Each device now migrates on the day of its date tag instead of on a
looping 24-hour clock: earlier days are finished, today's devices go
through backup and restore during working hours, later days haven't
started. The few that don't finish have a cause the dashboard shows
(broken client, broken WMI, low disk, a USMT error, unreachable PC).
A simulated team keeps two weeks of working days tagged. During working
hours it adds devices, reschedules or drops one, tags stuck PCs with
their problem, then fixes them and moves them to a new day; each morning
finished days are closed out and a new day is scheduled. Actions print
in the worker terminal as [Team] lines. Settings live in Demo.Team.
The visitor sandbox now keeps per-tag edits laid over the live roster,
so the team's changes keep showing around a visitor's own edits.
Overview calendar: today is a filled badge on the date number, so it no
longer looks like the selected day (outline plus tint).
First load:
- Draw the Overview as soon as the data arrives instead of on its next
2-second timer tick (about 2.5 s to 0.6 s on a typical connection).
- Start on DOMContentLoaded instead of window.onload.
- Load the ~900 KB PDF library only when someone exports a PDF.
- Fetch the startup files side by side, and build the docs search index
after the dashboard has drawn (unless the visit starts on /docs).
- Show a loading placeholder on the Overview until the data arrives.
Demo: the date-batch tags sit on fixed working-day offsets from today.
The seed records them in Demo-Dates.json and the worker relabels them when
the day changes, so a long-running container never looks stale.
Worker: pings every tracked device on a presence sweep (hourly in business
hours plus midnight) and records last online; inventory reports whether a
device was found in AD and SCCM; monitors from WmiMonitorID; snapshots of
the last reachable state replace "Cached"; newly tracked devices are swept
and published first; totals count source devices only; per-type timing in
Worker-State.
Front end: details dialog shows presence, reasons instead of N/A,
"Waiting for worker...", monitors and snapshot times, and refreshes in
place; Custom View gains Days Offline, Last Online, Found In and monitor
columns; manual pair changes show immediately; sync dialog shows timing by
type. Custom View gets a large scrollbar; "Exactly ticked" filter.
Remove all environment-specific data and make the dashboard adoptable
anywhere, then add a demo that runs the real PowerShell services.
- Strip cached device data, internal knowledge-base articles and
screenshots, server/site names and the old project notes; replace the
knowledge base with generic sample articles and diagrams
- Move every environment value into Config/usmt-dash.psd1
- Route all infrastructure access in the worker and Get-Software.ps1
through a provider (Providers/SCCM.ps1), with a config-driven asset
inventory view instead of a hard-coded custom class
- Add Providers/Demo.ps1: a deterministic simulated fleet whose
migrations progress in real time, and Demo/New-DemoData.ps1 to seed it
- Server: -Demo makes the API read-only; add get-config and get-logs;
serve static files only from web/, Documentation/ and an allowlist of
Data/ files with a path-traversal guard; gzip large responses
- Mirror both services' console output to Data/Logs via a Write-Host
proxy, and add a terminal panel to the dashboard that streams it
- Demo sandbox (web/demo/demo.js): visitor changes to tags, articles and
images are kept in the visitor's browser and echoed in the terminals;
nothing a visitor does reaches the server
- Vendor marked, highlight.js and html2pdf.js instead of loading CDNs
- Dockerfile and compose file for hosting the demo (Coolify-ready)
Johan Sandoval