vCenter Compute Dashboard

A web-based operations dashboard for VMware vSphere estates. It connects to one or more vCenter Servers, collects cluster, host, virtual machine, datastore and version data through VMware PowerCLI, and presents it as a single interactive dashboard covering capacity, redundancy, health, VM placement and ESXi version consistency across every site.

The repository also contains a companion PowerShell automation that emails a formatted daily health report and registers it as a scheduled task.

The system is designed to run inside an air-gapped environment. The web server has no internet access, so all front-end libraries are served locally and the dashboard makes no external requests.

---

Table of Contents

1. Components
2. Architecture and Data Flow
3. Dashboard Features
4. Cluster Metrics Reference
5. Repository Structure
6. Technology Stack
7. Prerequisites
8. Credentials File
9. Installation
10. Configuration
11. Backend API
12. Daily Health Report
13. Security Notes
14. Troubleshooting
15. License

---

Components

This repository contains two related but independent components.

1. Compute Dashboard (interactive web application)

• index.html - dashboard markup and navigation
• compute.js - all front-end logic and rendering
• style.css - dashboard styling, including dark mode
• run_compute.php - backend endpoint that runs the collector and returns JSON
• ComputeAnalysis.ps1 - PowerCLI collector that gathers all metrics from a vCenter
• assets/ - vendored third-party libraries (see Installation)

The user opens the dashboard in a browser, it requests data for each configured vCenter through the PHP endpoint, and it renders the results. Nothing is stored in a database; each refresh reflects the live state of the vCenter, subject to a short server-side cache.

2. Daily Health Report (scheduled automation)

• Send-HealthReport.ps1 - collects the estate state and emails an HTML report
• Setup-HealthReportTask.ps1 - registers the report as a daily scheduled task

This component runs headless on a schedule. It does not depend on the web application and produces its own self-contained HTML email.

---

Architecture and Data Flow

The dashboard has three layers: a browser front end, a PHP orchestration layer, and a PowerShell collector that talks to vCenter.

Browser (index.html + compute.js)
        |
        |  POST run_compute.php  { vcenter, forceRefresh }
        v
PHP (run_compute.php)
        |  - checks a 5-minute file cache
        |  - shells out to PowerShell when the cache is cold
        v
PowerShell (ComputeAnalysis.ps1)
        |  - reads credentials from vcenter.csv
        |  - connects with VMware PowerCLI
        |  - collects clusters, hosts, VMs, datastores, versions
        |  - emits a single JSON document on stdout
        v
PHP returns the JSON to the browser, which renders the dashboard.

Sequence for a full refresh:

1. On load, compute.js iterates the configured vCenter sites and issues one request per site to run_compute.php.
2. run_compute.php returns a cached response if one exists and is younger than five minutes; otherwise it executes ComputeAnalysis.ps1 for that vCenter.
3. ComputeAnalysis.ps1 connects to vCenter, collects all metrics, and prints a JSON document.
4. run_compute.php validates the JSON, caches it, and returns it.
5. compute.js aggregates the responses from all sites and renders the Overview, Health, ESXi and Analysis views.

The list of vCenter sites is defined in compute.js:

URL	Name
vcenter1.example.com	Site A
vcenter2.example.com	Site B
vcenter3.example.com	Site C
vcenter4.example.com	Site D

---

Dashboard Features

The dashboard is organized into views reachable from the top navigation bar.

Overview

A multi-site summary that aggregates every configured vCenter into one picture: cluster hotspots, host health issues, snapshot pressure and a searchable VM index built from all collected inventory.

Health

A morning-checklist view that evaluates the whole estate against a set of pass, warning and fail conditions, grouped into three categories:

• Availability and Protection: host connectivity, hardware status, HA enabled, active red alarms.
• Capacity and Performance: N+1 redundancy, host CPU and memory pressure, datastore capacity.
• Maintenance and Hygiene: host uptime, stale snapshots, VMware Tools status, connected installation media.

It also renders a compact host status grid per cluster showing state, CPU, memory, uptime and hardware health at a glance.

ESXi Version Consistency

A dedicated view whose purpose is to bring every host in the estate onto a single current ESXi version.

• It determines the estate target, defined as the newest ESXi version and build present on any host anywhere in the estate. Build numbers increase with each release, so the highest build is the most recent.
• It shows an estate-wide version distribution, with the target build highlighted and older builds listed below it.
• It reports, per cluster, how many hosts are behind the estate target and which ones, so upgrade work can be planned cluster by cluster.
• A cluster is only marked as on target when every one of its hosts is on the newest build; a cluster that is internally uniform but running an older build is correctly flagged as behind.

Sites

The manual entry point: a card per configured vCenter. Selecting a site runs a detailed analysis for that vCenter and jumps to the Analysis view.

Analysis

The per-cluster deep dive. For the selected vCenter it renders one card per cluster containing:

• A composite health score and CPU, memory and capacity metrics.
• N+1 redundancy status with the projected worst-case load after a single host failure.
• HA and DRS configuration, DRS automation level and active alarm count.
• Top resource-consuming VMs and a list of powered-off VMs with their allocated resources and virtual hardware version.
• VM health issues: outdated or missing VMware Tools, legacy virtual hardware versions, connected CD/DVD media, and disabled CPU or memory hot-add.
• Datastore capacity, sorted by usage.
• Per-host details, including a cross-cluster migration suggestion for hosts under memory pressure. The suggestion verifies that the target host can accept the VM by checking resource headroom, shared networks, shared datastores and CPU family compatibility (EVC) before recommending a destination.

Export

Export of the current analysis to CSV in several shapes (cluster and host metrics, full VM inventory, health issues). PDF export is present but disabled by default; because the server is offline, it requires the jsPDF and html2canvas libraries to be saved locally and referenced from index.html. The dashboard explains this when PDF export is selected.

Dark Mode

A light and dark theme toggle, persisted in the browser.

---

Cluster Metrics Reference

The following fields are the most significant ones produced per cluster by ComputeAnalysis.ps1 and consumed by the dashboard.

Field	Meaning
HealthScore	Composite 0-100 score derived from memory pressure, snapshots, N+1 status and VM health issues.
CpuUsage, MemUsage	Cluster-wide CPU and memory utilization as a percentage.
vCpuRatio	Provisioned virtual CPUs per physical core.
NPlus1Status	Whether the cluster can survive one host failure: Safe, At Risk or Insufficient.
NPlus1Detail	Projected worst-case memory and CPU percentage after removing the single most impactful host.
TotalPhysicalCores, TotalLogicalThreads	Physical cores and logical threads across connected hosts.
SnapshotCount, OldSnapshotCount	Total snapshots and those older than the configured age threshold.
PoweredOffCount, PoweredOffVMs	Powered-off VMs and a sample of the largest by allocated memory.
ToolsIssues, HWIssues, MediaMounted, HotAddIssues	VM-level hygiene findings.
Datastores	Per-datastore capacity and usage.
EsxiVersion	Per-cluster ESXi version data: each host's version and build, the cluster baseline, and per-host upgrade flags.
Hosts	Per-host CPU, memory, core count, uptime, hardware status and migration suggestion.
HAEnabled, DRSEnabled, DRSMode, RedAlarmCount	Cluster protection and alarm state.

Utilization is color-coded throughout the dashboard using these thresholds:

• Green: healthy, below 75 percent.
• Yellow: warning, from 75 up to 85 percent.
• Red: critical, at or above 85 percent.

---

Repository Structure

compute/
├── index.html                 Dashboard markup and navigation
├── compute.js                 Front-end logic, rendering and export
├── style.css                  Dashboard styling and dark mode
├── run_compute.php            Backend endpoint: cache and PowerShell orchestration
├── ComputeAnalysis.ps1        PowerCLI collector, emits JSON
├── Send-HealthReport.ps1      Daily HTML health report over email
├── Setup-HealthReportTask.ps1 Registers the daily scheduled task
├── README.md                  This document
├── .gitignore                 Excludes the assets directory
└── assets/                    Vendored libraries, not tracked in git
    ├── css/                   Bootstrap and Font Awesome styles
    ├── js/                    Bootstrap and SweetAlert2
    └── webfonts/              Font Awesome web fonts

The assets/ directory is intentionally excluded from version control by .gitignore. It holds third-party libraries and must be restored separately on a fresh checkout (see Installation).

---

Technology Stack

Front end:

• HTML5 and CSS3
• JavaScript (ES6+), no build step
• Bootstrap 5, Font Awesome and SweetAlert2, served locally

Backend:

• PHP for request handling, caching and process orchestration
• Windows PowerShell 5.1
• VMware PowerCLI for vCenter connectivity

The reference deployment uses XAMPP on Windows to serve the PHP application.

---

Prerequisites

• Windows Server, or Windows 10 or 11, with Windows PowerShell 5.1
• VMware PowerCLI installed for the account that runs the collector
• PHP 7.4 or later (for example via XAMPP)
• Network reachability and credentials for each vCenter Server
• The assets/ libraries restored into the project (see Installation)

---

Credentials File

Both the dashboard collector and the health report read vCenter credentials from a CSV file. The file maps a vCenter address to a username and password.

Format:

viserver,username,password
vcenter1.example.com,administrator@vsphere.local,ExamplePassword1
vcenter2.example.com,administrator@vsphere.local,ExamplePassword2
vcenter3.example.com,administrator@vsphere.local,ExamplePassword3
vcenter4.example.com,administrator@vsphere.local,ExamplePassword4

The viserver column must match the site URL configured in the application.

ComputeAnalysis.ps1 searches for the file in this order and uses the first that exists:

1. vcenter.csv in the script directory
2. %USERPROFILE%\Desktop\vcenter.csv

Send-HealthReport.ps1 uses its own list of candidate paths, configurable near the top of the script.

This file contains plaintext credentials. It must never be committed to version control and should be readable only by the account that runs the collector.

---

Installation

1. Clone the repository into the web root. For the reference XAMPP layout:

   C:\xampp\htdocs\powershell-app\compute\
   

2. Restore the assets/ directory. Because the server is offline and assets/ is not tracked in git, copy the Bootstrap, Font Awesome and SweetAlert2 files into assets/css, assets/js and assets/webfonts as referenced by index.html.

3. Install VMware PowerCLI for the account that will run the collector:

   Install-Module -Name VMware.PowerCLI -Scope AllUsers -Force

4. Create the credentials file (see Credentials File) in one of the searched locations.

5. Confirm the site list in compute.js matches your environment and the viserver values in the credentials file.

6. Start the web server. With XAMPP, start Apache and browse to the application, for example:

   http://localhost/powershell-app/compute/
   

   For a quick local test without XAMPP, PHP's built-in server can be used from the project directory:

   php -S localhost:8000
   

The dashboard loads all configured sites on startup. Because the account that Apache runs under is the identity that PowerCLI authenticates with for local resources, ensure that account has the access it needs.

---

Configuration

Collector thresholds are defined near the top of ComputeAnalysis.ps1:

Variable	Default	Purpose
SnapshotAgeDays	30	Snapshots older than this are flagged as stale.
MigrationMemThreshold	80	Host memory percentage above which a migration is suggested.
MigrationTargetCap	85	A migration target host is never pushed above this percentage.
AssumedVmCpuUtilization	0.20	Assumed average fraction of a VM's vCPUs actually in use, for target sizing.
NPlus1SafePct	85	After one host failure, survivors should stay below this percentage.
OldHWVersionNum	14	Virtual hardware at or below this version is treated as legacy.
HWUpgradeVersionNum	19	Virtual hardware below this version is treated as upgradeable.
ExcludeClusters	empty	Names of clusters to skip entirely.

Backend behavior is defined in run_compute.php:

• Responses are cached per vCenter for five minutes in a cache/ directory.
• A request may set forceRefresh to bypass the cache.
• Diagnostic output is written to compute_debug.log.

---

Backend API

POST run_compute.php

Request body (JSON):

{
  "vcenter": "vcenter1.example.com",
  "forceRefresh": false
}

Response (JSON, abbreviated):

{
  "success": true,
  "vCenter": "vcenter1.example.com",
  "generatedAt": "2026-01-01 09:00:00",
  "cached": false,
  "Clusters": [
    {
      "Name": "Cluster-01",
      "HealthScore": 92,
      "CpuUsage": 41.3,
      "MemUsage": 68.0,
      "vCpuRatio": 4.17,
      "NPlus1Status": "Safe",
      "HostCount": 3,
      "VmCount": 90,
      "SnapshotCount": 12,
      "EsxiVersion": {
        "BaseDisplay": "8.0.3 (build 25429389)",
        "Consistent": false,
        "UpgradeCount": 1,
        "HostCount": 3,
        "Hosts": [
          {
            "HostName": "esx-01.example.com",
            "Version": "8.0.3",
            "Build": "25429389",
            "Display": "8.0.3 (build 25429389)",
            "NeedsUpgrade": false
          }
        ],
        "Recommendation": "..."
      },
      "Hosts": [
        {
          "HostName": "esx-01.example.com",
          "ConnectionState": "Connected",
          "CpuUsage": 65.0,
          "MemUsage": 72.0,
          "CoreCount": 32
        }
      ]
    }
  ],
  "Errors": []
}

On failure the endpoint returns {"success": false, "message": "..."}. A single failing cluster does not abort the whole run; its error is recorded in the Errors array and the remaining clusters are still returned.

---

Daily Health Report

Send-HealthReport.ps1 is a standalone automation that produces a formatted HTML health report for the whole estate and emails it once per day. It is independent of the web dashboard.

What it does:

• Connects to each configured vCenter with PowerCLI.
• Evaluates the same classes of checks as the dashboard Health view: host connectivity and hardware, HA, active alarms, N+1 redundancy, host CPU and memory pressure, host uptime, stale snapshots, VMware Tools status and connected media.
• Builds a single self-contained HTML email using inline styles only, so it renders correctly in email clients without external resources.
• Sends the report by SMTP and also writes a timestamped copy, plus a latest.html, to an output directory, keeping the most recent copies.

Key configuration, near the top of the script:

Setting	Default
From address	vcenter-health@example.com
To address	it-team@example.com
SMTP server and port	smtp.example.com, port 25
Snapshot age threshold	7 days
CPU and memory thresholds	warning at 75 percent, critical at 85 percent
HTML output directory	health-report-html under the web application

The subject line is prefixed with the overall status, one of [OK], [WARNING] or [CRITICAL].

Scheduling

Setup-HealthReportTask.ps1 registers the report as a Windows scheduled task. Run it once from an elevated PowerShell session:

powershell.exe -ExecutionPolicy Bypass -File "Setup-HealthReportTask.ps1"

Defaults:

• Task name: vCenter-HealthReport, in the task folder \IT Operations
• Trigger: daily at 09:00
• Runs whether or not a user is logged on, with the highest privileges

To run it immediately for testing, or to inspect the last result:

Start-ScheduledTask -TaskName 'vCenter-HealthReport' -TaskPath '\IT Operations'
Get-ScheduledTaskInfo -TaskName 'vCenter-HealthReport' -TaskPath '\IT Operations'

To remove the task:

Unregister-ScheduledTask -TaskName 'vCenter-HealthReport' -Confirm:$false

---

Security Notes

• The credentials file holds plaintext vCenter passwords. Keep it outside the web root, restrict its file permissions, and never commit it.
• The dashboard performs read-only collection. It does not power VMs on or off, migrate them, or change any vCenter configuration. Migration entries are recommendations for an operator to act on, not automated actions.
• The application is intended for an internal, trusted network. It has no built-in authentication; place it behind existing network controls or add a web-server access control layer if wider exposure is a concern.
• The SMTP settings in the health report default to an unauthenticated relay on port 25. Adjust them to match your mail environment.

---

Troubleshooting

• Credentials file not found: confirm the CSV exists in one of the searched locations and that the account running the collector can read it.
• A site shows a connection error: verify network reachability to that vCenter, and that the viserver value in the CSV exactly matches the configured URL.
• Empty or partial data: check compute_debug.log in the application directory for the PowerShell output captured by the backend.
• Styling or icons missing: the assets/ directory has not been restored, or the files referenced by index.html are not present.
• PDF export unavailable: expected by default; save jsPDF and html2canvas locally and reference them from index.html as the dashboard describes.
• A host shows an unknown ESXi version: the host was likely disconnected or its product information was unavailable at collection time.

---

License

Proprietary. Intended for internal infrastructure use only.