Budget is not an issue actually. This is going to be deployed for customers but i want to it to be as cheap aspossible to get them maximum value for their money. Software stack is going to be minimal. Probably alpine linux or ubuntu server. Spec wise i think even an i3 level cpu is fine. Ram 8gb, hard disk 256 gb ssd should be more than enough. Dont require any fancy wireless stuff like wifi and Bluetooth.
Your use case sounds like something a nuc or sff from minisforum could handle, but if you want “cheap” and “small enterprise,” both ambiguous terms, the supermicro superservers should fit the bill.
Not really. Just has to run till at least 5years at least. Since this will be deployed at customer site, pine64 and android both are not feasible. Thanks for the suggestion though.
I have one in the lab at office. Were abt to be thrown out. Nursed it back to life somehow. Good to play around plus company foots the electricity bill so win-win.
You can basically get rack mount level performance from:
Supermicro SuperWorkstation Tower Servers
Lenovo Thinkstation P Series Server Towers
HPE ML series Tower Servers
Dell Precision Tower Workstations
In your situation, I’d be looking at ebay, serversupply, or other used hardware resalers that offer 2 generations back hardware. Used DDR4 based systems are abundant and cheap enough, go that route.
So this will be deployed for customer sites. I dont think they will be happy with second hand stuff from ebay 😀 For my personal stuff i would have been happy with that though.
Wow, thank you for sharing this! Grumblegrumble have to reinstall my system…
This straight on the back of a thread about flatpak verification and security - a reminder that a lot of the incredible work of a distribution, especially Debian, is a community of people curating packages with care, and not just for how quick they can be made to work together.
Also a highlight for the work toward fully replicatable systems - if I understand right, the exploit here was snuck in in the binary, not in the source code.
That’s not correct as far as I can tell. The backdoored code ended up in release tarballs (but not source tarballs because of autoconf fuckery), see eg. this mailing list discussion.
Well you only have to reinstall if you had affected versions installed.
For e.g. Debian stable, thats not the case. Or e.g. Arch sshd doesnt link to xz, so thats not a concern there.
Most systems wont be affected because their sshd doesnt link xz, didnt update to that version yet or simply isnt accessible from the outside.
Though it does show how vulnerable critical packages can be and how much better we need to protect them.
I understand the need for security, but default is powerful, which is why the eu requires os vendors to provide a browser choice screens instead of letting the os vendors to pick their own default browser. Without coupling this restriction with a browser choice screen, this would guarantee to increase edge market share simply because it’s the default.
Your assuming everyone understands how default apps work. And you are forgetting that most people get intimidated by the popups trying to get you not to change the default.
It is complete enshitification if you ask me. There should be no reason why a application can't have a button to make it the default.
It's possible that I've been lucky, and my experience of end users when it comes to Windows lately have been using it since Windows 10 which strongly pushed the Default Apps configuration vs previous versions. The world is a big place after all
Maybe I'm just over reacting. But people like us deal directly or indirectly with the fallout of such madness. I hope that the DOJ goes after Microsoft at some point soon or at least scares them a little.
Sysadmin
Oldest