dhicks

Is your wireless network set up as a "guest network" of some sort, with each device isolated from others on the network?

We have the same / very similar issues. Pupils log on to our Macs using their Google credentials. We use Moysle as an MDM, this allows us to present the login-with-Google process (complete with 2FA step for staff) as the login mechanism. As you've found, the machines seem to be filling up with profiles quite quickly, even though pupils should mostly be using Google Drive for storage. I need some time to sit and investigate a bit better, my plan being to have a login / periodic script that does a basic "disk free" check and, if below a certain threashold, clears out the oldest user profiles (going by creation / modified date of the folder in /users). Moysle scipts seem to be written in Bash, which I find slower to write.

We have a script run nightly that uses GAM to export a bunch of data about ChromeOS devices, then parses that data to get the most recent user and adds that to a spreadsheet. We keep the who-used-this-device data for a month, which can be handy if a child wanders off with another child's Chromebook.

For general display purposes (dashboards, slideshows, etc), we use ChromeOS devices in kiosk mode, which are then managable from the Google Workspace admin console. We tend to use cheap (£200) mini PCs from Amazon with ChromeOS Flex installed, but dedicated devices are available, including OPS devices if you want a neater, all-in-one solution. Otherwise, Raspberry Pi OS might be an option - either with a Raspberry Pi (I find the older Pi 3s can generally be powered from the USB port of most modern "smart" TVs, but the newer Pi 4s and 5s take a bit too much power, so you need a separate plug for the power supply) or, again, a cheap Mini PC formatted with Raspberry Pi OS. I had a script from a while back that could set up a default Raspberry Pi OS install (actual Pi or Intel hardware) to be a kiosk device: https://github.com/dhicks6345789/device-config The script might well need updating as I haven't touched it in a few years.

Hmm, I wonder if this might impact some of our parents.

If it's any help: no, we haven't (that I know of), and we have a similar-ish setup to yourself. Mostly Office 2019 on our Windows devices, with some now moving to Office 365, with the Google Drive client installed. We've not done anything to change cache size / bandwidth limits from the default. These are school devices, used from school with a nice gigabit fiber connection, so we haven't tried them in a limited-bandwidth situation.

We have similar hardware, with similar storage, network and processing capacity, however most of our services are now cloud based. I was thinking of maybe looking at something like Synology's larger rackmount devices, which if I remember correctly could also handle virtulisation / containerised services, we might be able to buy a simple all-in-one server to run our half-dozen or so VMs on.

Google Workspace can use a number of 2FA mechanisms, we've found physical USB keys (as you suggested) have worked well over the past couple of years or so. A previous post with some details: That's the Feitian K9, now around £20 each. We also use the NFC-capable keys as tokens for the door entry system (Paxton) and for printer access (PaperCut), we find that having one key for those things (especially getting into the building of a morning) seems to help the staff think of the USB keys as like house/car keys, and we have very few problems with lost keys (and I don't think anyone has managed to break one yet). We have nice, compact all-in-one PCs as classroom desktops, which tend to have the USB ports hidden around the side. We found we needed to buy a bunch of 30cm USB extension cables so people could easily plug the USB keys in. We don't enforce 2FA for pupil accounts yet, although we do allow it, so if our older pupils wanted to add 2FA to their accounts they could.

Another possible option is the Google redential Provider for Windows (GCPW), which lets users log lon to a Windows machine with their Google credentials. A search for "GCPW" on these forums or Google should get you further details.

I had a quick try just now on my work machine, but couldn't get it to pick up the integrated camera. It did pick up my HDMI/USB adaptor, though. I'm on an older Mac running ChromeOS, so probably something of an edge-case for compatability, although I did try on my own web page and it picked up both cameras okay.

The new release of Debian (version 13, codename "Trixie") is due this Saturday. I'll probably be switching my current (home) 2017-era Windows 10 Dell Inspiron 5567 over to that, not sure if that's installable on older Macs.

We had a similar problem with some snazzy OPS-slot PCs inside touch screens. The screen itself acts as a shield for WiFi and signals for the wireless keyboard - the OPS slot PC tucks inside and behind the screen. In the end we had to put the wireless keyboard receiver on a USB extension cable dangling down underneath the board for it to pick up a signal.

Resurrecting an older thread here - just got around to trying this, still no luck. Downloaded latest available version of GCPW, tried on a test machine and as soon as I've put my Google username and password in it skips straight past the "insert your USB key" to "There was an error - try another way".

I found CFM Systems' "Introduction to PowerBI" webinar to be very useful - understandable, not too long: https://cfmsystems.com/

We don't use Bromcom, we use Paxton / Inventry for door entry and iSAMS as an MIS. I was planning to look at creating a dashboard-style display showing a list of who's-in-the-building in some way, so staff and pupils could see who duty staff and so on are. I was thinking of that as an HTML / Javascript page loading maybe a CSV / JSON file generated server-side every minute or so. I'd guess it should be possible to get the same sort of data from Bromcom, so if there's no existing solution this might be something I could spend a bit of time looking at.

I was going to write my own dashboard-like display - probably a fairly simple web page (HTML / CSS (Bootstrap library) and Javascript) that loads a CSV file produced by a server-side script, periodically (every minute?) getting updates.

Hmm. Looking at the Google Admin Console user interface, I can see Admin -> Generative AI -> Gemini For Workspace, then a panel for "Feature Access" which lists Drive & Docs, GMail, Google Chat, Google Meet, all as on/off toggles. For some reason, Classroom is split out into its own panel. The appropriate API endpoints for the above are possibly (?) somewhere in the Google Admin SDK API -> Directory API -> resources.features: https://developers.google.com/workspace/admin/directory/reference/rest/v1/resources.features I'd need to write some code and start to figure out what came out of some of the calls to the above API to figure out if that was the right part of the API to be using.

Similar-ish: we would like a way for teaching staff to easily be able to enable / disable Gemini AI features for groups of pupils (generally by class / yeargroup). Ideally, a nice easy-to-use interface where you select the group of pupils and have an on/off button you can press. We could write the user interface ourselves (quite possibly using Gemini AI code assistance!), it's just a case of finding the correct Google admin API call for the backend to call to turn features off / on.

Yes, of course. I should be available most of today or next week - I'll send you a separate private message. With our setup, we get data via the iSAMS "batch" API once per night, then process that data into various places (class / yeargroup lists, etc). We also use the data for account creation and so on. We have a set of Python scripts do that - the script that gets the data in XML format from iSAMS is fairly simple, it should convert to other languages easily enough if wanted, or i suspect Power BI might be able to get and parse the XML data directly itself these days. For reporting, we have a separate group-wide project that is hauling data out of various systems to produce reports, so we're aiming not to reproduce any of that functionality just for our school. I've tended to use Python for reporting with the ReportLab library, which produces PDF output - snazzier Javascript-based report libraries are available, but that's not something I've looked into much myself.

Probably easier to ask Salamander directly, but if I understand correctly their new "cloud portal" product is a cloud-hosted portal to control your instance of Salamander's software installed on your server(s). So, you log on to a nice web-based service, accesible from any browser, rather than have to be on your local system or even on a specific server. I suspect by "cloud hosting" you're maybe aiming to have the server that runs Salamnder hosted off-site, which should be perfectly do-able with a number of providers (AWS, Google, etc) - it's something we've looked at, as it would make for one less VM to be hosted on-site.

Possibly not so much these days. The traditional training route for a software developer (GCSE / A-levels in maths / science, computer science degree or similar) has opened up somewhat, there are more routes available now. If you're doing "data science" of some sort then a basic knowledge of statistics would be a good idea, but a fair bit of software development, especially front-end work, is probably not going to need a huge amount of maths. A few years ago there was a fashion for coding "boot camps" - intensive courses, some over just a few weeks, which would aim to teach you the basics of software development, generally streamlined by concentrating on a specific platform / framework. I'm guessing such things are still available, possibly now with added "vibe coding", using AI tools to do a fair chunk of the development. Actually, starting from scratch as a programmer now, you might do quite nicely picking a platform and an AI tool and seeing how far you can get. Node.js seems to be a popular choice for AI coding tools to generate code for, so that might be a reasonable platform to aim for - from your existing "IT" side of things, you can probably figure out quite nicely how best to set up, securley host and so on a Node.js environment, which would give you a head start over a lot of others. Claude Code seems to be a popular AI coding tool at the moment or, if you're in a school, maybe have a look at Microsoft's Copilot or Google's Gemini if your school has those available. I'll hopefully get some time to give Gemini a bit of a go soon - I find it helpful with this sort of thing to have a well-defined target in mind, so I have a small web-based application I'm aiming to create, I'll see what help Gemini can be with that.

There's not a huge difference - the actual programming bit is easy, it's the problem solving / extracting requirements from the customer / end user that tends to be the more skillful part. Do you already know a particular language / framework / platform / IDE or whatever? Most developers would have a "tech stack" they know and work with on a regular basis - for instance, I tend to work with a Python or Go backend and tend to write self-contained HTML / CSS / JavaScript front ends (using the Bootstrap library) for web-based applications that call the backend via API calls (so like a self-contained application that runs in the browser, with remote procedure calls for server-based functionality). I find that a fairly solid, easily-hostable (on a small, cheap Linux or Windows VM), flexible kind of setup, but more the more modern platform seems to be Node.js with various frameworks (Angular / React, etc).

We get data from iSAMS via their API, rather than directly as an SQL query. Their "batch" API is actually more like a query language anyway - you get a bunch of XML data back that you can then process into whatever format you prefer, or put directly into something like PowerBI. If starting a project, I would probably aim to go with a tool like PowerBI (although, myself, I would probably choose a more open tool that allowed me more control over what I did with the resulting reports), SSRS is a bit more in the previous generation of data analysis tools.

CoPilot, then, in your case. I think, similar to Google's Gemini, you get a certain amount of functionality for free / included in a standard license, you can pay to license further functionality. A quick Google search suggests around £23 per user per month for the full licnese, which could get pricey for all staff in a school, so maybe start with a smaller, more select group of users, and/or discuss with whoever handles your Microsoft licensing to see if there is any bulk / schools pricing.

Ah ha. We're just looking at the licesning costs for a whole-school license, which I'm pretty sure includes most if not all features available. Is this people using the voice assistant from in front of desktop machines, with some kind of built-in microphone, or do you have specific microphones / devices of some kind? Most of our classroom and office machines are reasonably new all-in-one desktops, with a few NUC-alikes - most of our devices should have a small built-in array mic or a mic ona webcam.