#Requires -Version 5.1
<#
    JbTecWiz Support Centre -- generated fix script

    Fault : Random crashes and corruption -- testing for bad memory
    Fix   : Read what ECC has already recorded
    Source: https://jbtecwiz.com/support/win-hw-memory

    Run as     : Elevated PowerShell and the server's management controller
    Expect     : 30 minutes
    Risk       : low
    Reversible : yes

    WHEN THIS IS THE RIGHT FIX
    A server. ECC has almost certainly logged the failing DIMM by name.

    HOW TO UNDO IT
    Not applicable.

    This script walks the fix one step at a time and asks before each
    one. Steps with no command are things you do yourself -- it prints
    those and waits. Run with -DryRun to print without executing.

    --------------------------------------------------------------------
    NO WARRANTY - USE AT YOUR OWN RISK

    This script is provided by JbTecWiz as-is and with no warranty of any
    kind, express or implied. You run it entirely at your own risk.

    JbTecWiz accepts no liability for any loss or damage arising from its
    use, including but not limited to data loss, downtime, or configuration
    changes that turn out to be wrong for your system.

    You are responsible for reading this script before running it, for
    satisfying yourself that it suits the machine in front of you, and for
    having a working backup first. Some steps cannot be undone.
    --------------------------------------------------------------------
#>

[CmdletBinding()]
param(
    # Print every step and command without running anything.
    [switch]$DryRun,
    # Do not ask before each step. Read the script first if you use this.
    [switch]$Unattended
)

$ErrorActionPreference = 'Stop'
$script:Failed = 0

function Write-Rule { param([string]$Text)
    Write-Host ''
    Write-Host ('-' * 70) -ForegroundColor DarkGray
    if ($Text) { Write-Host $Text -ForegroundColor Cyan }
}

function Show-Prose { param([string]$Text, [string]$Colour = "Gray")
    if (-not $Text) { return }
    $words = $Text -split "\s+"; $line = "  "
    foreach ($w in $words) {
        if (($line.Length + $w.Length) -gt 74) { Write-Host $line -ForegroundColor $Colour; $line = "  " }
        $line += "$w "
    }
    if ($line.Trim()) { Write-Host $line -ForegroundColor $Colour }
}

function Invoke-Step {
    param(
        [int]$Number, [string]$Do, [string]$Why,
        [scriptblock]$Command, [switch]$Manual, [string]$Shell = "powershell"
    )
    Write-Rule "  Step $Number of 4"
    Show-Prose $Do "White"
    if ($Why) { Write-Host ""; Show-Prose $Why "DarkGray" }

    if ($Manual) {
        Write-Host ''
        Write-Host '  -> Do this yourself, then press Enter to carry on.' -ForegroundColor Yellow
        if (-not $Unattended -and -not $DryRun) { [void](Read-Host) }
        return
    }

    Write-Host ''
    foreach ($l in ($Command.ToString().Trim() -split "`n")) {
        Write-Host ("    " + $l.Trim()) -ForegroundColor Green
    }
    Write-Host ''

    if ($DryRun) { Write-Host "  (dry run -- not executed)" -ForegroundColor DarkGray; return }
    if (-not $Unattended) {
        $a = Read-Host "  Run this step? [Y]es / [S]kip / [Q]uit"
        if ($a -match "^[Qq]") { Write-Host "  Stopped at your request."; exit 0 }
        if ($a -match "^[Ss]") { Write-Host "  Skipped." -ForegroundColor DarkGray; return }
    }
    try { & $Command }
    catch {
        $script:Failed++
        Write-Host ("  Step $Number failed: " + $_.Exception.Message) -ForegroundColor Red
        Show-Prose "The rest of the fix may depend on this. Read the write-up before carrying on." "Red"
        if (-not $Unattended) {
            $c = Read-Host "  Carry on anyway? [y/N]"
            if ($c -notmatch "^[Yy]") { exit 1 }
        }
    }
}

Write-Rule
Write-Host '  Random crashes and corruption -- testing for bad memory' -ForegroundColor White
Write-Host '  Read what ECC has already recorded' -ForegroundColor Cyan
Write-Host ''
Write-Host '  Risk: low   Reversible   30 minutes'
Write-Host ''
Show-Prose 'No warranty. Use at your own risk - JbTecWiz accepts no liability. Read it before you run it, and have a backup.' 'DarkYellow'
Write-Rule

if (-not $Unattended -and -not $DryRun) {
    $go = Read-Host '  Ready? [y/N]'
    if ($go -notmatch "^[Yy]") { Write-Host "  Nothing was changed."; exit 0 }
}

Invoke-Step -Number 1 -Do 'Read the WHEA records -- these name the physical DIMM slot.' -Command {
    Get-WinEvent -LogName System -MaxEvents 500 | Where-Object ProviderName -like '*WHEA*' | Format-List TimeCreated,Id,Message
}

Invoke-Step -Number 2 -Do 'Check the management controller -- iDRAC, iLO, IMM or equivalent -- for correctable error counts per DIMM. A rising correctable count precedes an uncorrectable failure by days or weeks.' -Why 'This is the value of ECC: the failure is visible before it takes the machine down. A DIMM with a climbing correctable count should be replaced at the next window, not after the outage.' -Manual

Invoke-Step -Number 3 -Do 'Replace the identified DIMM and clear the counters in the management controller.' -Manual

Invoke-Step -Number 4 -Do 'If several DIMMs on the same channel report errors, suspect the CPU socket or a bent pin rather than the memory.' -Manual

Write-Rule "  Confirm it worked"
Show-Prose 'Correctable error counts stay at zero for a week after replacement.' 'White'
Write-Host ''
if (-not $DryRun) {
    Get-WinEvent -LogName System -MaxEvents 200 | Where-Object ProviderName -like '*WHEA*' | Measure-Object | Format-List Count
}

Write-Rule
if ($script:Failed -gt 0) {
    Write-Host ("  Finished with " + $script:Failed + " failed step(s).") -ForegroundColor Yellow
    Show-Prose 'Read the full write-up at https://jbtecwiz.com/support/win-hw-memory' 'Yellow'
} else {
    Write-Host '  Finished.' -ForegroundColor Green
}
Write-Host ''
Show-Prose 'To undo: Not applicable.' 'DarkGray'
Write-Rule