[Out from under a rock] Updating CORE (freebsd-update)

To su the user needs to be a member of the wheel group. That’s all. Use su - to get a complete root login environment.

For sudo to work like a “real” root login shell - FreeBSD and Linux all the same - use sudo su -

Not exactly the topic, I apologise for playing the wise guy.


Oh. I get it.

I would take the opportunity to jump to Scale, and forget about 13.3.

Oh. I did.

My last Core was 13.1-U6.1.

My killer feature was VM stability and docker capable jails.

Sounds like a new killer feature is going to be a shell in the GUI :frowning:

The point is that su asks for a password while it’s considered (at least by some including me) sufficiently low risk to allow for password less sudo in the context of e.g. a home lab.

And the - at the end of su - is important. It tells su to perform the equivalent of a full login, so you get e.g. root’s search path etc. Regardless if you use sudo or su directly.

1 Like

Very important distinction. Especially if you put a script called “ls” in your home directory that is only executable by su / wheel and which sends a email to the faculty member and every computer club admin in charge of the VAX a list of every user online at that moment, followed by an immediate system shutdown.

That’s how a classmate stopped being a system admin overnight. He was lucky he didn’t get expelled. It was a beautiful trap.


I wasn’t going to speculate on the longevity of the install, only chose to discuss how I had experienced it with a different long past customized OS and what one might need to consider for the present question.

What I meant by “update via source” is a process one can do on FreeBSD. You update /usr/src on your install (of FreeBSD tho it may exist for this too), and then you compile a fresh world and kernel (its all scripted, so fairly easy). The world is a small group of software which remains essentially constant through the life of FreeBSD which enables some basic function, I believe particularly the ability to program (write code/software). The kernel is the same concept on any operating system which is the smallest contained part of it that is a sort of limited but whole system manager for various necessary OS things.

However only Linux is technically named for its kernel. A Linux distro is the kernel plus whatever random group of things the distro devs choose to include as well as their own choice of customization of anything under the hood, and no two distros must be at all similar in the slightest beyond possibly using the same version of the kernel.

I also very often have terminal windows open on my box that are already switch user (su) to root, so its a good thing no one here at home has any kind of a clue outside of Windows/iPad… you can su to a non-root user too.
Aside from sudo which is super common in the Linux world, the OpenBSD forks have brought doas, which does basically the same thing though may be more controlled or more secure, I tend not to use either so I cannot say.

The classmate trap sounds very BOfH-like.

Makes a big difference. Linux is bloated and fickle. Systemd alone is something like 5% the size of the kernel itself. Ignores POSIX. Ignores security. …Linux makes a mountain out of a molehill for most issues at best, and Systemd has/will always sucked and makes several kernel APIs mandatory and thus never be considered high-secure and stable.

We can have a discussion about the GOODs and BADs of Linux v BSD in another thread. All linux isn’t bad, and I’ve run some fun Linux stuff a friend of mine was writing and I currently have one webserver running linux (who doesn’t). But that’s besides the point.

This thread is about the EOL type situations CORE will be in soon. If you choose to deal with that by “side grading” to Linux, then good for you. I’d rather update to base FreeBSD and tweak the system myself or let the OS go EOL and become a glorified external drive.

The shell in the UI is another sidebar (and moot if you ask me). Set up a user with a good shell, ssh key, and set root to use /bin/sh or better yet /bin/false to disable root login via ssh.

I’d take Theo (OpenBSD) over Pottering (SystemD) any day.


Even without systemd, Linux is still plagued by GRUB…

1 Like

I’ll just leave this here.


Agreed. Linux sucks. …now let’s see if we can bring back the Emacs/Vim war. Those Emacs users need a good beating now and again.


I’ve been snooping around the core-build repo.

  1. Hats off to the TrueNAS team. Looks like a lot of work and I just want to say “Thank you!”.
  2. I’m quickly realizing that a lot has changed in BSD since I’ve “been away”. I have two kids now, so I don’t get to play with servers and whatnot these days.
  3. I obviously have a lot of homework todo. I’m not looking to fork or anything, just trying to glean as many tips as I can.

Sorry, I’m scrolling up and reading/finding some of these posts I missed or didn’t address.

I’m proposing to grab the script freebsd-update and conf file from a jail or from the freebsd-src and running it from within the TrueNAS base. I’m guessing (based on the man page) it should do a fairly good job in keeping installed packages but I want to run the process on an old jail to see how the script works. Then I’m going to create a VM, install TrueNAS and try my theory to see what breaks (I can’t replicate my entire setup but I can get a feel).

Not something I’d like to do and if I were to go this far, I’d just clone the TrueNAS build repo and just update which FreeBSD branch to pull from for the build (e.g. freebsd 14) and blindly compile a “CORE-14” version. That in theory would allow me to roll back to “CORE-13” if things go horribly wrong (and, ‘IF’ I got an actual build).

Building a BSD from source was a cool thing to do at one point–and so was a custom kernel–but it’s typically only done once (because it’s not fun).

Plus, I imagine a build would lock up my mini for days!


Having a bit of memory and decent processor, or numerous cores, or speedy ones… will all help. My personal box has an AMD Phenom x6 and only up to near 3GHz, with 32GB ram, so it takes a some hours for world and I think like 45 mins for kernel, but I also believe I still remove some unneeded things from both so that may shorten the time slightly.

Hours?! The last time I built a BSD (I believe it was NetBSD—which I think has the largest number of code lines)—took me days! You must have quite the machine. …what was your IP and dog’s name again?

It doesn’t have to be:

1 Like

See!? Linux is bloated! Even a boot loader—based on a BSD version—has to use a “special” font. …come on people, remove the dependency and keep your code small. I honestly don’t need a boot loader to look fancy, after a few seconds it’s gone but that bloated code is still taking up room. #face-palm

1 Like

I think the main speed up was the switch to clang from gcc some years ago. I have it setup to use ccache which may be default these days, and the extra two cores may reduce my build time some. I think it was about 1.5 to 2 days back with v10 or 11.

45 minutes tops on a modern machine with 12+ cores.

I don’t have anything near that. I’m sporting a NAS mini 2. I’d have 4 cores at best, maybe. It is also EOL. My laptop is EOL, my phone is EOL. …phew! I’m not that old so hopefully I stick around. Apparently I’m a living version of the NetBSD logo, but I don’t need much; just LibC and Vim.

It maybe older but it can have all the tablets, tvs and phones streaming movies and whatnot so the family’s happy.

I’m a long time Core user with jails and VMs. I’ve been watching Scale and have heard all the compliments about recent releases. I’ve also planning for end of support issues such as a Windows 10 VM and no TMP support in either Core or Scale. Also, one of my Jales is end of support for Core and no plugin for Scale.

I built a test scale system on an old laptop, very simple with a small NVME boot drive and a single SATA SSD with 16-GB RAM. Setup was easy, new GUI is nice.


  • SMB transfers to and from Scale 1/2 the speed of Windows
  • The plex plugin did not match documentation or anything on Youtube

This was enough. For now I’ll keep running Core and slowly migrate from hyper converged to NAS + Virtualization on mini PC.

I’ll try Scale again in a Year