Point-in-time Recovery (PITR)

Recovering through the UI

After your deployment's first backup is complete, the point-in-time recovery option will display in the UI. To initiate a PITR, enter the time that you want to restore back to in Coordinated Universal Time.

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.

To restore to the most recent available time, select that option. Clicking Restore brings up the options for your recovery. Enter a name, select the version, region, and allocated resources for the new deployment. Click Recover to start the process.

If you use Key Protect and have a key, you must use the CLI to recover, and a command is provided for your convenience.

Recovery with the CLI

The Resource Controller supports provisioning of database deployments, and provisioning and restoring are the responsibility of the Resource Controller CLI. Use the resource service-instance-create command.

For PITR, use the point_in_time_recovery_time and point_in_time_recovery_deployment_id parameters. The point_in_time_recovery_deployment_id is the source deployment's ID and point_in_time_recovery_time is the timestamp in Coordinated Universal Time you want to restore to. To restore to the latest available point-in-time, use "point_in_time_recovery_time":" ".

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.

ibmcloud resource service-instance-create <SERVICE_INSTANCE_NAME> <service-id> <region> -p '{"point_in_time_recovery_deployment_id":"DEPLOYMENT_ID", "point_in_time_recovery_time":"TIMESTAMP", "version":" "}'

A pre-formatted command for a specific backup or PITR is available in detailed view of the backup.

When restoring through the CLI, optional parameters are available. Use them to customize resources or use a Key Protect key for BYOK encryption on the new deployment.

ibmcloud resource service-instance-create <SERVICE_INSTANCE_NAME> <service-id> standard <region> <--service-endpoints SERVICE_ENDPOINTS_TYPE> -p
'{"point_in_time_recovery_deployment_id":"DEPLOYMENT_ID", "point_in_time_recovery_time":"TIMESTAMP","key_protect_key":"KEY_PROTECT_KEY_CRN", "members_disk_allocation_mb":"DESIRED_DISK_IN_MB", "members_memory_allocation_mb":"DESIRED_MEMORY_IN_MB", "members_cpu_allocation_count":"NUMBER_OF_CORES", "version":" "}'

Recovery with the API

The Resource Controller supports provisioning of database deployments, and provisioning and restoring are the responsibility of the Resource Controller API. You need to complete the necessary steps to use the resource controller API before you can use it to restore from a backup.

After you have all the information, the create request is a POST to the /resource_instances endpoint.

curl -X POST \
  https://resource-controller.cloud.ibm.com/v2/resource_instances \
  -H 'Authorization: Bearer <>' \
  -H 'Content-Type: application/json' \
    -d '{
    "name": "<SERVICE_INSTANCE_NAME>",
    "target": "<region>",
    "resource_group": "<your-resource-group>",
    "resource_plan_id": "<service-id>"
    "point_in_time_recovery_time":"<TIMESTAMP>",
    "point_in_time_recovery_deployment_id":"<DEPLOYMENT_ID>"
  }'

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.

The parameters name, target, resource_group, and resource_plan_id are all required. The target is the region where you want the new deployment to be located, which can be a different region from the source deployment. Cross region restores are supported, except for restoring a eu-de backup to another region.

For PITR, use the point_in_time_recovery_time and point_in_time_recovery_deployment_id parameters. The point_in_time_recovery_deployment_id is the source deployment's ID and point_in_time_recovery_time is the timestamp in Coordinated Universal Time you want to restore to. To restore to the latest available point-in-time use "point_in_time_recovery_time":" ".

If you need to adjust resources or use a Key Protect key, add the optional parameters key_protect_key, members_disk_allocation_mb, members_memory_allocation_mb, and/or members_cpu_allocation_count, and their values to the body of the request.

Restoring a backup with Terraform

Before restoring, ensure that your point_in_time_recovery_time is no older than one week. If the timestamp is any older than 7 days, down to the second, validation fails.

Use the ibm_database_point_in_time_recovery data source to restore your database instance with point_in_time_recovery.

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.

Your Terraform script looks like this.

terraform {
  required_providers {
     ibm = {
       version = "1.44.3"
       source  = "IBM-Cloud/ibm"
    }
  }
}

variable "ibmcloud_api_key" {
  description = "<Enter your IBM Cloud API Key>"
}

provider "ibm" {
  region = "us-south"
  ibmcloud_api_key = var.ibmcloud_api_key
}

data "ibm_resource_group" "default_group" {
  is_default = true
}

resource "ibm_database" "mongodb_enterprise" {
  resource_group_id = data.ibm_resource_group.default_group.id
  name              = "testing-mongodb-pitr"
  service           = "databases-for-mongodb"
  plan              = "enterprise"
  location          = "us-south"

  point_in_time_recovery_deployment_id = "<crn>"
  point_in_time_recovery_time = "2022-09-14T14:47:45Z"

  group {
    group_id = "member"

    members {
      allocation_count = 3
    }

    cpu {
      allocation_count = 6
    }

    memory {
      allocation_mb = 14336
    }

    disk {
      allocation_mb = 20480
    }
  }
}

Point-in-time Recovery (PITR) Offline Restore through the UI

Initiate an Offline Restore through the IBM Cloud Dashboard just as you would for a standard PITR. Choose the third Point-in-time recovery option.

Point-in-time Recovery (PITR) Offline Restore through the CLI

Initiate an Offline Restore through the IBM Cloud CLI using a command like:

ibmcloud resource service-instance-create 3-aug-pitr-dr-cli-lorna databases-for-mongodb-development enterprise us-south -p '{"point_in_time_recovery_deployment_id":"crn:v1:staging:public:databases-for-mongodb-development:us-south:a/b9552134280015ebfde430a819fa4bb3:4e765b52-72f5-4664-9cc0-e1e2f6e1ecde::","point_in_time_recovery_time": "", "offline_restore": true}'

Specify the following parameters:

• point_in_time_recovery_deployment_id - This is the source CRN.
• point_in_time_recovery_time - Leave this blank, "".
• offline_restore - Set this value to true.

The command output looks like:

Creating service instance <INSTANCE_NAME> in resource group Default of account <ACCOUNT> as <USER>...
OK
Service instance <INSTANCE_NAME> was created.

Name:                <INSTANCE_NAME>
ID:                  crn:v1:staging:public:databases-for-mongodb-development:us-south:a/b9552134280015ebfde430a819fa4bb3:3558fedf-8170-4ab7-99ad-6baab449551b::
GUID:                3558fedf-8170-4ab7-99ad-6baab449551b
Location:            <LOCATION>
State:               provisioning
Type:                service_instance
Sub Type:            Public
Service Endpoints:   public
Allow Cleanup:       false
Locked:              false
Created at:          2023-08-03T09:36:37Z
Updated at:          2023-08-03T09:36:41Z
Last Operation:
                     Status    create in progress
                     Message   Started create instance operation

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.

Point-in-time Recovery (PITR) Offline Restore through the API

The Resource Controller supports provisioning of database deployments, and provisioning and restoring are the responsibility of the Resource Controller API. Complete the necessary steps to use the resource controller API before using it to restore from a backup.

After you have all the information, the create request is a POST to the /resource_instances that will look like:

-p '{"point_in_time_recovery_deployment_id":"<crn_of_source>","point_in_time_recovery_time": "", "offline_restore": true}

Specify the following parameters:

• point_in_time_recovery_deployment_id - This is the source CRN.
• point_in_time_recovery_time - Leave this blank, "".
• offline_restore - Set this value to true.

The point-in-time-recovery timestamp must be formatted as follows: %Y-%m-%dT%H:%M:%SZ.