anyone know how to set and use a var in a ucl config file pls?

anyone know how to set and use a var in a ucl config file pls?

just rainy summer here time to prepare the christmas tree

so, I have 2 datasets on a zfs pool. one of which is shared via smb. now I want some data from set A to be visible in set B via smb.

what I tried was to just symlink it, but that doesn't show up on smb

so far I had to copy the data from A to B

yourfate: Can you use MOUNT_NULLFS(8)?

for individual files?

I have used that for folders

oh for individual files it wont work...

but ye, for folders that is a good idea, thanks

I have acutally used it before but forgot :s

for folders it worked :D

no worries :) But yeah sorry, I dont know about individual files

one other think: can I somehow see in `ls` that its a mount?

like how I can see symlinks?

I don't think you can. You can see your mounts when running `mount`

right

anyone know how to set and use a var in a ucl config file pls?

i have a vm-bhyve disk of type "nvme" on its own partition (disk0_dev="custom", disk0_name="/dev/foo"). this isn't directly mountable like a normal device – are there some extra hoops i can jump through to mount it?

had a catastrophic hardware loss so i'm trying to access this vms data from a different machine and booting it doesn't seem like an option, since it just BSODs and reboots in circles…

RIP mergemaster in 15-current, long live etcupdate

now etcupdate just needs to be extended to easily update thin jails

how can we set up a pi4 8gb for cross-compiling?

hi. I have FreeBSD 14.1-RELEASE installed (upgrade from14.0). I am using packages and I like to build drm-61-kmod or rebuild drm-515-kmod. How should I do? I try tou rebuild drm-515 with make and I got an error dma related

nsoci: What's the error?

make[1]: stopped in /usr/ports/graphics/drm-515-kmod/work/drm-kmod-drm_v5.15.160_0/dmabuf

--- dma-fence.o ---

1 error generated.

*** [dma-fence.o] Error code 1

Can you paste all the relevant lines to a pastebin?

make[1]: stopped in /usr/ports/graphics/drm-515-kmod/work/drm-kmod-drm_v5.15.160_0/dmabuf

--- dma-buf.o ---

1 error generated.

*** [dma-buf.o] Error code 1

I am sorry. I need to go somewhere now. I will prepara later everything. Thank you.

had a question regarding freebsd releases

for a mission-critical server (think web server for a large organization), would I use the STABLE branch or RELENG?

right now we are on 14.1 RELENG but I see that 14 STABLE has a much longer EoL date of 2028

but according to the docs, STABLE is a development branch not suited for production uses

and there is 14 STABLE and 14.1 STABLE but I cannot find information about EoL dates for 14.1 nor the ISO for 14 STABLE

should we stick to RELENG and update every few months? how do we get the 5 years FreeBSD support schedule to our org?

STABLE isn't a version it's a branch that release versions are cut from. It has a long EOL because it is and will be every version of 14.x

isley: so we install 14.0 RELEASE, then update to 14.1 when that comes out, 14.2, so on and so forth for the 5 years?

since there is no such version as 14 STABLE by itself?

How is wireguard on FreeBSD these days? I was thinking of setting up a jail with it so I can connect to home from other places.

I always envision putting wiregaurd or tailscale/headscale in a jail, and that would let me and only me be able to ssh in remotely... but then I don't see how that would help me get into the host

beowuff: it's wonderful now on FreeBSD 13+

isley: am I on the right track with the release schedules?

Yay! I was also talking to someone in #wireguard, and they also said it's pretty good now. Can't wait to try it!

beowuff: yup

@scoobybejesus I just logged into my host, did a bastille console command into a jail, then ssh'd back into the host. Worked fine. Probably because of how I have the network setup so each jail gets it's own IP.

interesting. I think I have set skip on lo. maybe I can do the same thing if my jail has the proper private key

I would want this in a VPS where all my jails are loopback

i was thinking of doing routing through a wireguard/tailscale jail to connect a syncthing "lighthouse" (I think it goes by another name... relay server maybe)

maybe worth noting that we do support the container network thing that wg-on-linux does, where you can create a kernel wg(4) interface on the host and drop it into a jail for its sole network access

vnet jail

no epairs or any other middlemen

oh dear lord I just located a bunch of 12.1 jails

Dog help you! :-)

Better a jail than an entire system running in a rack somewhere.

when my dog sneezes, i say "dog bless you." and yeah, jails are at least a much smaller attack surface.. seems almost like a good problem to have

I was thinking that too. It's a good thing really.

Easier to deal with because I would make sure I had a backup copy of the jail and then I would upgrade the copy and test it then swap it in.

If the jail has writable data that is getting updated then I would still do it that way as a test run through and then do it again for real after I knew it worked okay.

yeah I usually just always snapshot the jail before modifying it

Writable data is always more tedious. I am upgrading databases this week and avoiding creating a "split-brain" is tedious.

by writable, do you mean base rw?

I mean like a web site which is online and so people are using it and changing data. If I fork the data and then upgrade the fork and then replace the site with the upgraded fork of it then any changes made after the fork are lost.

In order to avoid that one must make the data readonly to the world by some method. Usually by bringing a web site offline and putting up a maintenance page. Then upgrading. And then after the upgrading bringing it back online again.

For some legacy servers doing something that other people have set up that I have no idea anything about I don't want to just jump in and perform an upgrade code. I might find a problem that prevents it from being upgraded. And then I would be stuck betwixt and between.

Howdy, folks!

To avoid that I would have to try an upgrade and see if it works first. Run into any upgrade problems before I need to know.

Managed to compile my first FreeBSD program that involves the use of another library today.

That library being SDL 1.2.

Then if that works it means I have to do it again. But that's not necessarily bad. Since it means I just did it and know exactly what needs to be done.

More likely to make mistakes the first time. Less likely to make mistakes the second time.

I've yet to get VGL output to work on here, but it runs on X11 and aalib.

anyone know how to set and use a var in a ucl config file pls?

you mean like, telling the shell read the file?

rwp: we have a web proxy (nginx) on the edge of our network; when a new version of website is ready, we start it up and then just change what the proxy points to, if it works, fine, if not, roll back to the previous server. simple config setting change is all that is needed

rwp: of course you will still need a rollback strategy for your databases if needed. normal stuff I guess

Hello, all. viewing /var/run/devd.pipe is a simple way to see whcih keys produce which ACPI events, but the devd man page says it is deprecated. What would the new way do so?

aquamo4k, It's not a new version of a web site. That's not the problem.

Site A is up and running. Database A needs to be upgraded. It's a problem if we fork the database over to system B and then upgrade it there.

User's register themselves at Site A which stores data into the database A. We bring database B online. User returns and goes, "Hey! What happened to my account? It's gone!"

It's gone because when we forked the database without blocking writes to the old database then anything written to that old database is lost.

We can't just go, oh, yes, we know we will lose some data, but you just have to do it again. That's not acceptable.

So we bring things offline for the upgrade. All of that time people are getting a site offline result. That's not good. We want to make that time as short as possible.

There are of course many strategies to make this go easier all of which require more resources than most small organizations I am working with have available to them.

And so like my dad always told me, spit in one hand and wish in the other hand and see which hand fills up faster.

shit in one hand is the phrase

not spit

"Wish in one hand, shit in the other. See which one gets filled first." When a person wants the impossible.

Howdy, folks!

Managed to install some more packages on my new x86-64 FreeBSD PC.

great

Managed to get circus linux, GTKRadiant, netradiant, and TrenchBroom.

Though, i've uninstalled GTKRadiant since it really likes to SIGSEGV a lot.

Though, in terms of TB, i really don't mind it being an old version.

And yes, i also got quake-source (QuakeC source code for Q1/QW) for the heck of it.

Also for netradiant, i was expecting it to come with q3map2 and q2map. (It didn't.)

Even then, it's not like it's the end of the world.

And yes, getting circus linux was a perfect way to test the aalib support of SDL 1.2. (It works, but under VT, mouse input doesn't work.)

Managed to get a local high-score on cirucs linux. 158 pts.

And found out the "Lack of mouse support" issue on SDL 1.2's aalib backend also applies to terminal emulators.

Again, it's nothing much.

But it would be a challenge to play circus linux on a terminal.

vortexx: iirc you said you ran OpenBSD under FreeBSD, using... bhyve?

nicholaus04 o/