Skip to content

Latest commit

 

History

History
126 lines (105 loc) · 4.56 KB

restore-from-backups.md

File metadata and controls

126 lines (105 loc) · 4.56 KB
Raw

Restore from a backup

An Instance can be restored from a backups.oracle.db.anthosapis.com resource representing either a snapshot-based backup or an RMAN backup.

The following variables used in the examples below:

export NAMESPACE=<kubernetes namespace where the instance was created>
export PATH_TO_EL_CARRO_RELEASE=<the complete path to the downloaded release directory>

Locate a backup

The Instance resource contains the ID of the latest backup taken for the instance:

kubectl get instances.oracle.db.anthosapis.com -n $NAMESPACE
NAME     DB ENGINE   VERSION   EDITION      ENDPOINT        URL                DB NAMES      BACKUP ID                        READYSTATUS   READYREASON      DBREADYSTATUS   DBREADYREASON
mydb     Oracle      12.2      Enterprise   mydb-svc.db     10.128.0.33:6021   [pdb1, pdb2]  mydb-20210427-phys-885709718     True          CreateComplete   True            CreateComplete

Alternatively, IDs of older backups can be found by listing the backups.oracle.db.anthosapis.com resources in the same namespace that the database instance belongs to:

kubectl get backups.oracle.db.anthosapis.com -n $NAMESPACE
NAME            INSTANCE NAME   BACKUP TYPE   BACKUP SUBTYPE   DOP   BS/IC   GCS PATH   PHASE       BACKUP ID                        BACKUP TIME
rman1-inst      mydb            Physical      Instance         1     true               Succeeded   mydb-20210427-phys-885709718     20210427210913
snap1           mydb            Snapshot      Instance                                  Succeeded   mydb-20210427-snap-416248334     20210427182828

Prepare an Instance Resource Manifest for Restore

Once the ID for the backup to restore from is determined, you can restore the instance by uncommenting or adding the restore section in the Instance manifest. The four mandatory attributes to uncomment are:

  • backupType (Snapshot or Physical)
  • backupId
  • force
  • requestTime

The backupId comes from the previous step.

To avoid accidental restores, the force attribute needs to be explicitly set to true. Failure to do so trips the safeguard in the Instance controller and leads to an error message stating that you need to be explicit in requesting a restore (and acknowledging the downtime associated with it). Another safeguard is the requestTime attribute which is recommended to be set to the current time. If the same .yaml file that was previously used for a restore operation is sent to Kubernetes with the same value of requestTime, the request will be ignored. To invoke another restore operation, requestTime needs to be updated to the current time - this will ensure that the new request’s timestamp is later than the timestamp of the previous restore operation. Requests with earlier timestamps in requestTime are ignored as well.

Take 12.2 instance as an example.

cat $PATH_TO_EL_CARRO_RELEASE/samples/v1alpha1_instance.yaml
apiVersion: oracle.db.anthosapis.com/v1alpha1
kind: Instance
metadata:
  name: mydb
spec:
  type: Oracle
  version: "12.2"
  edition: Enterprise
  dbDomain: "gke"
  disks:
  - name: DataDisk
    size: 45Gi
    type: pd-standard
    storageClass: "standard-rwo"
  - name: LogDisk
    size: 55Gi
    type: pd-standard
    storageClass: "standard-rwo"
  services:
    Backup: true
    Monitoring: true
    Logging: true
  sourceCidrRanges: [ 0.0.0.0/0 ]
  images:
    # Replace below with the actual URIs hosting the service agent images.
    service: "gcr.io/${PROJECT_ID}/oracle-database-images/oracle-12.2-ee-seeded-${DB}"
  cdbName: ${DB}
  databaseResources:
    requests:
      memory: 4.0Gi

# Uncomment this section to trigger a restore.
  restore:
    backupType: "Snapshot"
    backupId: "mydb-20200705-snap-996678001"
    force: True
#    # once applied, new requests with same or older time will be ignored,
#    # current time can be generated using the command: date -u '+%Y-%m-%dT%H:%M:%SZ'
    requestTime: "2021-05-12T01:23:45Z"
#    # Physical backup specific attributes:
#    dop: 2
#    # The unit for time limit is minutes (but specify just an integer).
#    timeLimitMinutes: 180

Limitations

There are currently limitations for restoring from an RMAN backup:

  • Only backups created with spec.backupset either omitted or set to true can be restored from
  • Only backups created with spec.subType either omitted or set to Instance can be restored from
  • For local backups (ones that don't specify spec.gcsPath attribute and thus do not persist backup data in GCS) restore can be only be done for the latest such backup.