RestoreClusterDatabase function (clusapi.h)

Article
10/13/2021

[This function is available for use in the operating systems specified in the Requirements section. Support for this function was removed in Windows Server 2008 and this function does nothing and returns ERROR_CALL_NOT_IMPLEMENTED.]

Restores the cluster database and restarts the Cluster service on the node from which the function is called. This node is called the restoring node.

Syntax

DWORD RestoreClusterDatabase(
  [in]           LPCWSTR lpszPathName,
  [in]           BOOL    bForce,
  [in, optional] LPCWSTR lpszQuorumDriveLetter
);

Parameters

[in] lpszPathName

Null-terminated Unicode string specifying the path to the backup file. Cluster configuration information is contained in this location; this is sensitive data that should be protected. For example, this data can be protected by using an access control list to restrict access to the location where the data is stored.

[in] bForce

If FALSE, the restore operation will not be completed if either of the following circumstances applies:

Other nodes are currently active.
The partition layout of the current quorum resource is not identical to the partition layout of the quorum resource that was in place when the backup was made. (The term "partition layout" refers to the number of partitions on the disk and the offsets to each partition. The disk signature and drive letter assignments do not have to be identical.)

Setting bForce to TRUE causes the operation to proceed regardless of these preceding circumstances; however, the operation may still fail for other reasons.

[in, optional] lpszQuorumDriveLetter

Optional. Identifies the drive letter of the quorum resource on which the cluster database will be restored. Use this parameter only if the quorum resource has been replaced since the backup was made. The string must be formatted as follows:

The first character must be alphabetic, that is, in the range 'a'-'z' or 'A'-'Z'.
The second character must be a colon (':').
The third character must be a terminating null ('\0').

Return value

If the operation succeeds, the function returns ERROR_SUCCESS.

If the operation fails, the function returns a system error code. The following are possible error codes.

Return code	Description
ERROR_CLUSTER_NODE_UP	The operation failed because other cluster nodes are currently active. If you call RestoreClusterDatabase again with bForce set to TRUE, the cluster will attempt to shut down the Cluster service on the other active nodes.
ERROR_QUORUM_DISK_NOT_FOUND	The operation failed because the quorum disk described in the backup does not match the current quorum disk. If you call RestoreClusterDatabase again with bForce set to TRUE, the cluster will attempt to change the signature and drive letter of the current quorum disk to the values stored in the backup.

Remarks

If the restore operation is successful, the restoring node forms a cluster according to the configuration data in the restored cluster database. As other nodes join the cluster, they update their cluster databases from the database on the restoring node.

Note that cluster disks other than the quorum resource that have added or changed since the backup was made will not be recognized by the restored cluster database and will remain offline even if the restore operation is successful. New resources must be created for these disks (see Creating a Physical Disk Resource).

The following general procedure is recommended for any cluster restore routine:

Call RestoreClusterDatabase with bForce set to FALSE and no drive letter specified. This is the best approach because, if successful, the operation does not have to force configuration changes.

If the first call fails, let the user decide whether to force the procedure to continue or manually fix the problem. Be sure to communicate the implications of each decision.

Return value	Action if forced	Manual fix
ERROR_CLUSTER_NODE_UP	The restore operation will stop the Cluster service on all other nodes.	User manually shuts down the Cluster service on all other cluster nodes. The command Net Stop ClusSvc is sufficient; a complete power-down is unnecessary.
ERROR_QUORUM_DISK_NOT_FOUND	User must supply the drive letter of the quorum resource. The restore operation will change the disk's signature and drive letter to the values stored in the backup.	User repartitions the quorum disk so that the layout is identical to the layout stored in the backup.

If the user agrees to force continuation, call RestoreClusterDatabase with bForce set to TRUE and the drive letter specified (if applicable). Forcing does not guarantee success. If the restore operation fails again, test the return value and respond appropriately.

Examples

The following example illustrates the procedure described above. For a more complete example that includes BackupClusterDatabase, see the Backing Up and Restoring the Cluster Configuration. This example uses the ClusDocEx.h header file defined in the Failover Cluster documentation.


int main( void )
{
    WCHAR szPath[] = L"c:\\ClusBack\\19991215";
    WCHAR szInput[3];
    BOOL bForce = FALSE;
    DWORD dwResult = ERROR_SUCCESS;

    // First try: no force
    dwResult = RestoreClusterDatabase( szPath, FALSE, NULL );
    
    // Allow user to force shutdown if necessary.
    if( dwResult == ERROR_CLUSTER_NODE_UP )
    {
        wprintf( L"The operation failed because other cluster nodes are currently active. " );
        wprintf( L"The Cluster service must be shut down on all other nodes in order for this operation to succeed." );
        wprintf( L"Enter 'f' to force automatic shutdown, or any other key to exit for manual shutdown:  " );
        fgetws( szInput, 2, stdin );
        if( towupper( szInput[0] ) == L'F' )
            dwResult = RestoreClusterDatabase( szPath, TRUE, NULL );
    }

    // Allow user to locate quorum resource if necessary.
    if( dwResult == ERROR_QUORUM_DISK_NOT_FOUND )
    {
        wprintf( L"\n\nERROR: QUORUM DISK NOT FOUND\n" );
        wprintf( L"The restore routine cannot find a quorum resource with the same partition layout as the quorum resource described in the backup. " );
        wprintf( L"The existing quorum resource must have a layout (number of partitions and offsets to each partition) identical to the layout stored in the backup.\n" );
        wprintf( L"Enter the drive letter of the quorum resource to force continuation, or any non-letter key to exit:  " );
        fgetws( szInput, 3, stdin );
        if( iswalpha( szInput[0] ) )
        {
            szInput[1] = L':';
            szInput[2] = L'\0';
            dwResult = RestoreClusterDatabase( szPath, TRUE, szInput );
        }
    }

    // Only one force attempt per error, then report success or failure. 
    if( dwResult == ERROR_SUCCESS )
    {
        wprintf( L"\n\nSUCCESS\n" );
        wprintf( L"The restore routine succeeded. Start the Cluster service on the other cluster nodes to complete the restore operation." );
        wprintf( L"As nodes join the cluster, they will update their cluster databases to match the restored configuration." ); 
        return 0;
    }
    else
    {
        wprintf( L"RestoreClusterDatabase failed (%d)\n", dwResult );
        return 1;
    }

}

Requirements

Requirement	Value
Minimum supported client	None supported
Minimum supported server	Windows Server 2003 Enterprise, Windows Server 2003 Datacenter
Target Platform	Windows
Header	clusapi.h
Library	ClusAPI.lib
DLL	ClusAPI.dll

Share via