4

Environment: Windows 2016 FCI cluster with S2D running SQL Server cluster on 2 VMs in Azure.

New vhd disk has been added to a single VM in the cluster with intention to use it as local. Because 'AutoPool' is enabled, it grabbed the disk and marked its Usage property as 'Auto-Select'.

Question: how do I change that single disk to Manual-Select so that I could create local volume from it? Running: Set-PhysicalDisk -UniqueId mydiskId -Usage ManualSelect errors out with non-descriptive message.

Here's the output of physical disks with the one I'd like to remove highlighted in yellow, storage pools and storage settings:
S2D

Just came across this link https://docs.microsoft.com/en-us/powershell/module/storage/set-physicaldisk?view=winserver2012r2-ps that states "When using the Storage Spaces subsystem, Set-PhysicalDisk only works on physical disks that have been added to a storage pool"

This is not my case as the disk I'd like to remove is not part of any pool.
Is there any other command to remove my disk from auto pooling without actually disabling AutoPool setting, as per @CosmosDarvin answer in Is there any way to prevent Storage Spaces Direct from automatically adding disks? ?

Also, what are the repercussions if I set AutoPool to false? This is live cluster. BTW, I'm not planning on adding more vhd disks to pools frequently (maybe once a year) so disabling AutoPooling would be ok with me but I'd like to understand if there are any other effect this change may cause.

P. D.
  • 71
  • 4

1 Answers1

5

You can easily do what you want, check this link out:

https://docs.microsoft.com/en-us/powershell/module/storage/get-storagesubsystem

Get-StorageSubSystem Cluster* | Set-StorageHealthSetting -Name "System.Storage.PhysicalDisk.AutoPool.Enabled" -Value False

As a side effect: I'd check AlwaysOn Availability Groups (AGs) for SQL Server, rather than messing with FCI + super-slow S2D in Azure.

BaronSamedi1958
  • 12,510
  • 1
  • 20
  • 46
  • 1
    There was a hard requirement to use FCI and not AlwaysOn. What do you base your opinion on stating "super-slow S2D in Azure"? Thank you – P. D. Jan 26 '18 at 20:07
  • 1
    1) You're confused: FCI is still part of AlwaysOn. There are AlwaysOn FCI and AlwaysOn AGs. See: https://docs.microsoft.com/en-us/sql/sql-server/failover-clusters/windows/always-on-failover-cluster-instances-sql-server and https://docs.microsoft.com/en-us/sql/database-engine/availability-groups/windows/failover-clustering-and-always-on-availability-groups-sql-server – BaronSamedi1958 Jan 27 '18 at 16:17
  • 1
    2) It's not a biased opinion, it's empirical entirely. Get some write-intensive DB and do FCI vs AGs rat race in Azure to see the difference. Repeat on-premises. – BaronSamedi1958 Jan 27 '18 at 16:19
  • I have AutoPool.Enabled set to 'false' now. How do I go about reclaiming that disk that was auto-selected by S2D? It is still showing unchanged with Usage:Auto-Select and CanPool:True. I really need to reclaim that disk and make it show up in DIsk Manager so that I could use it as a local disk outside S2D. – P. D. Feb 02 '18 at 00:38
  • I just tested by adding new 64GB disk to one of the nodes and S2D picked it up and its Usage property to 'Auto-Select' :-( When querying AutoPool setting I get 'false': `Get-StorageSubSystem Cluster* | Get-StorageHealthSetting -Name "System.Storage.PhysicalDisk.AutoPool.Enabled" Name Value PSComputerName ---- ----- -------------- System.Storage.PhysicalDisk.AutoPool.Enabled False` This is beyond frustrating at this point.... – P. D. Feb 02 '18 at 01:59
  • We don't use S2D anymore so... I feel your pain! – BaronSamedi1958 Feb 02 '18 at 12:32
  • 4
    I can agree with @BaronSamedi1958 A 2 node S2D cluster for SQL is a very daredevil move, just due to how fragile it is in terms of resiliency.You can try to use for example a 2 node "classic" Failover Cluster with some sort of shared storage, like free Starwind / HPE VSA configured with a high amount of RAM cache, and create a simple HA VM with SQL Standard on top of it. – Strepsils Feb 07 '18 at 16:59
  • @Strepsils Could you elaborate on how Storage Spaces Direct is fragile? We are attempting to migrate a 2 node FCI to Azure and looking for the best solution. S2D and a FCI is the default response I am seeing from googling around. – Justin Aug 01 '18 at 16:03