my-rook-config/pkg/operator/ceph/pool/controller.go

/*
Copyright 2016 The Rook Authors. All rights reserved.

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 pool to manage a rook pool.
package pool

import (
	"context"
	"fmt"
	"reflect"
	"strings"

	"github.com/coreos/pkg/capnslog"
	cephclient "github.com/rook/rook/pkg/daemon/ceph/client"

	"github.com/pkg/errors"
	cephv1 "github.com/rook/rook/pkg/apis/ceph.rook.io/v1"
	"github.com/rook/rook/pkg/clusterd"
	"github.com/rook/rook/pkg/operator/ceph/cluster/mon"
	"github.com/rook/rook/pkg/operator/ceph/config"
	opcontroller "github.com/rook/rook/pkg/operator/ceph/controller"
	"github.com/rook/rook/pkg/operator/ceph/csi/peermap"
	"github.com/rook/rook/pkg/operator/ceph/reporting"
	"github.com/rook/rook/pkg/operator/k8sutil"
	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/runtime"
	"k8s.io/apimachinery/pkg/types"
	"k8s.io/client-go/tools/record"
	"sigs.k8s.io/controller-runtime/pkg/client"
	"sigs.k8s.io/controller-runtime/pkg/controller"
	"sigs.k8s.io/controller-runtime/pkg/handler"
	"sigs.k8s.io/controller-runtime/pkg/manager"
	"sigs.k8s.io/controller-runtime/pkg/reconcile"
	"sigs.k8s.io/controller-runtime/pkg/source"
)

const (
	poolApplicationNameRBD = "rbd"
	controllerName         = "ceph-block-pool-controller"
)

var logger = capnslog.NewPackageLogger("github.com/rook/rook", controllerName)

var cephBlockPoolKind = reflect.TypeOf(cephv1.CephBlockPool{}).Name()

// Sets the type meta for the controller main object
var controllerTypeMeta = metav1.TypeMeta{
	Kind:       cephBlockPoolKind,
	APIVersion: fmt.Sprintf("%s/%s", cephv1.CustomResourceGroup, cephv1.Version),
}

var _ reconcile.Reconciler = &ReconcileCephBlockPool{}

// ReconcileCephBlockPool reconciles a CephBlockPool object
type ReconcileCephBlockPool struct {
	client            client.Client
	scheme            *runtime.Scheme
	context           *clusterd.Context
	clusterInfo       *cephclient.ClusterInfo
	blockPoolContexts map[string]*blockPoolHealth
	opManagerContext  context.Context
	recorder          record.EventRecorder
}

type blockPoolHealth struct {
	internalCtx    context.Context
	internalCancel context.CancelFunc
	started        bool
}

// Add creates a new CephBlockPool Controller and adds it to the Manager. The Manager will set fields on the Controller
// and Start it when the Manager is Started.
func Add(mgr manager.Manager, context *clusterd.Context, opManagerContext context.Context, opConfig opcontroller.OperatorConfig) error {
	return add(opManagerContext, mgr, newReconciler(mgr, context, opManagerContext))
}

// newReconciler returns a new reconcile.Reconciler
func newReconciler(mgr manager.Manager, context *clusterd.Context, opManagerContext context.Context) reconcile.Reconciler {
	return &ReconcileCephBlockPool{
		client:            mgr.GetClient(),
		scheme:            mgr.GetScheme(),
		context:           context,
		blockPoolContexts: make(map[string]*blockPoolHealth),
		opManagerContext:  opManagerContext,
		recorder:          mgr.GetEventRecorderFor("rook-" + controllerName),
	}
}

func add(opManagerContext context.Context, mgr manager.Manager, r reconcile.Reconciler) error {
	// Create a new controller
	c, err := controller.New(controllerName, mgr, controller.Options{Reconciler: r})
	if err != nil {
		return err
	}
	logger.Info("successfully started")

	// Watch for changes on the CephBlockPool CRD object
	err = c.Watch(&source.Kind{Type: &cephv1.CephBlockPool{TypeMeta: controllerTypeMeta}}, &handler.EnqueueRequestForObject{}, opcontroller.WatchControllerPredicate())
	if err != nil {
		return err
	}

	// Build Handler function to return the list of ceph block pool
	// This is used by the watchers below
	handlerFunc, err := opcontroller.ObjectToCRMapper(opManagerContext, mgr.GetClient(), &cephv1.CephBlockPoolList{}, mgr.GetScheme())
	if err != nil {
		return err
	}

	// Watch for ConfigMap "rook-ceph-mon-endpoints" update and reconcile, which will reconcile update the bootstrap peer token
	err = c.Watch(&source.Kind{Type: &corev1.ConfigMap{TypeMeta: metav1.TypeMeta{Kind: "ConfigMap", APIVersion: corev1.SchemeGroupVersion.String()}}}, handler.EnqueueRequestsFromMapFunc(handlerFunc), mon.PredicateMonEndpointChanges())
	if err != nil {
		return err
	}

	return nil
}

// Reconcile reads that state of the cluster for a CephBlockPool object and makes changes based on the state read
// and what is in the CephBlockPool.Spec
// The Controller will requeue the Request to be processed again if the returned error is non-nil or
// Result.Requeue is true, otherwise upon completion it will remove the work from the queue.
func (r *ReconcileCephBlockPool) Reconcile(context context.Context, request reconcile.Request) (reconcile.Result, error) {
	// workaround because the rook logging mechanism is not compatible with the controller-runtime logging interface
	reconcileResponse, cephBlockPool, err := r.reconcile(request)

	return reporting.ReportReconcileResult(logger, r.recorder, request, &cephBlockPool, reconcileResponse, err)
}

func (r *ReconcileCephBlockPool) reconcile(request reconcile.Request) (reconcile.Result, cephv1.CephBlockPool, error) {
	// Fetch the CephBlockPool instance
	cephBlockPool := &cephv1.CephBlockPool{}
	err := r.client.Get(r.opManagerContext, request.NamespacedName, cephBlockPool)
	if err != nil {
		if kerrors.IsNotFound(err) {
			logger.Debug("CephBlockPool resource not found. Ignoring since object must be deleted.")
			// If there was a previous error or if a user removed this resource's finalizer, it's
			// possible Rook didn't clean up the monitoring routine for this resource. Ensure the
			// routine is stopped when we see the resource is gone.
			cephBlockPool.Name = request.Name
			cephBlockPool.Namespace = request.Namespace
			r.cancelMirrorMonitoring(cephBlockPool)
			return reconcile.Result{}, *cephBlockPool, nil
		}
		// Error reading the object - requeue the request.
		return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrap(err, "failed to get CephBlockPool")
	}
	// update observedGeneration local variable with current generation value,
	// because generation can be changed before reconile got completed
	// CR status will be updated at end of reconcile, so to reflect the reconcile has finished
	observedGeneration := cephBlockPool.ObjectMeta.Generation

	// Set a finalizer so we can do cleanup before the object goes away
	err = opcontroller.AddFinalizerIfNotPresent(r.opManagerContext, r.client, cephBlockPool)
	if err != nil {
		return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrap(err, "failed to add finalizer")
	}

	// The CR was just created, initializing status fields
	if cephBlockPool.Status == nil {
		updateStatus(r.opManagerContext, r.client, request.NamespacedName, cephv1.ConditionProgressing, nil, k8sutil.ObservedGenerationNotAvailable)
	}

	// Make sure a CephCluster is present otherwise do nothing
	cephCluster, isReadyToReconcile, cephClusterExists, reconcileResponse := opcontroller.IsReadyToReconcile(r.opManagerContext, r.client, request.NamespacedName, controllerName)
	if !isReadyToReconcile {
		// This handles the case where the Ceph Cluster is gone and we want to delete that CR
		// We skip the deletePool() function since everything is gone already
		//
		// Also, only remove the finalizer if the CephCluster is gone
		// If not, we should wait for it to be ready
		// This handles the case where the operator is not ready to accept Ceph command but the cluster exists
		if !cephBlockPool.GetDeletionTimestamp().IsZero() && !cephClusterExists {
			// don't leak the health checker routine if we are force-deleting
			r.cancelMirrorMonitoring(cephBlockPool)

			// Remove finalizer
			err = opcontroller.RemoveFinalizer(r.opManagerContext, r.client, cephBlockPool)
			if err != nil {
				return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrap(err, "failed to remove finalizer")
			}

			// Return and do not requeue. Successful deletion.
			return reconcile.Result{}, *cephBlockPool, nil
		}
		return reconcileResponse, *cephBlockPool, nil
	}

	// Populate clusterInfo during each reconcile
	clusterInfo, _, _, err := mon.LoadClusterInfo(r.context, r.opManagerContext, request.NamespacedName.Namespace)
	if err != nil {
		return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrap(err, "failed to populate cluster info")
	}
	r.clusterInfo = clusterInfo
	r.clusterInfo.NetworkSpec = cephCluster.Spec.Network

	// Initialize the channel for this pool
	// This allows us to track multiple CephBlockPool in the same namespace
	blockPoolChannelKey := blockPoolChannelKeyName(cephBlockPool)
	_, blockPoolContextsExists := r.blockPoolContexts[blockPoolChannelKey]
	if !blockPoolContextsExists {
		internalCtx, internalCancel := context.WithCancel(r.opManagerContext)
		r.blockPoolContexts[blockPoolChannelKey] = &blockPoolHealth{
			internalCtx:    internalCtx,
			internalCancel: internalCancel,
		}
	}

	// Set the name of the Ceph pool to be reconciled internally, which
	// may be different from the pool CR name.
	cephBlockPool.Spec.Name = getCephName(*cephBlockPool)

	// DELETE: the CR was deleted
	if !cephBlockPool.GetDeletionTimestamp().IsZero() {
		deps, err := cephBlockPoolDependents(r.context, r.clusterInfo, cephBlockPool)
		if err != nil {
			return reconcile.Result{}, *cephBlockPool, err
		}
		if !deps.Empty() {
			err := reporting.ReportDeletionBlockedDueToDependents(r.opManagerContext, logger, r.client, cephBlockPool, deps)
			return opcontroller.WaitForRequeueIfFinalizerBlocked, *cephBlockPool, err
		}
		reporting.ReportDeletionNotBlockedDueToDependents(r.opManagerContext, logger, r.client, r.recorder, cephBlockPool)
		// If the ceph block pool is still in the map, we must remove it during CR deletion
		// We must remove it first otherwise the checker will panic since the status/info will be nil
		r.cancelMirrorMonitoring(cephBlockPool)

		r.recorder.Event(cephBlockPool, corev1.EventTypeNormal, string(cephv1.ReconcileStarted), "starting blockpool deletion")

		logger.Infof("deleting pool %q", cephBlockPool.Name)
		poolSpec := cephBlockPool.Spec.ToNamedPoolSpec()
		err = deletePool(r.context, clusterInfo, &poolSpec)
		if err != nil {
			return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrapf(err, "failed to delete pool %q. ", cephBlockPool.Name)
		}

		// disable RBD stats collection if cephBlockPool was deleted
		if err := configureRBDStats(r.context, clusterInfo); err != nil {
			logger.Errorf("failed to disable stats collection for pool(s). %v", err)
		}

		// Remove finalizer
		err = opcontroller.RemoveFinalizer(r.opManagerContext, r.client, cephBlockPool)
		if err != nil {
			r.recorder.Event(cephBlockPool, corev1.EventTypeWarning, string(cephv1.ReconcileFailed), "failed to remove finalizer")
			return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrap(err, "failed to remove finalizer")
		}

		r.recorder.Event(cephBlockPool, corev1.EventTypeNormal, string(cephv1.ReconcileSucceeded), "successfully removed finalizer")

		// Return and do not requeue. Successful deletion.
		return reconcile.Result{}, *cephBlockPool, nil
	}

	// validate the pool settings
	if err := validatePool(r.context, clusterInfo, &cephCluster.Spec, cephBlockPool); err != nil {
		if strings.Contains(err.Error(), opcontroller.UninitializedCephConfigError) {
			logger.Info(opcontroller.OperatorNotInitializedMessage)
			return opcontroller.WaitForRequeueIfOperatorNotInitialized, *cephBlockPool, nil
		}
		return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrapf(err, "invalid pool CR %q spec", cephBlockPool.Name)
	}

	// Get CephCluster version
	cephVersion, err := opcontroller.GetImageVersion(cephCluster)
	if err != nil {
		return opcontroller.ImmediateRetryResult, *cephBlockPool, errors.Wrapf(err, "failed to fetch ceph version from cephcluster %q", cephCluster.Name)
	}
	r.clusterInfo.CephVersion = *cephVersion

	// CREATE/UPDATE
	reconcileResponse, err = r.reconcileCreatePool(clusterInfo, &cephCluster.Spec, cephBlockPool)
	if err != nil {
		if strings.Contains(err.Error(), opcontroller.UninitializedCephConfigError) {
			logger.Info(opcontroller.OperatorNotInitializedMessage)
			return opcontroller.WaitForRequeueIfOperatorNotInitialized, *cephBlockPool, nil
		}
		updateStatus(r.opManagerContext, r.client, request.NamespacedName, cephv1.ConditionFailure, nil, k8sutil.ObservedGenerationNotAvailable)
		return reconcileResponse, *cephBlockPool, errors.Wrapf(err, "failed to create pool %q.", cephBlockPool.GetName())
	}

	// enable/disable RBD stats collection based on cephBlockPool spec
	if err := configureRBDStats(r.context, clusterInfo); err != nil {
		return reconcile.Result{}, *cephBlockPool, errors.Wrap(err, "failed to enable/disable stats collection for pool(s)")
	}

	poolSpec := cephBlockPool.Spec.ToNamedPoolSpec()
	checker := newMirrorChecker(r.context, r.client, r.clusterInfo, request.NamespacedName, &poolSpec)
	// ADD PEERS
	logger.Debug("reconciling create rbd mirror peer configuration")
	if cephBlockPool.Spec.Mirroring.Enabled {
		// Always create a bootstrap peer token in case another cluster wants to add us as a peer
		reconcileResponse, err = opcontroller.CreateBootstrapPeerSecret(r.context, clusterInfo, cephBlockPool, k8sutil.NewOwnerInfo(cephBlockPool, r.scheme))
		if err != nil {
			updateStatus(r.opManagerContext, r.client, request.NamespacedName, cephv1.ConditionFailure, nil, k8sutil.ObservedGenerationNotAvailable)
			return reconcileResponse, *cephBlockPool, errors.Wrapf(err, "failed to create rbd-mirror bootstrap peer for pool %q.", cephBlockPool.GetName())
		}

		// Check if rbd-mirror CR and daemons are running
		logger.Debug("listing rbd-mirror CR")
		// Run the goroutine to update the mirroring status
		if !cephBlockPool.Spec.StatusCheck.Mirror.Disabled {
			// Start monitoring of the pool
			if r.blockPoolContexts[blockPoolChannelKey].started {
				logger.Debug("pool monitoring go routine already running!")
			} else {
				go checker.checkMirroring(r.blockPoolContexts[blockPoolChannelKey].internalCtx)
				r.blockPoolContexts[blockPoolChannelKey].started = true
			}
		}

		// Add bootstrap peer if any
		logger.Debug("reconciling ceph bootstrap peers import")
		reconcileResponse, err = r.reconcileAddBoostrapPeer(cephBlockPool, request.NamespacedName)
		if err != nil {
			return reconcileResponse, *cephBlockPool, errors.Wrap(err, "failed to add ceph rbd mirror peer")
		}

		// ReconcilePoolIDMap updates the `rook-ceph-csi-mapping-config` with local and peer cluster pool ID map
		err = peermap.ReconcilePoolIDMap(r.opManagerContext, r.context, r.clusterInfo, cephBlockPool)
		if err != nil {
			return reconcileResponse, *cephBlockPool, errors.Wrapf(err, "failed to update pool ID mapping config for the pool %q", cephBlockPool.Name)
		}

		// update ObservedGeneration in status at the end of reconcile
		// Set Ready status, we are done reconciling
		updateStatus(r.opManagerContext, r.client, request.NamespacedName, cephv1.ConditionReady, opcontroller.GenerateStatusInfo(cephBlockPool), observedGeneration)

		// If not mirrored there is no Status Info field to fulfil
	} else {
		// update ObservedGeneration in status at the end of reconcile
		// Set Ready status, we are done reconciling
		updateStatus(r.opManagerContext, r.client, request.NamespacedName, cephv1.ConditionReady, nil, observedGeneration)

		// Stop monitoring the mirroring status of this pool
		if blockPoolContextsExists && r.blockPoolContexts[blockPoolChannelKey].started {
			r.cancelMirrorMonitoring(cephBlockPool)
			// Reset the MirrorHealthCheckSpec
			checker.updateStatusMirroring(nil, nil, nil, "")
		}
	}

	// Return and do not requeue
	logger.Debug("done reconciling")
	return reconcile.Result{}, *cephBlockPool, nil
}

func (r *ReconcileCephBlockPool) reconcileCreatePool(clusterInfo *cephclient.ClusterInfo, cephCluster *cephv1.ClusterSpec, cephBlockPool *cephv1.CephBlockPool) (reconcile.Result, error) {
	poolSpec := cephBlockPool.Spec.ToNamedPoolSpec()
	err := createPool(r.context, clusterInfo, cephCluster, &poolSpec)
	if err != nil {
		return opcontroller.ImmediateRetryResult, errors.Wrapf(err, "failed to create pool %q.", cephBlockPool.GetName())
	}

	// Let's return here so that on the initial creation we don't check for update right away
	return reconcile.Result{}, nil
}

func getCephName(pool cephv1.CephBlockPool) string {
	// If the name is not overridden in the pool spec.name, set it to the name of the pool CR
	if pool.Spec.Name == "" {
		return pool.Name
	}
	if pool.Spec.Name == pool.Name {
		// The same pool name is set, nothing is being overridden
		return pool.Name
	}
	// The pool names that can be allowed to be overridden are enforced in the crd schema
	return pool.Spec.Name
}

// Create the pool
func createPool(context *clusterd.Context, clusterInfo *cephclient.ClusterInfo, clusterSpec *cephv1.ClusterSpec, p *cephv1.NamedPoolSpec) error {
	// Set the application name to rbd by default, but override for special pools
	appName := poolApplicationNameRBD
	if p.Name == "device_health_metrics" {
		appName = "mgr_devicehealth"
	} else if p.Name == ".mgr" {
		appName = "mgr"
	} else if p.Name == ".nfs" {
		appName = "nfs"
	}

	// create the pool
	logger.Infof("creating pool %q in namespace %q", p.Name, clusterInfo.Namespace)
	if err := cephclient.CreatePool(context, clusterInfo, clusterSpec, *p, appName); err != nil {
		return errors.Wrapf(err, "failed to create pool %q", p.Name)
	}

	logger.Infof("initializing pool %q", p.Name)
	args := []string{"pool", "init", p.Name}
	output, err := cephclient.NewRBDCommand(context, clusterInfo, args).Run()
	if err != nil {
		return errors.Wrapf(err, "failed to initialize pool %q. %s", p.Name, string(output))
	}
	logger.Infof("successfully initialized pool %q", p.Name)

	return nil
}

// Delete the pool
func deletePool(context *clusterd.Context, clusterInfo *cephclient.ClusterInfo, p *cephv1.NamedPoolSpec) error {
	pools, err := cephclient.ListPoolSummaries(context, clusterInfo)
	if err != nil {
		return errors.Wrap(err, "failed to list pools")
	}

	// Only delete the pool if it exists...
	for _, pool := range pools {
		if pool.Name == p.Name {
			err := cephclient.DeletePool(context, clusterInfo, p.Name)
			if err != nil {
				return errors.Wrapf(err, "failed to delete pool %q", p.Name)
			}
		}
	}

	return nil
}

func configureRBDStats(clusterContext *clusterd.Context, clusterInfo *cephclient.ClusterInfo) error {
	logger.Debug("configuring RBD per-image IO statistics collection")
	namespaceListOpt := client.InNamespace(clusterInfo.Namespace)
	cephBlockPoolList := &cephv1.CephBlockPoolList{}
	var enableStatsForPools []string
	err := clusterContext.Client.List(clusterInfo.Context, cephBlockPoolList, namespaceListOpt)
	if err != nil {
		return errors.Wrap(err, "failed to retrieve list of CephBlockPool")
	}
	for _, cephBlockPool := range cephBlockPoolList.Items {
		if cephBlockPool.GetDeletionTimestamp() == nil && cephBlockPool.Spec.EnableRBDStats {
			// add to list of CephBlockPool with enableRBDStats set to true and not marked for deletion
			enableStatsForPools = append(enableStatsForPools, getCephName(cephBlockPool))
		}
	}
	logger.Debugf("RBD per-image IO statistics will be collected for pools: %v", enableStatsForPools)
	monStore := config.GetMonStore(clusterContext, clusterInfo)
	if len(enableStatsForPools) == 0 {
		err = monStore.Delete("mgr.", "mgr/prometheus/rbd_stats_pools")
	} else {
		err = monStore.Set("mgr.", "mgr/prometheus/rbd_stats_pools", strings.Join(enableStatsForPools, ","))
	}
	if err != nil {
		return errors.Wrapf(err, "failed to enable rbd_stats_pools")
	}
	logger.Debug("configured RBD per-image IO statistics collection")
	return nil
}

func blockPoolChannelKeyName(p *cephv1.CephBlockPool) string {
	return types.NamespacedName{Namespace: p.Namespace, Name: p.Name}.String()
}

// cancel mirror monitoring. This is a noop if monitoring is not running.
func (r *ReconcileCephBlockPool) cancelMirrorMonitoring(cephBlockPool *cephv1.CephBlockPool) {
	channelKey := blockPoolChannelKeyName(cephBlockPool)

	_, poolContextExists := r.blockPoolContexts[channelKey]
	if poolContextExists {
		// Cancel the context to stop the go routine
		r.blockPoolContexts[channelKey].internalCancel()

		// Remove ceph block pool from the map
		delete(r.blockPoolContexts, channelKey)
	}
}