0

We have a fairly large Windows Server 2012 R2 machine that has Oracle 12c running on it. We used AWS MGN (Application Migration Service) to migrate that on-prem machine into AWS. The migration was tested and is successful - so we have a large Windows 2012 R2 EC2 instance running an Oracle 12c server on AWS now.

I'm experimenting with using AWS Backups to take snapshots of the Oracle instance in an attempt to do a restore on it as the first steps toward a full DR plan. The snapshot restores just fine, and we are able to login to the Windows instance, but it seems that the Oracle tables are in an inconsistent state and I did not get a clean snapshot from AWS backup. Microsoft VSS seems to be enabled on the restored instance.

The question is : Has anyone been successful using AWS Backups to backup and restore an Oracle database on a Windows Server? If so, what changes do we need to make? I included some of the terraform for reference:

resource "aws_backup_plan" "weekly" {
  name = "Legacy${var.environment_tag}Weekly"

  rule {
    enable_continuous_backup = false
    rule_name                = "Legacy${var.environment_tag}Weekly"
    target_vault_name        = aws_backup_vault.legacy.name
    schedule                 = var.backup_plan_weekly_schedule
    start_window             = 60  # minutes
    completion_window        = 180 # minutes

    lifecycle {
      cold_storage_after = 30  # days
      delete_after       = 120 # days
    }

    copy_action {
      destination_vault_arn = aws_backup_vault.legacy.arn

      lifecycle {
        cold_storage_after = 30  # days
        delete_after       = 120 # days
      }
    }

    copy_action {
      destination_vault_arn = aws_backup_vault.secondary.arn

      lifecycle {
        cold_storage_after = 30  # days
        delete_after       = 120 # days
      }
    }
  }

  advanced_backup_setting {
    backup_options = {
      WindowsVSS = "enabled"
    }
    resource_type = "EC2"
  }

  tags = merge(
    local.tags, {
      "Name" = "Legacy${var.environment_tag}Weekly"
    }
  )
}

resource "aws_backup_selection" "weekly" {
  iam_role_arn = aws_iam_role.legacy_backup.arn
  name         = "Legacy${var.environment_tag}Weekly"
  plan_id      = aws_backup_plan.weekly.id

  selection_tag {
    type  = "STRINGEQUALS"
    key   = "AWSBackup"
    value = "weekly"
  }
}

resource "aws_backup_vault" "legacy" {
  name        = "Legacy${var.environment_tag}"
  kms_key_arn = aws_kms_key.mgn.arn

  tags = merge(
    local.tags, {
      "Name" = "Legacy${var.environment_tag}"
    }
  )
}
BPS
  • 607
  • 8
  • 29

1 Answers1

0

It turns out that AWS Backup will keep the database in a 'backup' state even after the snapshot completes, so the restored instance still thinks it's not in a consistent state. The fix for this is to go into the restored instance, start the Oracle SQL prompt and run the command:

alter database end backup;
BPS
  • 607
  • 8
  • 29