Posts by Lee D

OVH Cloud warns of 87% price hikes to help it cover RAMpocalypse costs

The OpenVPN clients at home dial into the fixed, static-IP of the server to connect to the OpenVPN server there. That external server has an IP whitelist to only allow OpenVPN connections from my IPs (static IPs and dynamic DNS that my router automatically updates every 10 minutes or so), so nothing else can connect to OpenVPN at the external server.

Because OpenVPN assigns a 10. network to the connection (and I don't use 10. at home or at the server end), the only traffic that goes out over the VPN is that coming from the server itself (reverse proxy requests down to the home servers) and their responses. Firewalling at both ends (with UFW) means that nothing else can go over that interface but HTTP(S) and ICMP ping at either end of it.

The external server is then just set to use Apache reverse proxying to access the "10." address of one of the home machines for its content, depending on the service required. Multiple machines from my home each dial in on their own separate VPN connection to the external service, and each get a unique "10." address which allows the server to contact them. So there's no complex network configuration that could be mis-read or bring unintentional remote access to the whole subnet. Each machine at home has their own certificates for OpenVPN and they each verify the server's OpenVPN certificate on connection. They initiate the connections to the server, the server assigns them their (fixed) 10. IP based on their certificate, and then the Apache sites push any traffic for that service down the VPN connection via reverse proxy to the 10. addresses.

I went a stage further and made Apache not just reverse proxy, but also "load balance" so that if the 10. service is unavailable (i.e. the VPN is down), then it goes to "localhost" and retrieves a page that says that the VPN must be down. That was actually surprisingly simple.

So homeclient1 triggers an OpenVPN connection on itself to connect to externalserver's OpenVPN port.

externalserver accepts the connection and gives homeclient1, say, 10.0.0.1.

mydomain1 points to externalserver's public static IP address.

externalserver's Apache listens on that IP and then has a "site" defined for each service.

The Apache config of that site tells Apache to reverse proxy all accesses (except for LetsEncrypt /.well-known/acme) to the 10.0.0.1 address, which proxies the requests/responses to homeclient1's web server to actually serve content.

homeclient2 also connects to the server and is given 10.0.0.2, and Apache does the same for mydomain2 to that address. And so on.

If you're interested, I can post some Github gists's with the critical parts of my config.

I use OVH to operate a small dedicated server, that I use for a very complex home setup(*).

Given that it's a dedicated server, the same one I've been renting for years, and that it's never needed intervention in all the years I've had it, if they raise my prices by anywhere NEAR that amount... I'll find another host.

Sure, I'm not their target market but equally... you aren't gonna penalise me because your other customers want something expensive. I'm not there to subsidise your other customers. New servers, fine. Upgrades, fine. Those people who rent virtual servers and the like where your costs might increase, fine. But people with existing dedicated servers shouldn't be seeing a price hike unless they make changes, or the price of electricity changes.

(*) My external domains are pointed to the dedicated server. That means that they sit on a static, globally-accessible, always-on IP. They use LetsEncrypt to renew, so they're always online and "available". But the actual SITE they are providing is in my house. My home systems VPN up into the server (which avoids all kinds of port-forwarding, NAT, even double-NAT, etc. nightmares) from any of 5G, DSL or fibre connections. The dedicated server then uses Apache to reserve-proxy and load-balance the sites, getting the pages from my home systems via the VPN (with a special "fallback" load balancer which basically serves a page which says "VPN is down" instead of the site when things are disconnected from home).

No ports to open. No dynamic IP worries. No NAT-punching. No changes when my connection fails over. Static locations to access everything even internationally. TLS-wrapped and authenticated at the dedicated server end. Nothing else open or accessible on my home network to make it work.

Given that I host everything from my home TV (tvheadend), Plex, my car GPS tracker (Traccar), etc. on them, it works amazingly well and nothing "external" cares about my internal setup. My GPS tracker just sends the data to my dedicated server and it gets authenticated and proxied all the way back to my home system via secured channels (TLS, VPN, etc.).

I've gone from writing about SD-WAN to depending on it

(shrug)

I currently have a Draytek router, pretty much the stuff I've been running for the last 20 years, and before that I used (and contributed to) Freesco - a project to build a single-floppy Linux router distro. Both are/were capable of everything you describe.

There's nothing here that's special.

My router is currently balancing a DSL connection, a 5G connection (in fact, two as it's dual-SIM! And I can add another USB one with a dongle I keep in my laptop bag), and an Ethernet connection (which is about to be hooked up to 900Mbps fibre... tomorrow with any luck!). It has the routing to handle it all at max speed as well as the failover and even HA features to keep a hot-spare Draytek online should mine fail (I feel that's a BIT overkill for my purposes).

It fails over between connections, it load balances, I can prioritise routes for given IPs or protocols, choose specific routes for individual DNS paths (so my internal services are resolved internally), it applies QoS properly so I don't need to worry about knocking out a webcam session or increasing game latency when someone's using the Internet heavily on any connection, it even nicely updates a dynamic DNS address that's hosted on my own domain so that nothing needs to change to access home remotely (and, yes, I have a static IP on one of those connections anyway!).

(But I also employ a VPN punch-out that connects from home out through the NAT to a dedicated server that I rent for a pittance, which reverse-proxies everything back home however it can do so, over whatever connection and even fails over nicely to a screen that says "Whoops, looks like the home Internet is down!" when it can't... using basic Apache load balancing and reverse-proxy features. This means that I never have to "forward a port" to make my home systems appear online, on a static IP, and I can secure my internal services from outside access using authentication and having LetsEncrypt handle the TLS at the server end to then tunnel the traffic over the VPN back up).

Hell, the router natively supports port-knocking to open up ports and services, VPN, web filtering, OSPF, BGP, VLAN, 802.1x, LDAP, .... everything you'd expect it to support.

I don't know where this "SD-WAN" stuff comes from... it's just a term applied to basic networking and VPN etc. as far as I can tell. I've been "SD-WAN" since the days of running an old P90 with only a floppy disk to do much of the above (while being capable of all the above) for our 10Base2 network via ISA cards and 56K dial-up modems.

I live rurally, and we have a lot of power cuts, but DSL has proven reliable - if slow - for the most part and the 5G rarely kicks in. But I have FTTP coming soon and that's just going to plug straight into the same box. I would estimate... based on previous house moves and network reconfigurations and router replacements... about 5-10 minutes of config on that router to get everything including the fibre connection into all the balancing, QoS, firewalling, routing and path-based decisions (like which interfaces to prefer for NTP, DNS, VPN, etc.).

I did consider Starlink briefly but I can't bring myself to fund Musk to the tune of even a penny. I'd rather be offline.

But there's nothing special about this. This is just "IT" or "networking" guys doing what they do, day-in, day-out on a home or small business router.

Scot NHS trust probes access to medical records of 9-year-old girl after man arrested on suspicion of murder

Re: just idiots at this point

Need to start locking people up, so they lose their career and access to that data.

Then maybe they'll start changing their behaviour.

Sysadmin summoned to explain italics – to a user with at least two degrees

Re: the lawyer got angry

People always shoot the messenger.

Especially when the message is "You're a dumbass, and everyone else thinks so too".

Techie lured out of retirement to support software only he remembered

It's like the old resiliency/redundancy argument.

"But we're paying for an entire server, specced powerfully enough to run the company, to sit and do nothing most of the time!"

"Yes."

Re: Not quite the same (not retired yet)

The last place I left, I had trained up the only other IT guy there for nearly a decade to the point that he could take it over.

I warned that place I was going if they didn't treat us better, they didn't care a bit.

I left, keeping them informed at EVERY stage of... quite literally the first job offer that came up. Interest, interview, offer, acceptance, notice period, everything that they should know. They couldn't care less.

They asked who would take over. Well... there's only one choice, isn't there? Because you wouldn't let me have a larger team despite have a dozen of your own consultants and audits telling you that it was necessary. It would be the guy I trained up.

I warned them that he would also leave unless they treated him better. They didn't care.

So I left. They had a quick panic but by that time I had no interest (tried to "lure" me back with 2% raise... I'd already told them that the other was offering +20%, and 2% wasn't a raise, it was an insult).

The other guy took over. They suddenly realised that they SHOULD have someone else on the team, hired a complete idiot for him to manage while he tried to run the place on his own... that was worth than anything else.

Now, we had documented EVERYTHING. It was one of the things that the audits loved. Everything was written down. But nobody cared.

That other guy... he left a year or so later, tired of being the scapegoat for everything, not getting any further raises, being paid far less than I had been, and having to deal with an employee who was basically NEVER there and wasn't any use when he was. He got a far better job elsewhere, again almost immediately he looked.

So they hired out again. He asked to be involved in the hiring process, they refused. They finally relented when they realised that they didn't know HOW to hire someone who could do the job. He helped out and put in a firm recommendation for some candidates and a firm recommendation AGAINST one particular one. Guess which one they hired?

He warned them again, he warned them again when that guy started, he warned them that he had no clue and was actively breaking systems... they didn't care.

As my friend walked out the door (the 2nd in his position in as many years!), he watched the new guy delete the documentation VM, stop all the critical access control services, remove all the notes on that machine NOT to stop those services, all the funny hoops you have to jump through to restart them (notes I'd written and which I doubt I could remember myself without those notes!) etc. etc. Literally as my friend left on his last day, the doors locked and he couldn't get out because of what the guy did (fortunately, he and I installed that system, so he knew a bypass, and walked out... every door flashing like mad as it lost connection to the system...).

This was a Friday. The very next day, he's getting a barrage of calls. He ignores them all weekend. Monday morning our old workplace call him at his new job. Absolutely nobody can get access to the site. They're all locked you. "You have a new guy now... I handed everything over to them... I explained every system... I left copious notes and documentation... it's his problem."

In the end, he went back for ONE DAY at stupendous rates and - the guy had basically destroyed everything. The access control system was trashed and my friend restored from one of our backups. The servers were turned off for no reason. Anything the new guy didn't understand or like, he'd deleted... including the backups... Even the helpdesk - he'd trashed it and replaced it with some ad-based online shareware junk without telling anyone and now everyone had ads popping up on their computers, prompting numerous complaints. Everything was deleted, damaged, turned off or replaced with something that didn't work.

My friend and I both got calls, texts, messages, over the next year or so from various people there - a new IT director, the estates manager, etc. etc. ... my friend got them all the time, me not so regularly because I think they were utterly embarrassed at having to go that far back - I hadn't worked for them for 2 years by that point. We never answered them. They still try occasionally.

I wouldn't like to think what state their systems are in now.

AI slop pollutes the CVE pipeline with fake vulns

So at what point are we just going to go back to entirely ignoring slop, regardless of its source? And not pumping trillions into this worthless junk?

Three becomes one as Vodafone buys out merger partner

Re: Time to move.

I still have the original letters, in post-marked envelopes, from when a car insurer decided to retro-actively cancel my car insurance "for non-payment" despite that being news to both myself and the bank (who confirmed that the normal DD had been taken at the normal time by the company).

I got several letters from them in quick succession. Literally sent me one letter dated two weeks in the past (in an envelope postmarked the day before) to tell me it was cancelling my insurance for non-payment from a date... that had already passed in the previous week. Cited non-payment despite the fact that I also received from them a letter telling me explicitly that my payment had been taken successfully and the phrase "nothing more to pay". Which was dated the day before the above first letter.

Cue a LONG series of arguments with them about it not being possible, until I provided both "document reference numbers" (huge long /'d numbers on the bottom of each letter) which quite literally prompted an "Okay, hold on.... Oh. Oh!" from the person on the other end.

Basically they had cocked up, taken my payment as normal (confirmed with the bank), send me a letter to tell me that it was paid up, then waited two weeks, then sent a letter claiming that the payment had not gone through (it had) and that they were going to cancel my insurance, and then they sent another telling me they HAD cancelled my insurance because I'd ignored that letter. Despite the fact that the letters had all arrived together with the same post-mark date.

Apparently falsifying dates in those letters is a bit of a no-no, especially when those dates don't reflect reality, and when one is in the past and one is in the future but both got posted the same day. They absolutely shat themselves when they finally found all the letters that they'd sent me "on the system". It was immediately escalated and I got a lot of begging phone calls from, effectively, the top bod in that company asking me not to report them, that this could have left me marked as uninsured through no fault of my own, that it breached all their industry regulations, and please please don't tell anyone. I said if ANYTHING flags up for me driving uninsured on any day where I was SUPPOSED to have been covered, they'd be in court.

Couldn't trust them after that, cancelled the policy anyway (and got immediate confirmation of that, a big refund, and a letter from them clarifying certain things, without an actual admission of guilt, but so that I could present it in court if it ever came up), and about a year or so later they went bust.

Glad I held onto the letters because about 2 years later, their administrators tried to chase me for a debt of exactly the amount I was refunded and I told them where to go.

Excuses like 'AI did it' don't exist in the eyes of the law

Much like "the self-driving car crashed and killed someone".

Until THAT THING ITSELF is a legal entity that's solely taking responsibility, then I'm not going to let it have control of things I'm responsible for.

Whether that's "driving" or "data security".

And guess what happens when you try to get self-driving cars or AI models to be legally responsible for their output? Nobody's interested in giving you that guarantee and prices skyrocket (much like prices of aviation components, etc.).

Sorry, but I can have the power and the responsibility - or neither. It's not a mix-and-match situation.

Tech sector pours $1T into AI and sends customers the bill

I'm still of the opinion that, to pay this lot back, AI token cost has got to rise at least ten-fold to recoup their losses, pay back their loans, keep the lights on, etc.

I'm pretty sure that Copilot, ChatGPT, et al are just not going to be that useful or popular at 10 times the current cost.

Auditors tell UK government to do the math before banking on £45B AI savings

Nope

I will bet a significant amount of money that all the "savings" we see from AI in government will be wiped out by... the cost of AI in government.

Major glitch in Matrix as UK government projects watchdog rates shared services cluster red

Re: Idiocy

They can't even source consultants to test the system for them, even with offering backhanders, what makes you think they have the first clue about how to write an entire system?

Those kinds of skills are LONG GONE from the civil services.

AWS customer learns the hard way how even the smallest oversight can be mission-critical

Sorry, but this guy is just not a professional outfit.

Nothing documented. Nothing backed up. Nothing monitored. No break-glass accounts. Emails not updated and on a separate domain that you have access to.

Just amateur hour.

Not funny, just pathetic. For me, for him and his customers.

Windows Server Update Services buckle under Microsoft's metadata mountain

"One of its cloud-based options"... to replace... the on-site update-caching server....

Yeah, we'll get right on that.

Tech support chap told angry customer to think inside the box – and solved the problem

Re: "when many phones still had removable batteries"

I've had the XCover 5, 6 and 7 (Pro model, each time).

They're all still working, and all have removeable batteries, and I've passed them onto family.

The 7 only got rid of the headphone socket.

Best phone ever.

An unnamed US county – perhaps in Ohio – paid $1M extortion demand to cybercriminals

Blockchain mixers exist and basically are money-laundering facilities.

10 Bitcoin go in, via ten thousand individual transactions.

10 Bitcoin come out, via tens of thousands individual transactions to a dozen addresses, while being mixed in with COUNTLESS MILLIONS of other such transactions, with no correlation between the in-wallet and the out-wallet.

Now the person with the out-wallets is free to spend them on what they want, when they want, how they want, and they can't be traced back to the original funds. That pizza I bought officer? Well, clearly that was done with the proceeds of my Bitcoin gambling from the 0.01BTC that I put online five years ago. Can I prove it, you ask? No. And neither can you prove it wasn't.

At best you can trace it back to... a bitcoiner mixer account. Whatever the last one they used was. Which might be one of dozens or hundreds such mixing transactions.

Sorry, but it's just not that simple to trace this stuff, even if avoiding detection isn't just as simple as pressing a button.

Had this discussion many times, with managers, cybersecurity consultants, digital forensics people, etc.

Paying the ransom is indistinguishable from money-laundering / embezzlement. (How do we know you're not just paying yourself that untraceable money? Or a friend?)

And they've already committed a crime, so why would you ever trust them not to commit another just because they say so.

This is literally in the realm of "Granny gets done by scammer, then gives scammer MORE money because they promised they'd fix it if she did."

This shouldn't ever be the official position of ANY govermental organisation or even company.

Microsoft warns customers AI will mean busier Patch Tuesdays

Re: Double Entendre

If their AI was any good, the status quo would be maintained because their AI could fix the problems that other people's AI are discovering.

The fact that they have to tell you its going to get worse? It means that the AI can't fix things. It just knows how to brute-force everything, like any other dumb automaton.

Outlook for Mac bug makes font choice a purely decorative feature

Unit tests

Like Microsoft unit-tests anything any more.

Windows 95 detected installers by looking for magic words and hoping for the best

Regardless of whether it worked or not, it's indicative of a culture and attitude towards a major multinational operating system that just shouldn't be there.

It's a bodge you might use on backend testing, or in alpha builds, but it shouldn't ever have made its way into the OS proper.

This is what I find when I look at other's code all the time. I'm not interested in whether a particular bug exists, but the class of bug. I'm not interested in whether there are bodges, but what kind of bodges and where and why are they there and how long do they stay there.

I can understand, for example, that Wine and its various spin-offs have a lot of whitelists, and special-cases, and even things like detecting an exact executable to apply fixes, etc. until the code is written that means those workarounds are no longer required.

But when I look at people's code and see that kind of nonsense everywhere for no real reason it tells me far more about the kind of person who's writing that code, which steers my decision as to whether to use that code.

And everything I know about using and managing Windows tells me - without ever having looked - that the code must be absolute shite. So much stuff conflicting, so much lack of integration or a proper design, everything a bodge to a bodge to a bodge, all the way down, going back 40 years. They can't even get the UI/UX consistent, even with half a dozen complete "rewrites" forcing billions of people to learn an entirely new interface... there's still a ton of legacy sitting around AND the new UI/UX lacks a lot of things that you used to be able to do easily but no longer can. Then you look at the problems that the Samba, Wine, etc. projects have not to mention the file formats in question in the office suites, etc. It's an utter mess and nobody is ever going back to fix that.

And, to me, that's like picking up my brand new shiny car and then giving it immediately to the dodgy garage whose own door is held on with duct tape and hope, where any one mechanic doesn't want to touch anything the others have done and just wants to start over, but have to weld over their bodges instead, and where they put a large wooden stick in place of your steering wheel one day and say "Hey, we're working on it, it'll get there, but look, you can still use it..." That's what using MS software feels like to me. Linux seems, instead, to be a collection of elderly retired amateurs who spent their entire lives working on engines for the joy of it, and now operate as a bunch of enthusiasts who are quite willing to completely strip down your engine because they feel an extra dab of oil here or a wider seal there will make a difference.

Pretty much nothing today feels to me like it's been created by a bunch of professionals with a very specific and important job to fulfill who take it seriously, but Linux is far, far, far closer than I can ever achieve without, say, employing IBM's services or whatever. And with AI we seem to be letting any kid who's watched a YouTube tutorial work in the garage, and leave the work half-finished with your car in pieces on the side, saying "Your problem mate..."

New tool gives CLIs a warm and GUI feeling instead

All I see is:

"Our CLI interface is so unutterably awful in design that you have to ask a machine to tell you how to use it."

Lessons from the VMwars – nothing virtual about the Broadcom vs Tesco slugfest

Re: Linux Accounting software

You know what? If I were a millionaire starting a company, that's what I'd be choosing for managing the finances.

The whole "put your finances in the cloud running from a general purpose OS" shite is a prime area for compromise, and you'd be able to just limit that to a remote text-only session as a limited user very easily.

NASA says it will isolate volunteers from the outside world for a year

I'd jump at the chance.

Unfortunately it's mostly Americans that I want to get away from most at the moment.

Because we seem entirely incapable of having any kind of news, tech or social media website without constant mention of them and their nonsense.

End of era as the BBC switches off Radio 4 Long Wave service

Re: Irreplacable...

Or you could abandon the concept of it having to be "live" and just have it repeat a digital signal with enough redundancy that you can lose 99% of the signal and still pick up what it was saying perfectly even many, many minutes later. Much like how satellite downloads worked decades ago.

Honestly, things like DAB should just be buffered audio streams over highly-encoded channels and abandon any pretext of it being instantaneous and live. We got 90% of the way there with standardised tech and then specialised it too far down and nobody wanted to buy a DAB radio that was going to be obsolete in a year.

To be honest, they should just all be turned off, a massive national data-only broadcast network installed on every available frequency in concert with each other (much like Wifi multi-band sharing), and then we just transmit anything and everything over them, not just audio.

Why DAB and DVB wasn't just an entire digitalisation and democratisation of all the available airwaves, I don't understand.

Everyone just installs an standardised set of antennae. What you plug into it determines what you receive from it. Use every frequency for every purpose, so tthat everyone can receive pretty much everything, and then counter the problems with error correction and coding theory. TV, audio, data, weather images, who cares? Why are we still sectioning off the spectrum and requiring individual reception of each?

We have more than enough electronic power to do this in a small cheap commercial device nowadays, everything from SDRs to all kinds of processor oomph and wifi-like signal analysis to just pluck whatever signals we could get out of the air on whatveer frequencies are local to us and aren't bouncing off that hill. I mean, that's effectively how ADSL works too, it just finds the noisy bands out of hundreds of "buckets" of interfering frequencies on the copper and avoids using them.

Should be one huge datastream, broadcast over every available frequency as a bunch of "channels", and the more channels you can receive, the greater you reception for all the data being sent over them all. (obviously you would want to be able to receive a bunch of high frequencies to make the best of it)

We already do that with basic wifi (on three different frequency bands with dozens of individual channels all being used simultaneously for the same datastreams), we could extend that to AM, FM, LW, etc. radio-frequencies no problem.

When they made a big fuss about "opening up" the TV broadcast frequencies, I was hoping for something a bit more imaginative than "let;s just stuff 10 highly-compressed digital video streams down the space each analogue channel used to occupy".

Do it right and your phone picks up the "TV" and radio off the airwaves just as simply as it picks up 5G or Wifi or Bluetooth, wherever you are.

Flatpak-NG sounds like bad news for systemd refuseniks

Because, like someone interested in computers that's ageing, I was happy to take the leap and jump into "modern" Linux despite my reservations.

If this is "the way to do things", i'm happy to give it a fair shot.

And I did. And the sticking issues... were all the stuff that's supposed to be "new" and "better".

(P.S. I ran Slackware for a decade as a primary desktop, back in the Slackware 9.0-11.0 era... I'm not scared of some manual labour and a bit of a learning curve...)

I swear that Linux distributions are just setting out to destroy the whole point of Linux now.

I hate systemd with a passion. It's ALWAYS the cause of my problems. And I don't have complex needs.

I hate flakpak with a passion. And snap. I hate anything that's not a basic, old-fashioned, dependency-handling package manager going through an authorised channel. Snap - it's just whatever junk someone's thrown in there. The "steam" snap package doesn't even work on Ubuntu! It loads but no games run whatsoever. What kind of impression is that for the DEFAULT STORE on Ubuntu? Literally from a fresh install of the latest Ubuntu, I got five snap packages in before I gave up, deleted them all, redid them all as apt packages. Because they DIDN'T WORK for basic big-name open-source stuff, or were years out of date.

Then you have the myriad "let's bundle everything because we can't handle dependencies and we'd like random third-parties to be able to just bundle whatever version of whatever they like in with our software and never keep it up to date" - AppImage, Docker, etc. etc. etc.

Now we have Ubuntu shovelling (yes) AI into itself.

Seriously, I don't know what's going on any more, it's all so much horseshit.

UK Treasury hunts CTO on salary that may not compute for top tech talent

"the CTO needs to be "a trusted technical adviser to enable informed decisions" both inside HMT and across other Whitehall departments."

"This being 2026, the job ad mentions AI as one of the technologies the role is expected to champion."

Well... that's me out then, because my informed decision would be to keep AI as far away from Treasury and Whitehall data as possible.

KPMG's AI report becomes an accidental demo of AI hallucinations

Re: Lazy report writers

That's like asking a murderer nicely not to kill.

It's horseshit to think that anything in a prompt is regarded as some kind of orderto be obeyed.

Elon Musk is now worth more than $1,000,000,000,000

They're all rich now, so they don't need to work for a fascist twat any more, so they'll leave.

Malware scare keeps schoolkids home for a second day

Re: School can't school?

Tell the government, not me.

There are even official government QUANGO's that you use to sync MIS data and thousands of other services (e.g. Wonde).

If the Internet goes off, pretty much no current school in the UK is going to be able to operate without significant changes made extremely quickly, and legislative laxity on the official electronic submissions, etc. that they're required to do.

Not just a school here or there. Every single one of them in the country.

The time to argument this was 1995, not now.

Re: School can't school?

Yep.

And if you want to submit that record electronically to the DfE in a timely manner, as required... you'd better hope it's not longer than one day.

And when all your lesson plans for the year pivot around access to technology, and you're already squeezing everything you can into the school year to meet curriculum requirements, you can do a day's lessons. Sure. But then... you better get back on track quick, or rewrite your entire term's curriculum plans overnight for every single subject.

60 years ago, schools were basically free-for-alls on what they taught and where kids were. That's not been the case since the 90's at least.

Re: School can't school?

Heads-up, but they no longer rely on chalk and slates either.

They are reliant on computers. Registration is done on a computer to a cloud-based MIS. Registration is required BY LAW to operate the school.

Site access is controlled. Cards, readers, all feeding into attendance (above) and firelists. You cannot operate a school without something you can use as a firelist.

Timetabling is computerised, because it's a mathematically difficult problem and changes are frequent and have many knock-on effects (this music lesson has to move because the theatre is being used, which means this random group of kids from all years has to shift their timetables and be taken out of different lessons, which means that registration has to reflect that, and the firelists have to know where to find them if they're not present, and now you have to make sure that you're reporting the right absence / presence to the local council, DfE, etc. etc. for that day too...)

Schools are inherently reliant on computer systems and have been for the 25+ years I've worked in them. More so now than ever.

Malware compromise? Whoops, maybe you need to lock every student and teacher's Microsoft accounts where they have all their files in OneDrive, where all the policies and department sites are in Sharepoint, which is the SSO for their MIS and other services (like child-protection reporting, etc.), which has all their email, which has all their Teams groups in which they teach lessons directly to the students, who all have Microsoft accounts and devices....

Schools cannot "go back to pen and paper" for more than about a day or so. It's just not feasible with modern requirements, processes (e.g. child protection reporting), you can't just shove the kids in front of any machine (legal duty to have filters, keyword monitoring for extremism, etc. etc.), and almost everything they are taught, or record themselves, is in the cloud. 90% of the things they do require web services of some kind or other, whether that's a 3rd-party site for teaching, or an internal service for submitting and marking homework via the school portal, etc. etc. etc. Even exams are electronic now, with the plan to make them ALL electronic in just a few years.

FYI, government standards now basically require multiple 10Gbit leased lines, and Gigabit to the desktop minimum, with plans for 10Gbit / 40Gbit as suggested minimums for an average secondary school.

Sorry, but it's not the 70's any more.

Blockbuster new Raspberry Pi project turns any screen into old-school VCR

Many a Christmas morning was spent unwrapping 3 x 180min "good" VHS tapes from my parents, for both me and my brother.

When we got a video recorder in our own rooms, it was like "Wow!"

Canonical sends Ubuntu into the AI agent era

Re: We've had speech-to-text for decades

Agentic AI means "plugging your AI in to other things with permission enough to do dumb shite".

AI searching your files, I mean... fine... if that's what you want.

Agentic AI is then... deleting your files, moving them around, writing data into them, creating files, sending files over the network etc. etc. etc.

It's quite simply the single dumbest extension to AI ever.

"Hey, I just plugged ChatGPT into a sudo command line, what could go wrong?! Hey, I did ask it NICELY not to trash and corrupt my files, so everything will be fine!"

AI

I moved to Ubuntu at the end of last year because I'd had enough of Microsoft forcing features I absolutely, categorically, do not want... even though I'm FULLY aware of how they work, what they can do, etc.

So...erm... take that as a warning.

The good thing about Linux is, I can just change distro and move my home folder and a few config files over and carry on as before. It's really far easier than the Window->Linux migration I underwent, and that wasn't actually all that painful at all.

So... just be warned.

To me an AI is an app, and an app feature. They are not "agentic" (I'm not going to let them do stuff on my computer), they shouldn't need to "be contained" (they shouldn't be touching anything I haven't explicitly given them), and they are - in my opinion, which is the only one that matters on my machines - completely optional. For example, I've never used text-to-speech or speech-to-text, and have no intention to. My typing speed is more than fast enough, and far more accurate. I don't want anything AI changing settings or performing actions on my machine. Like any other program... they act only on my instruction and to my exact instruction, or not at all.

And with Linux... we should know better. We should know about permissioning, user consent, security, data access, network access, resource usage, optional packages, user-space, etc. etc. etc. and so to me even talking about AI in my Linux desktop is an absolute nonsense. If I want it, I'll apt-get it. If I don't, I expect everything else to just work as before. And if I do install it, I expect it to be locked down without even the POTENTIAL for "running on all the data in my home folder". I never granted that permission. It should not be doing it. SELinux should be blocking it ever doing anything like that.

Otherwise... we're just failing in the same way that Windows failed. And let me tell you how that went:

Microsoft said that Windows 10 would be the last version of Windows. And they were right. Windows 10 was my last version of Windows. Ever.

After 30+ years of always having some Windows and always having lots of Linux, my household is now entirely Linux.

Don't make me be saying the same about how my household was some Ubuntu... and now is entirely something else.

Satellite phone dreams orbit reality as direct-to-cell usage set to underwhelm

Re: Global roaming

Enough to keep an global satellite network in orbit?

Doubtful.

It'll go to Iridium / etc. pricing before it'll ever make a profit.

That an app 'Fits on a Floppy' is still a useful measure in 2026

Re: Router on a floppy

Freesco? Same.

And I used to make pacakges and write software and website tutorials for it.

Was very disappointed when it moved to 2.88Mb and the ISO images.

IIRC 2.2 kernels were too big to put on one floppy any more and it was based on 2.0.38

California may let Linux bypass age check

Re: And then, all of a sudden...

All the important stuff is headless and doesn't have a "user" anyway.

Why would you want to have to regularly identify yourself as an adult to, say, a traffic light controller? You wouldn't.

So nothing backend, datacentre, embedded, etc. could ever have nonsense like this anyway.

Case of have-to really, though, isn't it?

Because the rest of the world aren't going to accept that measure and, being open-source, it can just be coded out by even users if necessary.

You can't mandate open-source software to HAVE TO include something, because the terms of the licence means I can just pick it up, remove that, and redistribute it from (at worst) another jurisdiction.

One of those ridiculous unenforceable laws that politicians pass to make it seem like they're doing something when what they're doing is entirely impossible or vastly impractical.

But, hey, who knows? Maybe it'll be a driver towards the younger generation getting more interested in Linux.

OpenBSD 7.9 arrives, a diamond in the rough proud of every sharp edge

Literally just quoting the article, my friend:

"It uses FFS2, an improved version of the original Berkeley Fast File System developed by Kirk McKusick. This used to include a performance enhancement called soft updates (McKusick's own explanation) but these were removed in 2023. That means that turning off a running machine without shutting it down could cause disk corruption"

I can't take an OS seriously that doesn't have a filesystem that doesn't just corrupt itself on shutdown.

Especially not when their new bright idea is, just before your battery dies, it wakes up your laptop from suspend, writes to disk, and then forces an hibernate and shutdown.

If you make ONE miscalculation in that... you're going to corrupt that filesystem, aren't you? "Oh, the battery only has 5% left, let's wake it up in the laptop bag on the train, and while we're there we'll try to ... oh... the battery has died... because that 5% was nowhere near accurate, especially not when we just put a huge load on it... ah, but that's okay, the filesystem will... oh..."

Years after UK Post Office scandal broke, Accenture and OneView Commerce bag contract to replace Horizon

Re: Now why would you kill the internal project ?

Either you, personally, can be responsible for the next Horizon scandal, or you can blame a famous company that you paid hundreds of millions of pounds to do that job.

It's nothing to do with civil service, that's a no-brainer. There's a reason that IT are outsourcing the stuff that they don't want to take responsibility for (cloud storage, cloud backups, cloud email etc. etc.) - it's because given the choice I'd really rather pay someone "who knows better" to do all that and take the blame when they fail catastrophically.

I'm quite capable of running an email server. I do it for my personal server still. But if you want me to have corporate responsibility for it after you turned it into a core business requirement, then it's DUMB for a company to be reliant on me doing that, and it's dumb for me to just assume that responsibility alongside a thousand others. We pay a contractor, supplier, reseller, etc. and they do it. Problem solved. Of course it costs more.

But then my job isn't at risk because of an off-by-one error in something I wrote exposing the personal details of every subpostmaster and landing me on the front page of the news, brought up before government inquiries, etc.

I honestly cannot blame them for not "rolling their own". That's a no-brainer for outsourcing a highly-specialised and critical tasks. The question really is: Have they learned their lesson and put everything in the contract for penalty clauses, strict requirements, their own checking and compliance measures, even oversight to manage the problems and detect issues, etc. etc. etc. to ensure we don't have another Horizon?

Plex appeal fades as Lifetime Pass jumps to $750

I have a lifetime subscription, but would never have paid that price.

Even with a paid perpetual subscription already, they need to fix some things to stop me just disappearing.

Their media detectors are worthless and often don't detect a bunch of content. My usual trick is to copy an IMDB or TVDB series number and paste it into the search because that's the only way to get what you want even when you type in the full name and year of a piece of media.

They have a habit to overwrite my chosen imagery (e.g. clear, easy-to-read, easy-to-recognise old DVD covers with whatever junk reworked art the Netflix version has now, etc.)

They need to seriously fix their backend video processors... they glitch out on all kinds of perfectly-fine media files that VLC / FFMPEG just laugh at, meaning your show just STOPS and then it hangs up the processes and takes minutes to resolve. I mean, if nothing else, that's a DoS or security problem right there.

The app sucks and they removed audio playing from it... I mean... why? Why would you do that? They're talking about putting it back in but it's been over a year since you've been able to play an audio file via the app.

At this point, I'm already on the verge of ditching an already-fully-paid-for service, that runs locally, on my own machines, because of the problems it has.

Trebling the price, and thus peeing off your customers, losing future customers, and therefore indirectly making the service worse? Yeah, that's not going to help me stay at all. Maybe that's the plan.

But if the lifetime subscribers start to leave... Plex is dead.

The class of 2026 has heard enough about AI, thanks

WE DON'T WANT IT.

What is it with companies trying to foist shite on their users.

WE DON'T WANT IT.

Fine for those people who do, they can purchase it from you. But the rest of us don't want it.

Just like MS with their constant UI changes, and all the other nonsense.

"We've listened to our customers and then decided feck it, they don't matter, why would we want to listen to the likes of them, they're only paying our salaries..."

Windows boot partition runs out of space for Microsoft's May security update

Re: Time to give Windows

Ah, the days when a complete game was stored on tape at a ridiculously low bitrate, giving you maybe 40Kb for an entire game.

Re: Been here before…

Microsoft told me that Windows 10 was going to be the last version of Windows.

Personally speaking, how right they were...

UK begins antitrust inquiry into Microsoft's business software ecosystem

Word

Guarantee you that the final report is in Word format because that's the only format that they have the software to make.

Linux kernel maintainers pitch emergency killswitch after CopyFail and Dirty Frag chaos

Re: SELinux 'strict' ruleset provides this protection now

The problem being that almost every problem I encounter on Linux when doing basic things ties back to either a) systemd or b) SELinux rules.

And I'll be honest... I just don't have the time to bother to learn their stupid configuration / syntax / etc. in order to fix them.

Re: I find not building modules I don't need to be pretty effective.

Quite.

But the days of "make allmodconfig" are long gone and I haven't compiled my own kernel in years now.

It would be nice, however, to have systems configured so that module loading of new modules required explicit and secondary elevation (rather than just using sudo for install and it automatically loading lots of modules along the way).