Files
rb/scripts/od-backup-check.ps1
igodwin dfe37d3e4a Add od-backup-schedule runbook + od-backup-check script: schedule & monitor backups
Automate the nightly cold backup (Task Scheduler, filled-in local copy in
private tier), stagger the off-site upload, and MONITOR with three layers:
Task Scheduler last-run, a read-only health check (freshness/completeness/
size), and a dead-man's-switch heartbeat that pings an external monitor only
on success so silent failures and offline servers get caught. od-backup-check.ps1
is read-only (no DB/service), iex-safe, and pings <HEARTBEAT_URL> on PASS.
Cross-linked with od-db-backup.md and od-backup-verify.md; README updated.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-13 01:07:47 -07:00

148 lines
6.7 KiB
PowerShell

<#
.SYNOPSIS
Read-only health check of the latest Open Dental cold backup: freshness,
completeness, and size sanity. Optionally pings a dead-man's-switch heartbeat
URL on success so an external monitor alerts when a backup silently stops.
.DESCRIPTION
Points at the backup destination root (<DEST>) produced by od-db-backup.ps1,
finds the newest od-backup-<timestamp> folder, and verifies it is:
- FRESH : written within the max-age window (default 24h)
- COMPLETE : has data\, (optionally) OpenDentImages\, MANIFEST.txt, and a
backup.log that ends in success with no ERROR/CRITICAL lines
- SANE SIZE : data\ isn't near-empty, and isn't a fraction of the prior run
Prints a clear PASS/FAIL summary, appends a line to <DEST>\backup-check.log,
and — if healthy and a heartbeat URL was given — pings it. Pinging ONLY on
success is deliberate: if this check fails, crashes, or the whole server is
off, no ping arrives and the external monitor raises the alarm. "No news" is
never mistaken for good news.
Fully READ-ONLY: it does not touch the database, the service, or the backup
files. Safe to schedule and safe to run ad hoc.
.NOTES
Manual spot-check:
irm https://gitea.ivangodwin.com/ops/rb/raw/branch/main/scripts/od-backup-check.ps1 | iex
For unattended monitoring, schedule a filled-in LOCAL copy (with <DEST> and the
heartbeat URL baked in) — keep that copy in the private tier, not here. See
od-backup-schedule.md.
iex-safe: uses `return`, never `exit` (which would close an interactive
session). PUBLIC REPO: no client specifics hard-coded; prompt at run time.
#>
$ErrorActionPreference = 'Stop'
Write-Host '== Open Dental: backup health check (read-only) ==' -ForegroundColor Cyan
# --- Prompts (no param block; iex-safe) ---
$destRoot = Read-Host 'Backup destination root (<DEST>, e.g. E:\OD-Backups)'
if (-not (Test-Path $destRoot)) {
Write-Warning "Destination not found: $destRoot"
return
}
$maxAgeInput = Read-Host 'Max age (hours) for a healthy backup [24]'
$maxAgeHrs = if ($maxAgeInput -match '^\d+$') { [int]$maxAgeInput } else { 24 }
$expectImages = (Read-Host 'Should each backup include an OpenDentImages folder? (Y/n)') -ne 'n'
$heartbeat = Read-Host 'Heartbeat/ping URL to hit on SUCCESS (<HEARTBEAT_URL>) [blank to skip]'
$problems = New-Object System.Collections.Generic.List[string]
function Fail($msg) { $problems.Add($msg) }
function Get-DirSizeBytes($path) {
if (-not (Test-Path $path)) { return $null }
(Get-ChildItem -LiteralPath $path -Recurse -Force -File -ErrorAction SilentlyContinue |
Measure-Object -Property Length -Sum).Sum
}
# --- Find the newest backup folder ---
$folders = @(Get-ChildItem -LiteralPath $destRoot -Directory -Filter 'od-backup-*' -ErrorAction SilentlyContinue |
Sort-Object LastWriteTime -Descending)
if ($folders.Count -eq 0) {
Fail "No od-backup-* folders found under $destRoot."
}
else {
$latest = $folders[0]
Write-Host "Latest backup: $($latest.Name) ($($latest.LastWriteTime))" -ForegroundColor Green
# 1) Freshness
$ageHrs = [math]::Round(((Get-Date) - $latest.LastWriteTime).TotalHours, 1)
if ($ageHrs -gt $maxAgeHrs) {
Fail "Stale: newest backup is $ageHrs h old (limit $maxAgeHrs h). Backup may have stopped running."
}
if (((Get-Date) - $latest.LastWriteTime).TotalMinutes -lt 5) {
Write-Warning 'Newest folder was written <5 min ago — a backup may be in progress. Re-check after it finishes.'
}
# 2) Completeness
$dataDir = Join-Path $latest.FullName 'data'
$imgDir = Join-Path $latest.FullName 'OpenDentImages'
$manifest = Join-Path $latest.FullName 'MANIFEST.txt'
$logFile = Join-Path $latest.FullName 'backup.log'
if (-not (Test-Path $dataDir)) { Fail "Missing data\ directory in $($latest.Name)." }
if (-not (Test-Path $manifest)) { Fail "Missing MANIFEST.txt (backup may not have completed) in $($latest.Name)." }
if ($expectImages -and -not (Test-Path $imgDir)) {
Fail "Missing OpenDentImages\ in $($latest.Name) — a DB-only backup loses all documents."
}
if (Test-Path $logFile) {
$tail = Get-Content -LiteralPath $logFile -Tail 40 -ErrorAction SilentlyContinue
if ($tail -match 'ERROR|CRITICAL') { Fail "backup.log contains ERROR/CRITICAL lines in $($latest.Name)." }
if (-not ($tail -match 'completed successfully')) {
Fail "backup.log does not show a successful completion in $($latest.Name)."
}
} else {
Fail "Missing backup.log in $($latest.Name)."
}
# 3) Size sanity
$dataBytes = Get-DirSizeBytes $dataDir
if ($null -ne $dataBytes) {
$dataMB = [math]::Round($dataBytes / 1MB, 1)
Write-Host " data\ size: $dataMB MB"
if ($dataBytes -lt 1MB) { Fail "data\ is only $dataMB MB — implausibly small; likely a broken/partial backup." }
if ($folders.Count -gt 1) {
$prevData = Get-DirSizeBytes (Join-Path $folders[1].FullName 'data')
if ($prevData -and $dataBytes -lt ($prevData * 0.5)) {
Fail ("data\ shrank to {0}% of the previous run ({1} MB -> {2} MB) — possible truncation." -f `
[math]::Round($dataBytes / $prevData * 100), [math]::Round($prevData/1MB,1), $dataMB)
}
}
}
}
# --- Report ---
$healthy = ($problems.Count -eq 0)
$stamp = Get-Date -Format 'yyyy-MM-dd HH:mm:ss'
Write-Host ''
if ($healthy) {
Write-Host "PASS Backup is healthy as of $stamp." -ForegroundColor Green
} else {
Write-Host "FAIL Backup health problems:" -ForegroundColor Red
$problems | ForEach-Object { Write-Host " - $_" -ForegroundColor Red }
}
# On-disk trail at the destination
try {
$line = "$stamp $(if($healthy){'PASS'}else{'FAIL'}) $(if($healthy){'healthy'}else{$problems -join ' | '})"
Add-Content -LiteralPath (Join-Path $destRoot 'backup-check.log') -Value $line
} catch { Write-Warning "Could not write backup-check.log: $($_.Exception.Message)" }
# --- Dead-man's-switch heartbeat: ping ONLY on success ---
if ($heartbeat) {
if ($healthy) {
try {
Invoke-WebRequest -Uri $heartbeat -UseBasicParsing -TimeoutSec 20 -Method Get | Out-Null
Write-Host 'Heartbeat ping sent.' -ForegroundColor Green
} catch {
Write-Warning "Heartbeat ping failed: $($_.Exception.Message) (monitor will alarm on the missing check-in)."
}
} else {
Write-Host 'Unhealthy — NOT pinging heartbeat. The external monitor should alert on the missing ping.' -ForegroundColor Yellow
Write-Host '(healthchecks.io users: append /fail to the URL for an immediate alert.)'
}
}
if (-not $healthy) { Write-Warning 'Investigate now — see od-backup-schedule.md triage.' }