Checkpoint scheduled clean¶

Problem description¶

Karbor provides Operation Engine Service to support the scheduled operation for a protection plan. The scheduled operations will create lots of checkpoints as the triggers define the rules. The checkpoints will be created every day or every week or every month. Currently Karbor has no automatic clean feature and policy for the end user.

Use Cases¶

With more and more checkpoints created in the Bank, the end users need more and more storage capacity, and some checkpoints are not meaningful and necessary for the end users. The checkpoint scheduled clean feature is necessary to satisfy the end user’s requirement.

Proposed change¶

Karbor provides the end users some settings including max_backups and retention_duration for the scheduled operation. Karbor could clean the deprecated checkpoints which are created by the scheduled operation automatically as the end users define.

1. max_backups: the max amount of checkpoints.which are created by the scheduled operation. e.g. 10.
2. retention_duration: the retention time of checkpoints which are created by the scheduled operation. e.g. 20 weeks.

Karbor provides the default values for these two settings. The default value is -1, which means Karbor will not clean the checkpoints by default. When the end users launch a scheduled operation, they can input the values of these two settings and invoke the scheduled operation RESTful API.

Meanwhile a database table called checkpoint_records will be created to store the checkpoints which are created by the scheduled protect or protect now. This table’s data will keep a real-time mapping to the checkpoints in the Bank.

When the Operation Engine Service triggers an operation, and the operation will invoke the REST API to create a checkpoint, and the API Service will launch a checkpoint creation workflow and save the checkpoint information into database table called checkpoint_records. If the end users delete a checkpoint, the API Service will launch a checkpoint deletion workflow and delete this checkpoint information from the database.

When the operation is triggered in the Operation Engine Service, the operation could get the values of max_backups and retention_duration from the scheduled operation, and then the operation will judge these two values to confirm whether it will invoke the REST API to clean the deprecated checkpoints.

**POST** : /v1/{project_id}/scheduled_operations
```json
{
    ...
    "operation_definition": {
        ...
        "max_backups": 10,
        "retention_duration": 20
    }
}
```

Implementation¶

Dependencies¶

None

Testing¶

Unit tests in Karbor and the python-karborclient.

Documentation Impact¶

New docs to explain how to use the API.

References¶

None