6.4.1 Restoration Procedure

Table 6.6 Areas where Disk Failure Occurs
Pattern	System Area	Data Area
A	Failure	Failure
B	Normal	Failure
C	Failure	Normal

Figure 6.1 Workflow for Restoration after Disk Failure

If a disk failure occurs, perform the restoration procedure described in "Figure 6.1 Workflow for Restoration after Disk Failure".

Check which pattern shown in "Table 6.6 Areas where Disk Failure Occurs" applies for the area where the disk failure occurred.

If pattern A or pattern B applies
Use the data that is backed up to the cloud provider to restore the entire system.
Refer to "6.4.1.2 Restoring from Cloud Provider Data" for details about the restoration procedure.
If pattern C applies
The restoration procedure varies depending on whether a system backup exists.
- If there is no system backup
  Use the data that is backed up to the cloud provider to restore the entire system.
  Refer to "6.4.1.2 Restoring from Cloud Provider Data" for details about the restoration procedure.
- If there is a system backup
  Perform a system restore to restore the system while retaining the data area.
  Refer to "6.4.1.3 Restoring from a System Backup" for details about the restoration procedure.
  Note
  - When backing up the system, make sure that the network MAC address assigned to the virtual appliance is carried over.
  - If you make any structural changes, such as adding or removing a shared folder, after you have backed up the system, restore the entire system according to the procedure described in "6.4.1.2 Restoring from Cloud Provider Data".

6.4.1.2 Restoring from Cloud Provider Data

The procedure for restoring the system using data backed up to the cloud provider is described below.

Replace the failed physical disk with one that operates normally.
From the server virtualization software, delete the virtual machine that is on the physical disk where the failure occurred, and in which this product was running.
Information
If pattern B in "Table 6.6 Areas where Disk Failure Occurs" applies for the area where the disk failure occurred, this step also deletes the virtual disk from the system area that resides on the normal physical disk.
Deploy the virtual appliance of this product on the server virtualization software.
Refer to "2.2 Deploying Virtual Appliances" for details about how to deploy a virtual appliance.
Configure the virtual appliance settings for this product.
Perform the work described in the sections from "2.3 Setting Up Virtual Appliances" to "2.7 Setting NAS Access Users".
Register a cloud provider.
Refer to "3.1 Registering a Cloud Provider" for details about how to register a cloud provider.
Create a datastore on the cloud provider that you registered in step 5.
Refer to "3.2 Registering a Datastore and Cache" for details about how to create a datastore.
Note
- Enter the same content as before the failure occurred for the item.
  In particular, if the content of the following items differ from the content before the failure occurred, the restoration process fails.
  - Cache capacity and Provider name, which are required in the basic settings screen
  - Bucket name, which is required in the bucket selection screen
  - Datastore encryption and Datastore encryption password, which are required in the advanced settings screen
- If the cache capacity is smaller than before the failure occurred, the restoration process fails due to insufficient cache capacity. Delete the datastore and then start the procedure from step 6.
  If a bucket that has no data is selected, the restoration process cannot be performed even by performing step 7. Delete the datastore and then start the procedure from step 5 or step 6.
  For the selected bucket, if the settings for Datastore encryption and Datastore encryption password do not match, step 6 fails. Perform the procedure again from step 6.
- Specify the bucket that was used before the failure occurred.
  If the registration of the datastore and cache is performed, the csgdp03002 message is output. This message indicates that the specified bucket is already in use.
  When restoring from cloud provider data, this message is output because a bucket that has data is specified. Ignore this message and continue the restoration procedure.
  If a bucket that has no data is specified, this message is not output. Check whether the bucket specification is correct.
Use the following procedure to perform meta data recovery for the datastore that was created in step 6. When you perform meta data recovery, the meta data that is in the cloud provider is restored to the cache.
1. Execute the following CSG REST API to perform meta data recovery for the datastore. You can check the ID in the Datastore screen of CSG Web GUI.
```
POST /v1/datastores/{id}/metadata/recovery
```
2. Execute the following CSG REST API periodically as you wait for meta data recovery for the datastore to complete.
```
GET /v1/datastores/{id}/metadata/recovery
```
  You can check the progress of the meta data recovery with the status key for CSG REST API response.
  - While the process is in progress
    The status key is "Active".
  - When this product is accessible with the read-only mode
    The status key is "Active (Readable)".
    Registering the shared folders allows data to be read.
    Use the event log to check the last time when the untransferred data became "None".
    Use the checked time when performing a restoration with the backup software.
  - When the process is completed
    The status key is "N/A".
    Use the event log to check that the meta data recovery is completed.
  - If a process error occurs
    The status key is "Error".
    You can determine the cause of the process error from the event log. Reboot the system after identifying and removing the cause of the error.
    A reboot will automatically restart the meta data recovery.
    If you restart the meta data recovery but it still terminates with an error, you must re-execute the procedure from step 2.
    See
    The event log for this product appears in the Logs panel of the CSG Web GUI dashboard.
    Note
    In the read-only mode state, creating, deleting, and writing files to the shared folders cause errors.
Use the following procedure to allow NAS access for the shared folders on the datastore.
1. Execute the following CSG REST API to check the names of the shared folders in the datastore.
  The "datastore_id" parameter can be checked from the Datastore screen of CSG Web GUI.
```
GET /v1/datastore_folders
```
2. Specify a shared folder name that you checked in step 8-a to register that shared folder.
  Refer to "3.3 Registering a Shared Folder" for details about how to register a shared folder.
  Point
  Perform this procedure for each shared folder in the datastore.
  Note
  For configurable items (shared folder name, owner, and group) of the shared folder, the values must be the same as before the failure.
  During the read-only mode, if you enter content that differs from before the failure, registration of the shared folder terminates with an error.
  If you enter content that differs from before the failure after the recovery is complete, the content you entered will be set to the shared folder. (The shared folder is registered based on the information that you entered.)

See

Refer to the "Reference Guide" for details about the CSG REST APIs mentioned in step 7 and step 8.

Fast Recovery Function

Depending on the amount of data, the recovery may take some time before being complete. This product provides a fast recovery function that can perform read only in advance even if a recovery process is running. This allows the restore to start before the meta data recovery is completed.

In meta data recoveries, the execution state transitions in the order of "Processing (Active)" -> "Read-only mode (Active(Readable))" -> "Completed (N/A)" according to the progress of the process. The recovered data can be referenced in the "Read-only mode (Active(Readable))" and subsequent states, and can be updated in the "Completed (N/A)" state.

Table 6.7 Metadata Recovery Status and Access Type
Status (Status key display)	Type of Access
Status (Status key display)	Reference	Update
Active	NG	NG
Active(Readable)	OK	NG
N/A	OK	OK

The state of the meta data recovery transitions from top to bottom in this table.

Note

Because the "Active(Readable)" state is a state in which only a restore for meta data that are referenced is complete, a data transfer from the cloud provider is always generated when the file is first accessed. This may require more processing time than an access after the restore is complete.
If you are restoring from the "Active(Readable)" state using a backup software that requires writing, you will need to copy and restore the data to a temporary area of the local storage.
In the "Active(Readable)" state, mounting can be performed in the read-write mode, but creating and writing files will cause errors.
In the "Active" or "Active(Readable)" state, the status of the datastore on the GUI is "Normal".

6.4.1.3 Restoring from a System Backup

The procedure for using a system backup to restore the system is described below.

Replace the failed physical disk with one that operates normally.
Restore the system backup.
Refer to the manual of the backup software for details about how to perform a restore.
Note
Only the system area gets restored. For the data area, use an existing disk that functions normally.
Use the following procedure to activate the datastore function.
1. Log in to the console of this product using the administrator account (administrator).
2. Execute the following command to activate the datastore function.
```
# csgadm datastore activate datastoreID
```
  datastoreID, which is specified as the command operand, is the datastore ID. You can check the datastore ID in the Datastore screen of CSG Web GUI (by selecting "ID" in "Display settings").
Information
The reason why the datastore function becomes disabled after restoring a system backup is to prevent corruption of data in the bucket in case a restore is performed accidentally while the backup source system is running. In case the datastore function has become disabled, NAS access is not available.
Execute the following command to restart the system.
```
# csgadm power restart
```

If, when restoring from a system backup, the network MAC address that you assigned to the virtual appliance is not carried over, use the following procedure to reset the IP address.

Note

Because communication with the cloud provider is disabled until the IP address is reset, the system takes approximately ten minutes or longer to restart.

For single network configurations
Run the Configuration Wizard to change the IP address in "Network settings".
Refer to "6.7.1 For a Single Network Configuration" for how to set the IP address. The system restarts at the end of the work procedure, so please restart the system at this time. If you did not choose when asked, restart the system using the following command.
```
# csgadm power restart
```
For multi-network configurations
1. Set the IP address of the first network set in the Configuration Wizard.
  For the configuration method, refer to "6.7.1 For a Single Network Configuration".
  Perform the following procedure from the first network.
2. Delete the second and all subsequent networks.
  The CSG REST API is used to remove the second and all subsequent networks.
  By default, the network that is being used to communicate with the cloud provider cannot be deleted. Delete it by specifying the option that forcibly deletes networks.
3. Re-register all networks from the second one.
  Prevent the network ID that is used for communication with the cloud provider from being changed.
  Use the CSG REST API to check the network ID specified for the cloud provider.
  - If there is a cloud provider that uses the network whose ID is "1", make sure to register a network that can communicate with the relevant cloud provider first.
  - If there is a cloud provider that uses the network whose ID is "2", make sure to register a network that can communicate with the relevant cloud provider next.
  If only networks whose ID is "0" are used to communicate with the cloud provider, networks can be registered in any order.
  Refer to "2.5.7 Configuring Multi-network Configurations" for registering the network.
4. Restart the system.
  Log in to the console of this product using the administrator account (administrator) and execute the following command.
```
# csgadm power restart
```

See

You can also recover the environment with the CSG REST API. Refer to "Meta Data Recovery" and "Shared Folder List (on Recovery)" in the "Reference Guide".