Udostępnij za pośrednictwem

Step 2: Create the Scale-Out File Server and storage pools

  • Artykuł

 

Applies To: Windows Server 2012 R2

This topic discusses how to set up a clustered file server and create storage pools. It includes how to prepare storage for clustering, how to create the cluster and storage pools, as well as how to set storage pool defaults and create a clustered file server role (scale-out file server).

Before using this topic, you should already have installed the necessary server roles, configured MPIO, and done the other steps in Step 1: Set up the nodes of the file server.

Step 1: Prepare Storage Spaces for clustering and pooling

Before creating the cluster, it's a good idea to wipe out any existing Storage Spaces and Failover Clustering data, check that all physical disks are healthy, online and writeable, and initialize any RAW disks. Preparing the storage before clustering helps reduce cluster validation warnings.

To prepare storage by using Server Manager

  1. If your storage enclosures and servers were previously used for something else, completely erase all Storage Spaces and Failover Clustering data from the physical disks and storage enclosures.

    For a script that can completely erase everything from a Storage Spaces configuration (and we do mean everything, so be careful!), see Completely Clearing an Existing Storage Spaces Configuration. This script (Clear-SdsConfig.ps1) can take around 30 minutes to wipe 180 disks.

  2. Use Remote Desktop to connect to each cluster node, and then go to Server Manager > File and Storage Services > Storage Pools. Select the Primordial pool for the cluster node and confirm that the Physical Disks tile lists all disks in the shared storage enclosures. Also confirm that:

    • None of the disks have a checkmark in the Read Only column

    • None of the disks display any warnings in the Health (!) column

    • All of the disks show the correct Slot number, MediaType, and Chassis.

  3. Validate the health and performance of your physical disks and enclosures. Doing so helps ensure that there aren't low-performing disks that could reduce overall storage performance.

    To validate your storage, use the Storage Spaces Physical Disk Validation Script. Note that this script (ValidateStorageHardware.ps1) preconditions your SSDs and does performance testing, so it'll take quite a while to run (up to a day or two).

  4. Replace any disks that fail the validation script above, if possible.

    Doing so will increase storage performance - the performance of a storage tier in Storage Spaces is limited by the performance of the slowest disk in the tier.

  5. We recommend using the Storage Spaces Automated Provisioning Script (ProvisionStorageConfiguration.ps1) to create clustered storage pools and virtual disks according to best practices. Enter the following info according to the design you created earlier, and run the Clear-SdsConfig.ps1 script again if any of the disks are missing after running the ValidateStorageHardware.ps1 script :

    • Number of cluster nodes

    • Amount of capacity to reserve for storing backups (which you'll provision separately)

    • The physical disk redundancy for the workload pool (for example, specify 2 to use three-way mirrors).

  6. If you don't use the Storage Spaces Automated Provisioning Script (ProvisionStorageConfiguration.ps1), instead Initialize any disks with the RAW partition style so that Storage Spaces can use them.

    To do so, in Computer Management, go to Disk Management. The Initialize Disk dialog box appears if you have any uninitialized disks - choose the GPT partition style.

To prepare storage by using Windows PowerShell

  1. If your storage enclosures and servers were previously used for something else, completely erase all Storage Spaces and Failover Clustering data from the physical disks and storage enclosures.

    For a script that can completely erase everything from a Storage Spaces configuration (and we do mean everything, so be careful!), see Completely Clearing an Existing Storage Spaces Configuration. This script can take around 30 minutes to wipe 180 disks.

  2. Validate the health and performance of your physical disks and enclosures. Doing so helps ensure that there aren't low-performing disks that could reduce overall storage performance.

    To validate your storage, use the Storage Spaces Physical Disk Validation Script and the procedure in Storage Spaces Deployment Guide for Automation Scripts. Note that this script preconditions your SSDs and does performance testing, so it'll take quite a while to run (up to a day or two).

  3. Replace any disks that fail the validation script above, if possible.

    Doing so will increase storage performance - the performance of a storage tier in Storage Spaces is limited by the performance of the slowest disk in the tier.

  4. Use the Storage Spaces Automated Provisioning Script (ProvisionStorageConfiguration.ps1) to create clustered storage pools and virtual disks according to best practices. Enter the following info according to the design you created earlier:

    • Number of cluster nodes

    • Amount of capacity to reserve for storing backups (which you'll provision separately)

    • The physical disk redundancy for the workload pool (for example, specify 2 to use three-way mirrors).

Step 2: Create the cluster

Once your storage is prepped, you can cluster the file server nodes.

To create the cluster by using Server Manager

To create the file server cluster, use the following procedure from one of the cluster nodes or a management computer in the same domain as the cluster:

  1. In Server Manager click Tools > Failover Cluster Manager.

  2. In the Actions pane, click Validate Configuration….

  3. Use the Validate a Configuration Wizard to specify the cluster nodes and then run all tests. This could take an hour or more, so get comfortable.

  4. Fix any problems identified by the wizard.

    If there aren't any problems, leave the Create the cluster now using the validated nodes… checkbox selected and then click Finish to create the cluster.

    If you're having trouble getting disks to appear, check the List Disks section of the report for more details.

  5. If you didn't already launch the Create Cluster Wizard, in the Actions pane of Failover Cluster Manager, click Create Cluster….

  6. On the Access Point for Administering the Cluster page shown in Figure 1, enter the name you want to use when administering the cluster itself - you'll create a clustered file server instance that other computers will connect to later (with a different name).

    Figure 1 The Access Point for Administering the Cluster page of the Create Cluster Wizard

  7. On the same page, enter static IP addresses for each network interface.

  8. On the Confirmation page, unless you used the ProvisionStorageConfiguration.ps1 PowerShell script to automatically create your pools and virtual disks, clear the Add all eligible storage to the cluster checkbox - you'll pool the disks later.

    Note

    You might see "Name resolution not yet available" in Failover Cluster Manager for the first 15 minutes or so after creating the cluster due to DNS propagation delays. Server Manager also might report problems. The easiest thing to do is wait for the DNS data to propagate.

To create the cluster by using Windows PowerShell

From a management computer on one of the subnets connected to the nodes, open a Windows PowerShell session and then use the following steps.

  1. Run cluster validation and resolve any outstanding issues.

    To do so, enter the following commands from a management computer that has the Failover Clustering cmdlets and is in the same domain as the cluster, replacing FSNode01,FSNode02 with the names of your cluster nodes. This could take a few minutes or longer and requires Microsoft Edge or Internet Explorer on the management computer to view the validation report.

    # Specify which computers you want to run the script on
[string[]]$ComputerName = "FSNode01","FSNode02"

if (Test-Connection $ComputerName -Count 1) {
    $report = Test-Cluster -Node $ComputerName
    Invoke-Item $report.versioninfo.filename
} #/endif

  2. To create the failover cluster, enter the following commands from a management computer in the same domain as the cluster, again replacing FS01,FS02 with the names of your cluster nodes and specifying appropriate IP addresses. If you didn't use the ProvisionStorageConfiguration.ps1 script to create your storage pools, add the -NoStorage parameter.

    New-Cluster -Name FileServerCluster01 -Node FSNode01,FSNode02 -StaticAddress 10.10.1.60,10.10.2.60

Step 3: Create clustered storage pools

Create the storage pools specified by your storage design. If you used the ProvisionStorageConfiguration.ps1 script to provision your storage, skip this step.

You can use Failover Cluster Manager to create a simple storage pool, or Windows PowerShell if you want to automate creating multiple pools to create or need to change advanced pool settings.

To create storage pools by using Failover Cluster Manager

  1. In Failover Cluster Manager, go to <ClusterName> > Storage > Pools.

  2. Right-click Pools, and then click New Storage Pool.

    The New Storage Pool wizard appears.

  3. On the Specify a storage pool name and subsystem page, enter a name for the storage pool, and then select the group of available physical disks (also known as a primordial pool) that you want to use. In the Available to column, verify that the cluster name is listed.

  4. On the Select physical disks for the storage pool page, select each physical disk that you want to include in the storage pool.

    For a failover cluster, you must select at least three physical disks. If you're using enclosure awareness and multiple pools, make sure that you evenly allocate disks from all enclosures to each pool.

  5. Select any SSDs that you want to use as journal disks for use by parity spaces in the pool (see Software-Defined Storage Design Considerations Guide for more details), and then in the Allocation column, choose Journal from the drop-down list.

  6. Repeat these steps for any additional pools you're creating.

To create storage pools by using Windows PowerShell

If you didn't use the ProvisionStorageConfiguration.ps1 script to provision your storage, from a management computer on one of the subnets connected to the nodes, open a Windows PowerShell session and then use the following commands to create a single storage pool with all physical disks, and 4 KB logical sectors.

$DisksToAdd = Get-PhysicalDisk -CanPool $true
New-StoragePool -PhysicalDisks $DisksToAdd -FriendlyName StoragePool1 -StorageSubSystemFriendlyName "Clustered Storage Spaces*" -LogicalSectorSizeDefault 4096

You can also use the ProvisionStorageConfiguration.ps1 script to automatically create storage pools and virtual disks according to recommended best practices. For more information, see Storage Spaces Deployment Guide for Automation Scripts .

Step 4: Set storage pool defaults

After creating your storage pools there are a few settings that you should use Windows PowerShell to change (if you used the ProvisionStorageConfiguration.ps1 script to provision your storage, skip this step):

  • When to retire missing physical disks

  • Enclosure awareness

  • Default number of columns

The first one you should change if you want Storage Spaces to automatically repair virtual disks after a disk failure (we recommend this for all but the smallest environments); the other two are necessary only if you use a GUI to create virtual disks.

Consult your storage design, open a Windows PowerShell session with administrative permissions on the file server cluster, and then use this chunk of code to change these settings, modifying the first three lines to suit your design.

$PoolName = "Pool1"
$EnclosureAwareDefault = $true
$MirrorNumberofColumnsDefault = 4

Write-Output "Current storage pool and resiliency type settings"
Get-StoragePool -FriendlyName $PoolName |
Format-Table FriendlyName, EnclosureAwareDefault, RetireMissingPhysicalDisks, RepairPolicy -AutoSize
Get-ResiliencySetting -StoragePool (Get-StoragePool $PoolName) -Name Mirror |
Format-Table -Property Name,NumberOfColumnsDefault,NumberOfDataCopiesDefault, InterleaveDefault -AutoSize

Set-StoragePool -FriendlyName $PoolName -RetireMissingPhysicalDisks Always
Set-StoragePool -FriendlyName $PoolName -EnclosureAwareDefault $EnclosureAwareDefault
Set-ResiliencySetting -StoragePool (Get-StoragePool $PoolName) -Name Mirror -NumberOfColumnsDefault $MirrorNumberofColumnsDefault

Write-Output "New storage pool and resiliency type settings"
Get-StoragePool -FriendlyName $PoolName |
Format-Table FriendlyName, EnclosureAwareDefault, RetireMissingPhysicalDisks, RepairPolicy -AutoSize
Get-ResiliencySetting -StoragePool (Get-StoragePool $PoolName) -Name Mirror |
Format-Table -Property Name,NumberOfColumnsDefault,NumberOfDataCopiesDefault, InterleaveDefault -AutoSize

Step 5: Set the cluster quorum witness

This solution uses a witness disk to provide quorum to the file server cluster. The following procedures describe how to create an appropriate storage space (more generically called a virtual disk) and how to assign it as the witness disk, using Failover Cluster Manager or Windows PowerShell.

Later you can use this virtual disk as a file share witness for other clusters in your solution if you want.

To set up the cluster quorum witness by using Failover Cluster Manager

Connect to the cluster using an account with administrative permissions and then use the following procedure to create a new virtual disk for use as the disk witness and set the cluster quorum to use the disk witness.

Create a virtual disk for use as a disk witness

  1. In Failover Cluster Manager, select the cluster, and then go to Storage > Pools.

  2. Select a pool, and then click New Virtual Disk.

  3. On the Storage Pool page, select the first pool (if you have multiple pools).

  4. On the Virtual Disk Name page, enter a name for the virtual disk (for example, File Server Cluster Quorum).

    Don't create storage tiers on this virtual disk - the cluster quorum doesn't need the added performance of an SSD tier. Because your SSDs are likely set to Automatic allocation, some of the virtual disk will get written to SSD anyway, but not enough to worry about.

  5. On the Storage Layout page, click Mirror.

  6. On the Resiliency Settings page, click Two-way mirror.

  7. On the Specify the size of the virtual disk page, enter 4 (GB).

  8. On the Assign to a drive letter or folder page of the New Volume Wizard, choose Don't assign to a drive letter or folder (you'll probably want to save the drive letter for a more commonly used volume).

To set up the cluster quorum witness by using Windows PowerShell

In a Windows PowerShell session with administrative credentials that's connected to the file server cluster, enter the following commands to create a new virtual disk for use as the disk witness and then set the cluster quorum to use the disk witness.

Create the virtual disk and assign it as the witness disk

  1. If you didn't use the ProvisionStorageConfiguration.ps1 script to create your virtual disks (which creates a quorum disk for you), run these commands to create the quorum virtual disk:

    $WitnessDiskName = "WitnessDisk1"

New-Volume -StoragePoolFriendlyName "Pool1" -FriendlyName $WitnessDiskName -Size 4GB -ResiliencySettingName Mirror -NumberOfColumns 3 -FileSystem NTFS -PhysicalDiskRedundancy 2

  2. Then use the following commands to add CSVs to all of the virtual disks the ProvisionStorageConfiguration.ps1 script created, remove the CSV from the virtual disk to be used as a disk witness, and make sure that the disk witness is online, replacing fsnode01 with the computer name of the node on which you'll run the script:

    # Specify the node from which you're running scripts so we can make sure resources are on the node you're administering
$currentnode = "fsnode01"

# This gets the name of the virtual disk used for cluster quorum - probably created by ProvisionStorage.ps1
$ClusterQuorumFileSystemLabel = (Get-VirtualDisk *quorum*).FriendlyName

# Bring virtual disks online and make them CSVs
Get-ClusterResource *disk* | Add-ClusterSharedVolume | Start-ClusterResource

# Find the CSV that is supposed to be the disk quorum (it's easy to find once it's a CSV)
$QuorumCSV = Get-ClusterSharedVolume -Name (Get-ClusterSharedVolumeState |
Where-Object -Property VolumeName -EQ (Get-Volume -FileSystemLabel "Cluster Quorum Disk").Path).Name

# Remove the CSV that's supposed to be the disk queue
Remove-ClusterSharedVolume -Name $QuorumCSV[0].Name

#Make sure that the virtual disk is on the current node
Get-ClusterGroup -Name "Available Storage" | Move-ClusterGroup -Node $currentnode
Get-ClusterGroup -Name "Available Storage" | Start-ClusterGroup

  3. Then use the following command to set the cluster quorum:

    $WitnessDiskName = Get-ClusterResource *disk*
Set-ClusterQuorum -DiskWitness $WitnessDiskName

    In the output from these commands, the virtual disk should be listed with the the QuorumType as NodeAndDiskMajority, as shown in this example output:

    Cluster        : FSCLUSTER
QuorumResource : Cluster Virtual Disk (WitnessDisk1)
QuorumType     : NodeAndDiskMajority

  4. Then run the following commands to confirm that the storage pools and virtual disk are listed as Online and that the correct quorum resource is listed.

    Get-ClusterResource |
Where-Object {$_.ResourceType -eq "Storage Pool" -or $_.ResourceType -eq "Physical Disk"} |
Format-Table -Autosize
Get-ClusterQuorum

    Here's an example of the output:

    Name                                    State  OwnerGroup        ResourceType
----                                    -----  ----------        ------------
Cluster Pool 1                          Online Cluster Group     Storage Pool
Cluster Disk 18     Online Cluster Group Physical Disk

Cluster              QuorumResource
-------              --------------
FSCLUSTER            Cluster Disk 18

Step 6: Create the Scale-Out File Server role

The last step in setting up the cluster services for your file server is creating the clustered file server role, which is when you create the Scale-Out File Server instance on which your continuously available file shares are hosted.

To create a Scale-Out File Server role by using Server Manager

  1. In Failover Cluster Manager, select the cluster, go to Roles, and then click Configure Role….

    The High Availability Wizard appears.

  2. On the Select Role page, click File Server.

  3. On the File Server Type page, click Scale-Out File Server for application data.

  4. On the Client Access Point page, type a name for the Scale-Out File Server.

  5. Verify that the role was successfully set up by going to Roles and confirming that the Status column shows Running next to the clustered file server role you created, as shown in Figure 2.

    Figure 2 Failover Cluster Manager showing the Scale-Out File Server with the Running status

Note

After creating the clustered role, there might be some network propagation delays that could prevent you from creating file shares on it for a few minutes, or potentially longer.

To create a Scale-Out File Server role by using Windows PowerShell

In a Windows PowerShell session that's connected to the file server cluster, enter the following commands to create the Scale-Out File Server role, changing FSCLUSTER to match the name of your cluster, and SOFS to match the name you want to give the Scale-Out File Server role:

Add-ClusterScaleOutFileServerRole -Name SOFS -Cluster FSCLUSTER

Note

After creating the clustered role, there might be some network propagation delays that could prevent you from creating file shares on it for a few minutes, or potentially longer. If the SOFS role fails immediately and won't start, it might be because the cluster's computer object doesn't have permission to create a computer account for the SOFS role. For help with that, see this blog post: Scale-Out File Server Role Fails To Start With Event IDs 1205, 1069, and 1194.

Next steps

After completing the steps in this topic, you should have a clustered file server and storage pool(s), and will be ready to create virtual disks and shares, as discussed in Step 3: Create storage spaces and file shares on the Scale-Out File Server.