", is.Name)
	for _, d := range ds {
		dc.enqueueMachineDeployment(d)
	}
}

// getDeploymentsForMachineSet returns a list of Deployments that potentially
// match a MachineSet.
func (dc *controller) getMachineDeploymentsForMachineSet(machineSet *v1alpha1.MachineSet) []*v1alpha1.MachineDeployment {
	deployments, err := dc.GetMachineDeploymentsForMachineSet(machineSet)
	if err != nil || len(deployments) == 0 {
		return nil
	}
	// Because all MachineSet's belonging to a deployment should have a unique label key,
	// there should never be more than one deployment returned by the above method.
	// If that happens we should probably dynamically repair the situation by ultimately
	// trying to clean up one of the controllers, for now we just return the older one
	if len(deployments) > 1 {
		// ControllerRef will ensure we don't do anything crazy, but more than one
		// item in this list nevertheless constitutes user error.
		glog.V(4).Infof("user error! more than one deployment is selecting machine set %s with labels: %#v, returning %s",
			machineSet.Name, machineSet.Labels, deployments[0].Name)
	}
	return deployments
}

// updateMachineSet figures out what deployment(s) manage a MachineSet when the MachineSet
// is updated and wake them up. If the anything of the MachineSets have changed, we need to
// awaken both the old and new deployments. old and cur must be *extensions.MachineSet
// types.
func (dc *controller) updateMachineSetToDeployment(old, cur interface{}) {
	curMachineSet := cur.(*v1alpha1.MachineSet)
	oldMachineSet := old.(*v1alpha1.MachineSet)
	if curMachineSet.ResourceVersion == oldMachineSet.ResourceVersion {
		// Periodic resync will send update events for all known machine sets.
		// Two different versions of the same machine set will always have different RVs.
		return
	}

	curControllerRef := metav1.GetControllerOf(curMachineSet)
	oldControllerRef := metav1.GetControllerOf(oldMachineSet)
	controllerRefChanged := !reflect.DeepEqual(curControllerRef, oldControllerRef)
	if controllerRefChanged && oldControllerRef != nil {
		// The ControllerRef was changed. Sync the old controller, if any.
		if d := dc.resolveDeploymentControllerRef(oldMachineSet.Namespace, oldControllerRef); d != nil {
			dc.enqueueMachineDeployment(d)
		}
	}

	// If it has a ControllerRef, that's all that matters.
	if curControllerRef != nil {
		d := dc.resolveDeploymentControllerRef(curMachineSet.Namespace, curControllerRef)
		if d == nil {
			return
		}
		glog.V(4).Infof("MachineSet %s updated.", curMachineSet.Name)
		dc.enqueueMachineDeployment(d)
		return
	}

	// Otherwise, it's an orphan. If anything changed, sync matching controllers
	// to see if anyone wants to adopt it now.
	labelChanged := !reflect.DeepEqual(curMachineSet.Labels, oldMachineSet.Labels)
	if labelChanged || controllerRefChanged {
		ds := dc.getMachineDeploymentsForMachineSet(curMachineSet)
		if len(ds) == 0 {
			return
		}
		glog.V(4).Infof("Orphan MachineSet %s updated.", curMachineSet.Name)
		for _, d := range ds {
			dc.enqueueMachineDeployment(d)
		}
	}
}

// deleteMachineSet enqueues the deployment that manages a MachineSet when
// the MachineSet is deleted. obj could be an *v1alpha1.MachineSet, or
// a DeletionFinalStateUnknown marker item.
func (dc *controller) deleteMachineSetToDeployment(obj interface{}) {
	machineSet, ok := obj.(*v1alpha1.MachineSet)

	// When a delete is dropped, the relist will notice a Machine in the store not
	// in the list, leading to the insertion of a tombstone object which contains
	// the deleted key/value. Note that this value might be stale. If the MachineSet
	// changed labels the new deployment will not be woken up till the periodic resync.
	if !ok {
		tombstone, ok := obj.(cache.DeletedFinalStateUnknown)
		if !ok {
			utilruntime.HandleError(fmt.Errorf("Couldn't get object from tombstone %#v", obj))
			return
		}
		machineSet, ok = tombstone.Obj.(*v1alpha1.MachineSet)
		if !ok {
			utilruntime.HandleError(fmt.Errorf("Tombstone contained object that is not a MachineSet %#v", obj))
			return
		}
	}

	controllerRef := metav1.GetControllerOf(machineSet)
	if controllerRef == nil {
		// No controller should care about orphans being deleted.
		return
	}
	d := dc.resolveDeploymentControllerRef(machineSet.Namespace, controllerRef)
	if d == nil {
		return
	}
	glog.V(4).Infof("MachineSet %s deleted.", machineSet.Name)
	dc.enqueueMachineDeployment(d)
}

// deleteMachine will enqueue a Recreate Deployment once all of its Machines have stopped running.
func (dc *controller) deleteMachineToMachineDeployment(obj interface{}) {
	machine, ok := obj.(*v1alpha1.Machine)

	// When a delete is dropped, the relist will notice a Machine in the store not
	// in the list, leading to the insertion of a tombstone object which contains
	// the deleted key/value. Note that this value might be stale. If the Machine
	// changed labels the new deployment will not be woken up till the periodic resync.
	if !ok {
		tombstone, ok := obj.(cache.DeletedFinalStateUnknown)
		if !ok {
			utilruntime.HandleError(fmt.Errorf("Couldn't get object from tombstone %#v", obj))
			return
		}
		machine, ok = tombstone.Obj.(*v1alpha1.Machine)
		if !ok {
			utilruntime.HandleError(fmt.Errorf("Tombstone contained object that is not a machine %#v", obj))
			return
		}
	}
	glog.V(4).Infof("Machine %s deleted.", machine.Name)
	if d := dc.getMachineDeploymentForMachine(machine); d != nil && d.Spec.Strategy.Type == common.RecreateMachineDeploymentStrategyType {
		// Sync if this Deployment now has no more Machines.
		machineSets, err := ListMachineSets(d, IsListFromClient(dc.controlMachineClient))
		if err != nil {
			return
		}
		machineMap, err := dc.getMachineMapForMachineDeployment(d, machineSets)
		if err != nil {
			return
		}
		numMachines := 0
		for _, machineList := range machineMap {
			numMachines += len(machineList.Items)
		}
		if numMachines == 0 {
			dc.enqueueMachineDeployment(d)
		}
	}
}

func (dc *controller) enqueueMachineDeployment(deployment *v1alpha1.MachineDeployment) {
	key, err := KeyFunc(deployment)
	if err != nil {
		utilruntime.HandleError(fmt.Errorf("Couldn't get key for object %#v: %v", deployment, err))
		return
	}

	dc.machineDeploymentQueue.Add(key)
}

func (dc *controller) enqueueRateLimited(deployment *v1alpha1.MachineDeployment) {
	key, err := KeyFunc(deployment)
	if err != nil {
		utilruntime.HandleError(fmt.Errorf("Couldn't get key for object %#v: %v", deployment, err))
		return
	}

	dc.machineDeploymentQueue.AddRateLimited(key)
}

//  enqueueMachineDeploymentAfter will enqueue a deployment after the provided amount of time.
func (dc *controller) enqueueMachineDeploymentAfter(deployment *v1alpha1.MachineDeployment, after time.Duration) {
	key, err := KeyFunc(deployment)
	if err != nil {
		utilruntime.HandleError(fmt.Errorf("Couldn't get key for object %#v: %v", deployment, err))
		return
	}

	dc.machineDeploymentQueue.AddAfter(key, after)
}

// getDeploymentForMachine returns the deployment managing the given Machine.
func (dc *controller) getMachineDeploymentForMachine(machine *v1alpha1.Machine) *v1alpha1.MachineDeployment {
	// Find the owning machine set
	var is *v1alpha1.MachineSet
	var err error
	controllerRef := metav1.GetControllerOf(machine)
	if controllerRef == nil {
		// No controller owns this Machine.
		return nil
	}
	if controllerRef.Kind != "MachineDeployment" { //TODO: Remove hardcoded string
		// Not a Machine owned by a machine set.
		return nil
	}
	is, err = dc.controlMachineClient.MachineSets(machine.Namespace).Get(controllerRef.Name, metav1.GetOptions{})
	if err != nil || is.UID != controllerRef.UID {
		glog.V(4).Infof("Cannot get machineset %q for machine %q: %v", controllerRef.Name, machine.Name, err)
		return nil
	}

	// Now find the Deployment that owns that MachineSet.
	controllerRef = metav1.GetControllerOf(is)
	if controllerRef == nil {
		return nil
	}
	return dc.resolveDeploymentControllerRef(is.Namespace, controllerRef)
}

// resolveControllerRef returns the controller referenced by a ControllerRef,
// or nil if the ControllerRef could not be resolved to a matching controller
// of the correct Kind.
func (dc *controller) resolveDeploymentControllerRef(namespace string, controllerRef *metav1.OwnerReference) *v1alpha1.MachineDeployment {
	// We can't look up by UID, so look up by Name and then verify UID.
	// Don't even try to look up by Name if it's the wrong Kind.
	if controllerRef.Kind != controllerKind.Kind {
		return nil
	}
	d, err := dc.controlMachineClient.MachineDeployments(namespace).Get(controllerRef.Name, metav1.GetOptions{})
	if err != nil {
		return nil
	}
	if d.UID != controllerRef.UID {
		// The controller we found with this Name is not the same one that the
		// ControllerRef points to.
		return nil
	}
	return d
}

func (dc *controller) handleErr(err error, key interface{}) {
	if err == nil {
		dc.machineDeploymentQueue.Forget(key)
		return
	}

	if dc.machineDeploymentQueue.NumRequeues(key) < maxRetries {
		glog.V(2).Infof("Error syncing deployment %v: %v", key, err)
		dc.machineDeploymentQueue.AddRateLimited(key)
		return
	}

	utilruntime.HandleError(err)
	glog.V(2).Infof("Dropping deployment %q out of the queue: %v", key, err)
	dc.machineDeploymentQueue.Forget(key)
}

// getMachineSetsForDeployment uses ControllerRefManager to reconcile
// ControllerRef by adopting and orphaning.
// It returns the list of MachineSets that this Deployment should manage.
func (dc *controller) getMachineSetsForMachineDeployment(d *v1alpha1.MachineDeployment) ([]*v1alpha1.MachineSet, error) {
	// List all MachineSets to find those we own but that no longer match our
	// selector. They will be orphaned by ClaimMachineSets().
	machineSets, err := dc.machineSetLister.List(labels.Everything())
	if err != nil {
		return nil, err
	}
	deploymentSelector, err := metav1.LabelSelectorAsSelector(d.Spec.Selector)
	if err != nil {
		return nil, fmt.Errorf("machine deployment %s has invalid label selector: %v", d.Name, err)
	}
	// If any adoptions are attempted, we should first recheck for deletion with
	// an uncached quorum read sometime after listing MachineSets (see #42639).
	canAdoptFunc := RecheckDeletionTimestamp(func() (metav1.Object, error) {
		fresh, err := dc.controlMachineClient.MachineDeployments(d.Namespace).Get(d.Name, metav1.GetOptions{})
		if err != nil {
			return nil, err
		}
		if fresh.UID != d.UID {
			return nil, fmt.Errorf("original Machine Deployment %v is gone: got uid %v, wanted %v", d.Name, fresh.UID, d.UID)
		}
		return fresh, nil
	})
	cm := NewMachineSetControllerRefManager(dc.machineSetControl, d, deploymentSelector, controllerKind, canAdoptFunc)
	ISes, err := cm.ClaimMachineSets(machineSets)
	return ISes, err
}

// getMachineMapForDeployment returns the Machines managed by a Deployment.
//
// It returns a map from MachineSet UID to a list of Machines controlled by that RS,
// according to the Machine's ControllerRef.
func (dc *controller) getMachineMapForMachineDeployment(d *v1alpha1.MachineDeployment, machineSets []*v1alpha1.MachineSet) (map[types.UID]*v1alpha1.MachineList, error) {
	// Get all Machines that potentially belong to this Deployment.
	selector, err := metav1.LabelSelectorAsSelector(d.Spec.Selector)
	if err != nil {
		return nil, err
	}
	machines, err := dc.machineLister.List(selector)
	if err != nil {
		return nil, err
	}
	// Group Machines by their controller (if it's in rsList).
	machineMap := make(map[types.UID]*v1alpha1.MachineList, len(machineSets))
	for _, is := range machineSets {
		machineMap[is.UID] = &v1alpha1.MachineList{}
	}
	for _, machine := range machines {
		// Do not ignore inactive Machines because Recreate Deployments need to verify that no
		// Machines from older versions are running before spinning up new Machines.
		controllerRef := metav1.GetControllerOf(machine)
		if controllerRef == nil {
			continue
		}
		// Only append if we care about this UID.
		if machineList, ok := machineMap[controllerRef.UID]; ok {
			machineList.Items = append(machineList.Items, *machine)
		}
	}
	return machineMap, nil
}

// reconcileClusterMachineDeployment will sync the deployment with the given key.
// This function is not meant to be invoked concurrently with the same key.
func (dc *controller) reconcileClusterMachineDeployment(key string) error {
	startTime := time.Now()
	glog.V(3).Infof("Started syncing deployment %q (%v)", key, startTime)
	defer func() {
		glog.V(3).Infof("Finished syncing deployment %q (%v)", key, time.Since(startTime))
	}()

	_, name, err := cache.SplitMetaNamespaceKey(key)
	if err != nil {
		return err
	}
	deployment, err := dc.controlMachineClient.MachineDeployments(dc.namespace).Get(name, metav1.GetOptions{})
	if errors.IsNotFound(err) {
		glog.V(4).Infof("Deployment %v has been deleted", key)
		return nil
	}
	if err != nil {
		return err
	}

	// If MachineDeployment is frozen and no deletion timestamp, don't process it
	if deployment.Labels["freeze"] == "True" && deployment.DeletionTimestamp == nil {
		glog.V(3).Infof("MachineDeployment %q is frozen. However, it will still be processed if it there is an scale down event.", deployment.Name)
	}

	// Validate MachineDeployment
	internalMachineDeployment := &machine.MachineDeployment{}

	err = v1alpha1.Convert_v1alpha1_MachineDeployment_To_cluster_MachineDeployment(deployment, internalMachineDeployment, nil)
	if err != nil {
		return err
	}

	// validationerr := validation.ValidateMachineDeployment(internalMachineDeployment)
	// if validationerr.ToAggregate() != nil && len(validationerr.ToAggregate().Errors()) > 0 {
	// 	glog.V(2).Infof("Validation of MachineDeployment failled %s", validationerr.ToAggregate().Error())
	// 	return nil
	// }

	// TODO: Enable Validation again.
	// // Validate MachineClass
	// _, secretRef, err := dc.validateMachineClass(&deployment.Spec.Template.Spec.Class)
	// if err != nil || secretRef == nil {
	// 	return err
	// }

	// Resync the MachineDeployment after 10 minutes to avoid missing out on missed out events
	defer dc.enqueueMachineDeploymentAfter(deployment, 10*time.Minute)

	// Deep-copy otherwise we are mutating our cache.
	// TODO: Deep-copy only when needed.
	d := deployment.DeepCopy()

	// Manipulate finalizers
	if d.DeletionTimestamp == nil {
		dc.addMachineDeploymentFinalizers(d)
	}

	everything := metav1.LabelSelector{}
	if reflect.DeepEqual(d.Spec.Selector, &everything) {
		dc.recorder.Eventf(d, v1.EventTypeWarning, "SelectingAll", "This deployment is selecting all machines. A non-empty selector is required.")
		if d.Status.ObservedGeneration < d.Generation {
			d.Status.ObservedGeneration = d.Generation
			dc.controlMachineClient.MachineDeployments(d.Namespace).Update(d)
		}
		return nil
	}

	// List MachineSets owned by this Deployment, while reconciling ControllerRef
	// through adoption/orphaning.
	machineSets, err := dc.getMachineSetsForMachineDeployment(d)
	if err != nil {
		return err
	}
	// List all Machines owned by this Deployment, grouped by their MachineSet.
	// Current uses of the MachineMap are:
	//
	// * check if a Machine is labeled correctly with the Machine-template-hash label.
	// * check that no old Machines are running in the middle of Recreate Deployments.
	machineMap, err := dc.getMachineMapForMachineDeployment(d, machineSets)
	if err != nil {
		return err
	}

	if d.DeletionTimestamp != nil {
		if finalizers := sets.NewString(d.Finalizers...); !finalizers.Has(DeleteFinalizerName) {
			return nil
		}
		if len(machineSets) == 0 {
			dc.deleteMachineDeploymentFinalizers(d)
			return nil
		}
		glog.V(4).Infof("Deleting all child MachineSets as MachineDeployment %s has set deletionTimestamp", d.Name)
		dc.terminateMachineSets(machineSets, d)
		return dc.syncStatusOnly(d, machineSets, machineMap)
	}

	// Update deployment conditions with an Unknown condition when pausing/resuming
	// a deployment. In this way, we can be sure that we won't timeout when a user
	// resumes a Deployment with a set progressDeadlineSeconds.
	if err = dc.checkPausedConditions(d); err != nil {
		return err
	}

	if d.Spec.Paused {
		return dc.sync(d, machineSets, machineMap)
	}

	// rollback is not re-entrant in case the underlying machine sets are updated with a new
	// revision so we should ensure that we won't proceed to update machine sets until we
	// make sure that the deployment has cleaned up its rollback spec in subsequent enqueues.
	if d.Spec.RollbackTo != nil {
		return dc.rollback(d, machineSets, machineMap)
	}

	scalingEvent, err := dc.isScalingEvent(d, machineSets, machineMap)

	if err != nil {
		return err
	}
	if scalingEvent {
		return dc.sync(d, machineSets, machineMap)
	}

	switch d.Spec.Strategy.Type {
	case common.RecreateMachineDeploymentStrategyType:
		return dc.rolloutRecreate(d, machineSets, machineMap)
	case common.RollingUpdateMachineDeploymentStrategyType:
		return dc.rolloutRolling(d, machineSets, machineMap)
	}
	return fmt.Errorf("unexpected deployment strategy type: %s", d.Spec.Strategy.Type)
}

func (dc *controller) terminateMachineSets(machineSets []*v1alpha1.MachineSet, deployment *v1alpha1.MachineDeployment) {
	var (
		wg               sync.WaitGroup
		numOfMachinesets = len(machineSets)
	)
	wg.Add(numOfMachinesets)

	for _, machineSet := range machineSets {
		go func(machineSet *v1alpha1.MachineSet) {
			defer wg.Done()
			// Machine is already marked as 'to-be-deleted'
			if machineSet.DeletionTimestamp != nil {
				return
			}
			dc.controlMachineClient.MachineSets(machineSet.Namespace).Delete(machineSet.Name, nil)
		}(machineSet)
	}
	wg.Wait()
}

/*
	SECTION
	Manipulate Finalizers
*/

func (dc *controller) addMachineDeploymentFinalizers(machineDeployment *v1alpha1.MachineDeployment) {
	clone := machineDeployment.DeepCopy()

	if finalizers := sets.NewString(clone.Finalizers...); !finalizers.Has(DeleteFinalizerName) {
		finalizers.Insert(DeleteFinalizerName)
		dc.updateMachineDeploymentFinalizers(clone, finalizers.List())
	}
}

func (dc *controller) deleteMachineDeploymentFinalizers(machineDeployment *v1alpha1.MachineDeployment) {
	clone := machineDeployment.DeepCopy()

	if finalizers := sets.NewString(clone.Finalizers...); finalizers.Has(DeleteFinalizerName) {
		finalizers.Delete(DeleteFinalizerName)
		dc.updateMachineDeploymentFinalizers(clone, finalizers.List())
	}
}

func (dc *controller) updateMachineDeploymentFinalizers(machineDeployment *v1alpha1.MachineDeployment, finalizers []string) {
	// Get the latest version of the machineDeployment so that we can avoid conflicts
	machineDeployment, err := dc.controlMachineClient.MachineDeployments(machineDeployment.Namespace).Get(machineDeployment.Name, metav1.GetOptions{})
	if err != nil {
		return
	}

	clone := machineDeployment.DeepCopy()
	clone.Finalizers = finalizers
	_, err = dc.controlMachineClient.MachineDeployments(machineDeployment.Namespace).Update(clone)
	if err != nil {
		// Keep retrying until update goes through
		glog.Warning("Updated failed, retrying")
		dc.updateMachineDeploymentFinalizers(machineDeployment, finalizers)
	}
}