S2D: Cache Bindings Error / Rebind
Genutzte Skripte:
Skript von Author: Darryl van der Peijl
Website: http://www.DarrylvanderPeijl.nl/
Email: DarrylvanderPeijl@outlook.com
Date created: 3.january.2018
Last modified: 12.february.2018
Version: 1.1
<#
.DESCRIPTION
Get-CacheDiskStatus is a script that can be used verify the boundings to cache devices
with Storage Spaces Direct
.INPUTS
None
.OUTPUTS
None
.NOTES
Author: Darryl van der Peijl
Website: http://www.DarrylvanderPeijl.nl/
Email: DarrylvanderPeijl@outlook.com
Date created: 3.january.2018
Last modified: 12.february.2018
Version: 1.1
.LINK
http://www.DarrylvanderPeijl.nl/
Tweets by DarrylvdPeijl
#>
Param (
[parameter(Position=0, Mandatory=$true)]
[ValidateNotNullOrEmpty()]
[string]
$ClusterName
)
function Get-PCStorageReportSSBCache ()
{
BEGIN {
$csvf = New-TemporaryFile
function Format-SSBCacheDiskState
([string] $DiskState)
{
$DiskState -replace 'CacheDiskState',''
}
}
<#
These are the possible DiskStates
typedef enum
{
CacheDiskStateUnknown = 0,
CacheDiskStateConfiguring = 1,
CacheDiskStateInitialized = 2,
CacheDiskStateInitializedAndBound = 3, <- expected normal operational
CacheDiskStateDraining = 4, <- expected during RW->RO change (waiting for dirty pages -> 0)
CacheDiskStateDisabling = 5,
CacheDiskStateDisabled = 6, <- expected post-disable of S2D
CacheDiskStateMissing = 7,
CacheDiskStateOrphanedWaiting = 8,
CacheDiskStateOrphanedRecovering = 9,
CacheDiskStateFailedMediaError = 10,
CacheDiskStateFailedProvisioning = 11,
CacheDiskStateReset = 12,
CacheDiskStateRepairing = 13,
CacheDiskStateIneligibleDataPartition = 2000,
CacheDiskStateIneligibleNotGPT = 2001,
CacheDiskStateIneligibleNotEnoughSpace = 2002,
CacheDiskStateIneligibleUnsupportedSystem = 2003,
CacheDiskStateIneligibleExcludedFromS2D = 2004,
CacheDiskStateIneligibleForS2D = 2999,
CacheDiskStateSkippedBindingNoFlash = 3000,
CacheDiskStateIgnored = 3001,
CacheDiskStateNonHybrid = 3002,
CacheDiskStateInternalErrorConfiguring = 9000,
CacheDiskStateMarkedBad = 9001,
CacheDiskStateMarkedMissing = 9002,
CacheDiskStateInStorageMaintenance = 9003 <- expected during FRU/maint
}
CacheDiskState;
#>
PROCESS {
$log = get-clusterlog -Node $env:computername -Destination C:\Clusterlog
dir C:\clusterlog\*cluster.log | sort -Property BaseName |% {
$node = "<unknown>"
if ($_.BaseName -match "^(.*)_cluster$") {
$node = $matches[1]
}
Write-Output ("-"*40) "Node: $node"
##
# Parse cluster log for the SBL Disk section
##
$sr = [System.IO.StreamReader]$_.FullName
$in = $false
$parse = $false
$(do {
$l = $sr.ReadLine()
# Heuristic ...
# SBL Disks comes before System
if ($in) {
# in section, blank line terminates
if ($l -notmatch '^\s*$') {
$l
} else {
# parse was good
$parse = $true
break
}
} elseif ($l -match '^\[=== SBL Disks') {
$in = $true
} elseif ($l -match '^\[=== System') {
break
}
} while (-not $sr.EndOfStream)) > $csvf
##
# With a good parse, provide commentary
##
if ($parse) {
$d = import-csv $csvf
##
# Table of raw data, friendly cache device numbering
##
$idmap = @{}
$d |% {
$idmap[$_.DiskId] = $_.DeviceNumber
}
$d | sort IsSblCacheDevice,CacheDeviceId,DiskState | ft -AutoSize @{ Label = 'DiskState'; Expression = { Format-SSBCacheDiskState $_.DiskState }},
DiskId,DeviceNumber,@{
Label = 'CacheDeviceNumber'; Expression = {
if ($_.IsSblCacheDevice -eq 'true') {
'= cache'
} elseif ($idmap.ContainsKey($_.CacheDeviceId)) {
$idmap[$_.CacheDeviceId]
} elseif ($_.CacheDeviceId -eq '{00000000-0000-0000-0000-000000000000}') {
"= unbound"
} else {
# should be DiskStateMissing or OrphanedWaiting? Check live.
"= not present $($_.CacheDeviceId)"
}
}
},HasSeekPenalty,PathId,BindingAttributes,DirtyPages
##
# Now do basic testing of device counts
##
$dcache = $d |? IsSblCacheDevice -eq 'true'
$dcap = $d |? IsSblCacheDevice -ne 'true'
Write-Output "Device counts: cache $($dcache.count) capacity $($dcap.count)"
##
# Test cache bindings if we do have cache present
##
if ($dcache) {
# first uneven check, the basic count case
$uneven = $false
if ($dcap.count % $dcache.count) {
$uneven = $true
Write-Warning "Capacity device count does not evenly distribute to cache devices"
}
# now look for unbound devices
$unbound = $dcap |? CacheDeviceId -eq '{00000000-0000-0000-0000-000000000000}'
if ($unbound) {
Write-Warning "There are $(@($unbound).count) unbound capacity device(s)"
}
# unbound devices give us the second uneven case
if (-not $uneven -and ($dcap.count - @($unbound).count) % $dcache.count) {
$uneven = $true
}
$gdev = $dcap |? DiskState -eq 'CacheDiskStateInitializedAndBound' | group -property CacheDeviceId
if (@($gdev).count -ne $dcache.count) {
Write-Warning "Not all cache devices in use"
}
$gdist = $gdev |% { $_.count } | group
# in any given round robin binding of devices, there should be at most two counts; n and n-1
# single ratio
if (@($gdist).count -eq 1) {
Write-Output "Binding ratio is even: 1:$($gdist.name)"
} else {
# group names are n in the 1:n binding ratios
$delta = [math]::Abs([int]$gdist[0].name - [int]$gdist[1].name)
if ($delta -eq 1 -and $uneven) {
Write-Output "Binding ratios are as expected for uneven device ratios"
} else {
Write-Warning "Binding ratios are uneven"
}
# form list of group sizes
$s = $($gdist |% {
"1:$($_.name) ($($_.count) total)"
}) -join ", "
Write-Output "Groups: $s"
}
}
##
# Provide summary of diskstate if more than one is present in the results
##
$g = $d | group -property DiskState
if (@($g).count -ne 1) {
write-output "Disk State Summary:"
$g | sort -property Name | ft @{ Label = 'DiskState'; Expression = { Format-SSBCacheDiskState $_.Name}},@{ Label = "Number of Disks"; Expression = { $_.Count }}
} else {
$gname = (($g.name) -replace 'CacheDiskState','')
write-output "All disks are in $gname"
}
}
}
}
END {
del $csvf
}
}
Function Get-CacheDiskStatus
{
Param (
[parameter(Position=0, Mandatory=$true)]
[ValidateNotNullOrEmpty()]
[string]
$ClusterName
)
Get-clusternode -Cluster $ClusterName | % {
Invoke-Command -ComputerName $_.NodeName -ScriptBlock ${Function:Get-PCStorageReportSSBCache}
}
}
Get-CacheDiskStatus -ClusterName $Clustername
Eigenes Skript mit Anleitung vom MS Support für das Rebinden von Disks zu Cache Devices:
#Erste Checks #Health Service muss Online sein Get-ClusterResource *Health* #Alles sollte Healthy sein Get-VirtualDisk Get-Volume Get-StoragePool Get-Disk #Alles sollte Healthy sein Get-PhysicalDisk | Sort-Object DeviceID #Keine laufenden Jobs Get-StorageJob #Disk auswählen die im Cache-Binding Skript angemeckert wurde: #PhysicalDisk anhand ObjectID filtern Hinter PD Get-PhysicalDisk -ObjectId *dc6f5de* #Variabel für pDisk erstellen $Disk = Get-PhysicalDisk -ObjectId *dc6f5de* #pDisk in Pause setzen Set-PhysicalDisk -InputObject $Disk -Usage Retired #Check Retire Status Get-PhysicalDisk | Sort-Object Usage #Check Footprint sollte jetzt bei 0 sein! Get-PhysicalDisk -ObjectId *dc6f5de* | fl uni*,VirtualDiskFootprint #Storage Job abwarten Get-StorageJob #Check pDisk Get-PhysicalDisk | Sort-Object Usage #Remove pDisk from Pool $PDToRemove = Get-PhysicalDisk -ObjectId *dc6f5de* Get-StoragePool Remove-PhysicalDisk -PhysicalDisks $PDToRemove -StoragePoolFriendlyName "S2D on Cluster-4" #Unclaiming Disk #Info zu Unterschied zwischen Server 2016 und Server 2016 #Version für 2016 benötigt die UniquieID der Disk Get-PhysicalDisk $Disk | FL UniquieId $DiskUQID = Get-PhysicalDisk -UniqueId XXXXXXXXXXXXXX #Check Variable $DiskUQID.UniqueId Set-ClusterStorageSpacesDirectDisk -CanBeClaimed $false -PhysicalDisk $DiskUQID.UniqueId #Version für 2019 funktioniert mit der pDisk $Disk oben schon deklariert. Set-ClusterStorageSpacesDirectDisk -CanBeClaimed 0 -PhysicalDisk $Disk #Check pDisk Get-PhysicalDisk | Sort-Object Usage #Cleanen des Disks Get-Disk #Korrekte Nummer Auswählen und Ausführung des Clearens $disknumber = "7" Get-Disk | ? Number -e $disknumber | Set-Disk -isoffline:$false Get-Disk | ? Number -e $disknumber | Set-Disk -isreadonly:$false Get-Disk | ? Number -e $disknumber | Clear-Disk -RemoveData -RemoveOEM -Confirm:$false Get-Disk | ? Number -e $disknumber | Set-Disk -isreadonly:$true Get-Disk | ? Number -e $disknumber | Set-Disk -isoffline:$true #Reclaiming the Disk Get-PhysicalDisk | Sort-Object CanPool #Health Service muss Online sein Get-ClusterResource *Health* (get-clusterresource health | Get-ClusterParameter).Value #Kann jetzt wieder verbunden werden Set-ClusterStorageSpacesDirectDisk -CanBeClaimed $true -PhysicalDisk $DiskUQID.UniqueId #pDisk sollte sich eigentlich automatisch wieder in das Pool integrieren sonst manuell: #$PDToAdd = Get-PhysicalDisk -ObjectId *dc6f5de9-a3fa-bc32* #Add-PhysicalDisk -PhysicalDisks $PDToAdd -StoragePoolFriendlyName "S2D on Cluster-4" #Check Disks Get-PhysicalDisk | Sort-Object SerialNumber Get-PhysicalDisk | Sort-Object Usage #Rebalance Volume auf Disks Get-StorageJob Get-StoragePool -FriendlyName "S2D *" | Optimize-StoragePool
Fehler erkannt im Cache Bindings Skript
Disk Id = dc6fxxxxxxxxxxxxxx
Ausführen des Skriptes Rebind S2D Disks mit der entsprechenden Disk
Ergebnis:
Disk Id = dc6fxxxxxxxxxxxxxx wieder einem Cache Device zugeordnet
Info von MS:
Binding der Disks kann man auch im PerfMon überprüfen.
Node für Node unter:
Links die Disk Ids der Capacity Drives
Rechts die Disk Ids der Cache Drives
Schreibe einen Kommentar