Add support for S3Repositories (#345)

api/v1beta1/solrbackup_types.go

@@ -47,6 +47,10 @@ type SolrBackupSpec struct {
 	// +optional
 	Collections []string `json:"collections,omitempty"`
 
+	// The location to store the backup in the specified backup repository.
+	// +optional
+	Location string `json:"location,omitempty"`
+
 	// Persistence is the specification on how to persist the backup data.
 	// +optional
 	Persistence *PersistenceSource `json:"persistence,omitempty"`
@@ -202,7 +206,8 @@ type SolrBackupStatus struct {
 	CollectionBackupStatuses []CollectionBackupStatus `json:"collectionBackupStatuses,omitempty"`
 
 	// Whether the backups are in progress of being persisted
-	PersistenceStatus BackupPersistenceStatus `json:"persistenceStatus"`
+	// +optional
+	PersistenceStatus *BackupPersistenceStatus `json:"persistenceStatus,omitempty"`
 
 	// Version of the Solr being backed up
 	// +optional
@@ -221,6 +226,10 @@ type CollectionBackupStatus struct {
 	// Solr Collection name
 	Collection string `json:"collection"`
 
+	// BackupName of this collection's backup in Solr
+	// +optional
+	BackupName string `json:"backupName,omitempty"`
+
 	// Whether the collection is being backed up
 	// +optional
 	InProgress bool `json:"inProgress,omitempty"`
@@ -284,7 +293,7 @@ func (sb *SolrBackup) SharedLabelsWith(labels map[string]string) map[string]stri
 	return newLabels
 }
 
-// HeadlessServiceName returns the name of the headless service for the cloud
+// PersistenceJobName returns the name of the persistence job for the backup
 func (sb *SolrBackup) PersistenceJobName() string {
 	return fmt.Sprintf("%s-solr-backup-persistence", sb.GetName())
 }

api/v1beta1/solrcloud_types.go

@@ -391,6 +391,10 @@ type SolrBackupRepository struct {
 	//+optional
 	GCS *GcsRepository `json:"gcs,omitempty"`
 
+	// An S3Repository for Solr to use when backing up and restoring collections.
+	//+optional
+	S3 *S3Repository `json:"s3,omitempty"`
+
 	// Allows specification of a "repository" for Solr to use when backing up data "locally".
 	// Repositories defined here are considered "managed" and can take advantage of special operator features, such as
 	// post-backup compression.
@@ -410,6 +414,54 @@ type GcsRepository struct {
 	BaseLocation string `json:"baseLocation,omitempty"`
 }
 
+type S3Repository struct {
+	// The S3 region to store the backup data in
+	Region string `json:"region"`
+
+	// The name of the S3 bucket that all backup data will be stored in
+	Bucket string `json:"bucket"`
+
+	// Options for specifying S3Credentials. This is optional in case you want to mount this information yourself.
+	// However, if you do not include these credentials, and you do not load them yourself via a mount or EnvVars,
+	// you will likely see errors when taking s3 backups.
+	//
+	// If running in EKS, you can create an IAMServiceAccount that uses a role permissioned for this S3 bucket.
+	// Then use that serviceAccountName for your SolrCloud, and the credentials should be auto-populated.
+	//
+	// +optional
+	Credentials *S3Credentials `json:"credentials,omitempty"`
+
+	// An already-created chroot within the bucket to store data in. Defaults to the root path "/" if not specified.
+	// +optional
+	BaseLocation string `json:"baseLocation,omitempty"`
+
+	// The full endpoint URL to use when connecting with S3 (or a supported S3 compatible interface)
+	// +optional
+	Endpoint string `json:"endpoint,omitempty"`
+
+	// The full proxy URL to use when connecting with S3
+	// +optional
+	ProxyUrl string `json:"proxyUrl,omitempty"`
+}
+
+type S3Credentials struct {
+	// The name & key of a Kubernetes secret holding an AWS Access Key ID
+	// +optional
+	AccessKeyIdSecret *corev1.SecretKeySelector `json:"accessKeyIdSecret,omitempty"`
+
+	// The name & key of a Kubernetes secret holding an AWS Secret Access Key
+	// +optional
+	SecretAccessKeySecret *corev1.SecretKeySelector `json:"secretAccessKeySecret,omitempty"`
+
+	// The name & key of a Kubernetes secret holding an AWS Session Token
+	// +optional
+	SessionTokenSecret *corev1.SecretKeySelector `json:"sessionTokenSecret,omitempty"`
+
+	// The name & key of a Kubernetes secret holding an AWS credentials file
+	// +optional
+	CredentialsFileSecret *corev1.SecretKeySelector `json:"credentialsFileSecret,omitempty"`
+}
+
 type ManagedRepository struct {
 	// This is a volumeSource for a volume that will be mounted to all solrNodes to store backups and load restores.
 	// The data within the volume will be namespaced for this instance, so feel free to use the same volume for multiple clouds.

config/crd/bases/solr.apache.org_solrbackups.yaml

@@ -67,6 +67,9 @@ spec:
                 items:
                   type: string
                 type: array
+              location:
+                description: The location to store the backup in the specified backup repository.
+                type: string
               persistence:
                 description: Persistence is the specification on how to persist the backup data.
                 properties:
@@ -1068,6 +1071,9 @@ spec:
                     asyncBackupStatus:
                       description: The status of the asynchronous backup call to solr
                       type: string
+                    backupName:
+                      description: BackupName of this collection's backup in Solr
+                      type: string
                     collection:
                       description: Solr Collection name
                       type: string
@@ -1127,7 +1133,6 @@ spec:
                 description: Whether the backup was successful
                 type: boolean
             required:
-            - persistenceStatus
             - solrVersion
             type: object
         type: object

config/crd/bases/solr.apache.org_solrclouds.yaml

@@ -1022,6 +1022,92 @@ spec:
                     name:
                       description: 'A name used to identify this local storage profile.  Values should follow RFC-1123.  (See here for more details: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#dns-label-names)'
                       type: string
+                    s3:
+                      description: An S3Repository for Solr to use when backing up and restoring collections.
+                      properties:
+                        baseLocation:
+                          description: An already-created chroot within the bucket to store data in. Defaults to the root path "/" if not specified.
+                          type: string
+                        bucket:
+                          description: The name of the S3 bucket that all backup data will be stored in
+                          type: string
+                        credentials:
+                          description: "Options for specifying S3Credentials. This is optional in case you want to mount this information yourself. However, if you do not include these credentials, and you do not load them yourself via a mount or EnvVars, you will likely see errors when taking s3 backups. \n If running in EKS, you can create an IAMServiceAccount that uses a role permissioned for this S3 bucket. Then use that serviceAccountName for your SolrCloud, and the credentials should be auto-populated."
+                          properties:
+                            accessKeyIdSecret:
+                              description: The name & key of a Kubernetes secret holding an AWS Access Key ID
+                              properties:
+                                key:
+                                  description: The key of the secret to select from.  Must be a valid secret key.
+                                  type: string
+                                name:
+                                  description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names TODO: Add other useful fields. apiVersion, kind, uid?'
+                                  type: string
+                                optional:
+                                  description: Specify whether the Secret or its key must be defined
+                                  type: boolean
+                              required:
+                              - key
+                              type: object
+                            credentialsFileSecret:
+                              description: The name & key of a Kubernetes secret holding an AWS credentials file
+                              properties:
+                                key:
+                                  description: The key of the secret to select from.  Must be a valid secret key.
+                                  type: string
+                                name:
+                                  description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names TODO: Add other useful fields. apiVersion, kind, uid?'
+                                  type: string
+                                optional:
+                                  description: Specify whether the Secret or its key must be defined
+                                  type: boolean
+                              required:
+                              - key
+                              type: object
+                            secretAccessKeySecret:
+                              description: The name & key of a Kubernetes secret holding an AWS Secret Access Key
+                              properties:
+                                key:
+                                  description: The key of the secret to select from.  Must be a valid secret key.
+                                  type: string
+                                name:
+                                  description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names TODO: Add other useful fields. apiVersion, kind, uid?'
+                                  type: string
+                                optional:
+                                  description: Specify whether the Secret or its key must be defined
+                                  type: boolean
+                              required:
+                              - key
+                              type: object
+                            sessionTokenSecret:
+                              description: The name & key of a Kubernetes secret holding an AWS Session Token
+                              properties:
+                                key:
+                                  description: The key of the secret to select from.  Must be a valid secret key.
+                                  type: string
+                                name:
+                                  description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names TODO: Add other useful fields. apiVersion, kind, uid?'
+                                  type: string
+                                optional:
+                                  description: Specify whether the Secret or its key must be defined
+                                  type: boolean
+                              required:
+                              - key
+                              type: object
+                          type: object
+                        endpoint:
+                          description: The full endpoint URL to use when connecting with S3 (or a supported S3 compatible interface)
+                          type: string
+                        proxyUrl:
+                          description: The full proxy URL to use when connecting with S3
+                          type: string
+                        region:
+                          description: The S3 region to store the backup data in
+                          type: string
+                      required:
+                      - bucket
+                      - region
+                      type: object
                   required:
                   - name
                   type: object

controllers/solrbackup_controller.go

@@ -131,15 +131,17 @@ func (r *SolrBackupReconciler) Reconcile(ctx context.Context, req ctrl.Request)
 	if backup.Status.Finished && backup.Status.FinishTime == nil {
 		now := metav1.Now()
 		backup.Status.FinishTime = &now
-		backup.Status.Successful = backup.Status.PersistenceStatus.Successful
+		if backup.Spec.Persistence != nil {
+			backup.Status.Successful = backup.Status.PersistenceStatus.Successful
+		}
 	}
 
-	if !reflect.DeepEqual(oldStatus, backup.Status) {
+	if !reflect.DeepEqual(oldStatus, &backup.Status) {
 		logger.Info("Updating status for solr-backup")
 		err = r.Status().Update(ctx, backup)
 	}
 
-	if backup.Status.Finished {
+	if err != nil && backup.Status.Finished {
 		requeueOrNot = reconcile.Result{}
 	}
 
@@ -187,12 +189,13 @@ func (r *SolrBackupReconciler) reconcileSolrCloudBackup(ctx context.Context, bac
 	// This should only occur before the backup processes have been started
 	if backup.Status.SolrVersion == "" {
 		// Prep the backup directory in the persistentVolume
-		err := util.EnsureDirectoryForBackup(solrCloud, backupRepository, backup.Name, r.config)
+		err = util.EnsureDirectoryForBackup(solrCloud, backupRepository, backup, r.config)
 		if err != nil {
 			return solrCloud, collectionBackupsFinished, actionTaken, err
 		}
 
 		// Make sure that all solr nodes are active and have the backupRestore shared volume mounted
+		// TODO: we do not need all replicas to be healthy. We should just check that leaders exist for all shards. (or just let Solr do that)
 		cloudReady := solrCloud.Status.BackupRestoreReady && (solrCloud.Status.Replicas == solrCloud.Status.ReadyReplicas)
 		if !cloudReady {
 			logger.Info("Cloud not ready for backup backup", "solrCloud", solrCloud.Name)
@@ -238,11 +241,12 @@ func reconcileSolrCollectionBackup(backup *solrv1beta1.SolrBackup, solrCloud *so
 		if started && collectionBackupStatus.StartTime == nil {
 			collectionBackupStatus.StartTime = &now
 		}
+		collectionBackupStatus.BackupName = util.FullCollectionBackupName(collection, backup.Name)
 	} else if collectionBackupStatus.InProgress {
 		// Check the state of the backup, when it is in progress, and update the state accordingly
-		finished, successful, asyncStatus, error := util.CheckBackupForCollection(solrCloud, collection, backup.Name, httpHeaders, logger)
-		if error != nil {
-			return false, error
+		finished, successful, asyncStatus, err := util.CheckBackupForCollection(solrCloud, collection, backup.Name, httpHeaders, logger)
+		if err != nil {
+			return false, err
 		}
 		collectionBackupStatus.Finished = finished
 		if finished {
@@ -271,6 +275,9 @@ func reconcileSolrCollectionBackup(backup *solrv1beta1.SolrBackup, solrCloud *so
 }
 
 func (r *SolrBackupReconciler) persistSolrCloudBackups(ctx context.Context, backup *solrv1beta1.SolrBackup, solrCloud *solrv1beta1.SolrCloud, logger logr.Logger) (err error) {
+	if backup.Status.PersistenceStatus == nil {
+		backup.Status.PersistenceStatus = &solrv1beta1.BackupPersistenceStatus{}
+	}
 	if backup.Status.PersistenceStatus.Finished {
 		return nil
 	}