Replace the limits with the objectIDs in the backup and recover of vi…

…neyard cluster. (#1402)

As titled. The `objectIDs` is more meaningful for users.

Fixes #1322

Signed-off-by: Ye Cao <[email protected]>

charts/vineyard-operator/templates/backup-crd.yaml

@@ -45,8 +45,10 @@ spec:
  properties:
  backupPath:
  type: string
- limit:
- type: integer
+ objecIDs:
+ items:
+ type: string
+ type: array
  persistentVolumeClaimSpec:
  properties:
  accessModes:

docs/notes/cloud-native/vineyard-operator.rst

@@ -1741,9 +1741,10 @@ the vineyard operator. The main fields are described as follows.
  - The namespace of vineyardd cluster.
  - nil
 
- * - limit
- - int
- - The number of objects to be backed up
+ * - objectIDs
+ - []string
+ - The object IDs that need to be backed up.
+ If it is empty, all objects will be backed up.
  - nil
 
  * - backupPath
@@ -1806,7 +1807,6 @@ up the data. The following is the yaml file of the backup:
  spec:
  vineyarddName: vineyardd-sample
  vineyarddNamespace: vineyard-system
- limit: 1000
  backupPath: /var/vineyard/dump
  persistentVolumeSpec:
  storageClassName: manual

k8s/apis/k8s/v1alpha1/backup_types.go

@@ -31,9 +31,10 @@ type BackupSpec struct {
  // +kubebuilder:validation:Required
  VineyarddNamespace string `json:"vineyarddNamespace,omitempty"`
 
- // the number of objects to be backed up
+ // the specific objects to be backed up
+ // if not specified, all objects will be backed up
  // +kubebuilder:validation:Required
- Limit int `json:"limit,omitempty"`
+ ObjectIDs []string `json:"objecIDs,omitempty"`
 
  // the path of backup data
  // +kubebuilder:validation:Required

k8s/cmd/README.md

@@ -165,7 +165,7 @@ vineyardctl create backup [flags]
 ```
  --backup-name string the name of backup job (default "vineyard-backup")
  -h, --help help for backup
- --limit int  the limit of objects to backup (default 1000)
+ --objectIDs strings the specific objects to be backed up
  --path string the path of the backup data
  --pv-pvc-spec string the PersistentVolume and PersistentVolumeClaim of the backup data
  --vineyardd-name string the name of vineyardd
@@ -594,12 +594,12 @@ vineyardctl deploy backup-job [flags]
 ### Examples
 
 ```shell
- # deploy a backup job for the vineyard cluster on kubernetes
- # you could define the pv and pvc spec from json string as follows
+ # deploy a backup job for all vineyard objects of the vineyard
+ # cluster on kubernetes and you could define the pv and pvc
+ # spec from json string as follows
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 \
  --path /var/vineyard/dump \
  --pv-pvc-spec '{
  "pv-spec": {
@@ -632,7 +632,7 @@ vineyardctl deploy backup-job [flags]
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
+ --path /var/vineyard/dump \
  --pv-pvc-spec \
  '
  pv-spec:
@@ -652,21 +652,31 @@ vineyardctl deploy backup-job [flags]
  storage: 1Gi
  '
 
- # deploy a backup job for the vineyard cluster on kubernetes
- # you could define the pv and pvc spec from json file as follows
- # also you could use yaml file instead of json file
+ # deploy a backup job for specific vineyard objects of the vineyard
+ # cluster on kubernetes.
  cat pv-pvc.json | vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
- -
+ --objectIDs "o000018d29207fd01,o000018d80d264010" \
+ --path /var/vineyard/dump
 
  # Assume you have already deployed a pvc named "pvc-sample", you
  # could use them as the backend storage for the backup job as follows
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
+ --path /var/vineyard/dump \
+ --pvc-name pvc-sample
+
+ # The namespace to deploy the backup and recover job must be the same
+ # as the vineyard cluster namespace.
+ # Assume the vineyard cluster is deployed in the namespace "test", you
+ # could deploy the backup job as follows
+ vineyardctl deploy backup-job \
+ --vineyard-deployment-name vineyardd-sample \
+ --vineyard-deployment-namespace test \
+ --namespace test \
+ --path /var/vineyard/dump \
  --pvc-name pvc-sample
 ```
 
@@ -675,7 +685,7 @@ vineyardctl deploy backup-job [flags]
 ```
  --backup-name string the name of backup job (default "vineyard-backup")
  -h, --help help for backup-job
- --limit int  the limit of objects to backup (default 1000)
+ --objectIDs strings the specific objects to be backed up
  --path string the path of the backup data
  --pv-pvc-spec string the PersistentVolume and PersistentVolumeClaim of the backup data
  --pvc-name string the name of an existing PersistentVolumeClaim

k8s/cmd/commands/deploy/deploy_backup_job.go

@@ -41,12 +41,12 @@ var (
  Also, you could also specify the existing pv and pvc name to use`)
 
  deployBackupJobExample = util.Examples(`
- # deploy a backup job for the vineyard cluster on kubernetes
- # you could define the pv and pvc spec from json string as follows
+ # deploy a backup job for all vineyard objects of the vineyard 
+ # cluster on kubernetes and you could define the pv and pvc 
+ # spec from json string as follows
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 \
  --path /var/vineyard/dump \
  --pv-pvc-spec '{
  "pv-spec": {
@@ -79,7 +79,7 @@ var (
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
+ --path /var/vineyard/dump \
  --pv-pvc-spec \
  '
  pv-spec:
@@ -99,23 +99,32 @@ var (
  storage: 1Gi
  '
 
- # deploy a backup job for the vineyard cluster on kubernetes
- # you could define the pv and pvc spec from json file as follows
- # also you could use yaml file instead of json file
+ # deploy a backup job for specific vineyard objects of the vineyard 
+ # cluster on kubernetes.
  cat pv-pvc.json | vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
- -
+ --objectIDs "o000018d29207fd01,o000018d80d264010" \
+ --path /var/vineyard/dump
  
  # Assume you have already deployed a pvc named "pvc-sample", you 
  # could use them as the backend storage for the backup job as follows
  vineyardctl deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 --path /var/vineyard/dump \
+ --path /var/vineyard/dump \
  --pvc-name pvc-sample
- `)
+ 
+ # The namespace to deploy the backup and recover job must be the same
+ # as the vineyard cluster namespace.
+ # Assume the vineyard cluster is deployed in the namespace "test", you
+ # could deploy the backup job as follows
+ vineyardctl deploy backup-job \
+ --vineyard-deployment-name vineyardd-sample \
+ --vineyard-deployment-namespace test \
+ --namespace test \
+ --path /var/vineyard/dump \
+ --pvc-name pvc-sample`)
 )
 
 // deployBackupJobCmd deploy the backup job of vineyard cluster on kubernetes

k8s/cmd/commands/flags/backup_flags.go

@@ -50,8 +50,8 @@ func ApplyBackupNameOpts(cmd *cobra.Command) {
 
 func ApplyBackupCommonOpts(cmd *cobra.Command) {
  cmd.Flags().
- IntVarP(&BackupOpts.Limit, "limit", "", 1000,
- "the limit of objects to backup")
+ StringSliceVarP(&BackupOpts.ObjectIDs, "objectIDs", "", []string{},
+ "the specific objects to be backed up")
  cmd.Flags().
  StringVarP(&BackupOpts.BackupPath, "path", "", "",
  "the path of the backup data")

k8s/config/crd/bases/k8s.v6d.io_backups.yaml

@@ -33,8 +33,10 @@ spec:
  properties:
  backupPath:
  type: string
- limit:
- type: integer
+ objecIDs:
+ items:
+ type: string
+ type: array
  persistentVolumeClaimSpec:
  properties:
  accessModes:

k8s/controllers/k8s/backup_controller.go

@@ -19,6 +19,7 @@ package k8s
 import (
  "context"
  "strconv"
+ "strings"
  "time"
 
  "github.com/pkg/errors"
@@ -54,8 +55,8 @@ type FailoverConfig struct {
 
 // BackupConfig holds all configuration about backup
 type BackupConfig struct {
- Name string
- Limit string
+ Name  string
+ ObjectIDs string
  FailoverConfig
 }
 
@@ -246,7 +247,8 @@ func BuildBackupCfg(c client.Client, name string, backup *k8sv1alpha1.Backup,
  path string, withScheduler bool) (BackupConfig, error) {
  backupCfg := BackupConfig{}
  backupCfg.Name = name
- backupCfg.Limit = strconv.Itoa(backup.Spec.Limit)
+ backupCfg.ObjectIDs = strings.Join(backup.Spec.ObjectIDs, ",")
+
  failoverCfg, err := newFailoverConfig(c, backup, path, withScheduler)
  if err != nil {
  return backupCfg, errors.Wrap(err, "unable to build failover config")

k8s/pkg/templates/backup/job.yaml

@@ -41,8 +41,10 @@ spec:
  image: ghcr.io/v6d-io/v6d/backup-job
  imagePullPolicy: IfNotPresent
  env:
- - name: LIMIT
- value: "{{ $backupConfig.Limit }}"
+ {{- if $backupConfig.ObjectIDs }}
+ - name: ObjectIDs
+ value: "{{ $backupConfig.ObjectIDs }}"
+ {{- end }}
  - name: BACKUP_PATH
  value: {{ $failoverConfig.Path }}
  - name: ENDPOINT

k8s/test/e2e/deploy-raw-backup-and-recover/e2e.yaml

@@ -77,7 +77,7 @@ setup:
  go run k8s/cmd/main.go deploy backup-job \
  --vineyard-deployment-name vineyardd-sample \
  --vineyard-deployment-namespace vineyard-system \
- --limit 1000 \
+ --objectIDs "$localobjectid,$distributedobjectid" \
  --path /var/vineyard/dump \
  --pv-pvc-spec '{
  "pv-spec": {

k8s/test/e2e/failover-demo/backup.py

@@ -24,7 +24,9 @@
 
 env_dist = os.environ
 
-limits = int(env_dist['LIMIT'])
+objectids = None
+if "ObjectIDs" in env_dist:
+ objectids = env_dist["ObjectIDs"]
 path = env_dist['BACKUP_PATH']
 socket = '/var/run/vineyard.sock'
 endpoint = env_dist['ENDPOINT']
@@ -59,13 +61,23 @@
 # get instance id
 instance = vineyard_client.instance_id
 
-objs = vineyard_client.list_objects(pattern='*',limit=limits)
+objIDs = []
+if objectids is not None:
+ objs = objectids.split(',')
+ for _,o in enumerate(objs):
+ objIDs.append(vineyard.ObjectID(o))
+else:
+ objs = vineyard_client.list_objects(pattern='*')
+ for _,o in enumerate(objs):
+ objIDs.append(o.id)
+
+
 # serialize all persistent objects
-for i,o in enumerate(objs):
+for _,o in enumerate(objIDs):
  try:
- meta = vineyard_client.get_meta(o.id, sync_remote=True)
+ meta = vineyard_client.get_meta(o, sync_remote=True)
  if not meta['transient'] and meta['typename'] not in ['vineyard::ParallelStream','vineyard::StreamCollection']:
- objname = str(o.id).split("\"")[1]
+ objname = str(o).split("\"")[1]
  objpath = path + '/' + objname
  # for local objects
  if not meta['global'] and meta['instance_id']==instance:
@@ -74,7 +86,7 @@
  os.makedirs(objpath)
  vineyard.io.serialize(
  objpath,
- o.id,
+ o,
  vineyard_ipc_socket=socket,
  vineyard_endpoint=service,
  )
@@ -85,15 +97,15 @@
  os.makedirs(objpath)
  vineyard.io.serialize(
  objpath,
- o.id,
+ o,
  type="global",
  vineyard_ipc_socket=socket,
  vineyard_endpoint=service,
  deployment='kubernetes',
  hosts=hosts,
  )
  except vineyard.ObjectNotExistsException:
- print(o.id,' trigger ObjectNotExistsException')
+ print(o,' trigger ObjectNotExistsException')
 
 # wait other backup process ready
 print('Succeed',flush=True)

k8s/test/e2e/failover-demo/backup.yaml

@@ -6,7 +6,6 @@ metadata:
 spec:
  vineyarddName: vineyardd-sample
  vineyarddNamespace: vineyard-system
- limit: 1000
  backupPath: /var/vineyard/dump
  persistentVolumeSpec:
  storageClassName: manual

k8s/test/e2e/failover/e2e.yaml

@@ -65,12 +65,12 @@ setup:
  - namespace: vineyard-job
  resource: deployment/build-distributed-object-step2
  for: condition=Available
- - name: install backup
+ - name: install backup with the specific objects
  command: |
  go run k8s/cmd/main.go create backup \
  --vineyardd-name vineyardd-sample \
  --vineyardd-namespace vineyard-system \
- --limit 1000 \
+ --objectIDs $localobjectid,$distributedobjectid \
  --path /var/vineyard/dump \
  --pv-pvc-spec '{
  "pv-spec": {