spork_css: Yep. That's exactly what I did.

On the up-side, I've never had the ports' numbers change or anything. So, the NIC's are consistent over reboot. So, as long as you know what is what, it works fine.

Just kind of a tiny thorn in the side.

I dunno why Dell does that on some of their BCM cards - there are actually multiple PCI devices per card that handle two ports each, but the ordering on the bus is backwards (chip B has ports 0, 1 and chip A has ports 2 and 3)

there's some SMBIOS magic that Linux uses to derive physical port numbers

this server used to have vmware on it and I think it too had weird numbering as I was always having to test if I had something in the right port...

what's the newest stable version of python recommended to install from pkg?

seems weird "python" installs an eol version 3.9

looks like ungoogled-chromium isn't in pkgs anymore :(

iridium-browser seems to be next best privacy browser option but doesn't seem all that great

oh wtf not even iridium is available?

i'm on an updated 14.4, why doesn't pkg search iridium or pkg search iridium-browser return anything?

Anyone having issues packaging rust? Seems it always being killed at the packing phase.

i think its swap/memory problem

that's a problem for me

Yeah, it is. I guess 48G isn't enough or maybe just need to not use tmpfs?

Lost a stick last week due to issues with it so maybe it will just go away once I get that back later this week.

i have 32GB of ram on laptop and i have tmpfs = all in poudriere. and it never yet to fail

Rust requires a lot of memory to build, so its usually worth building it by itself or with other heavy builds like llvm.

angry_vincent: it can fail for sure, even with 64GB of RAM. It depends what else you're building it with.

angry_vincent: In the last couple days?

robert12: I was thinking that too but I just reran it was the only thing at the time.

I had the same problem (well past 2 days ago), so now I have a separate port list for heavy ports like llvm, rust, etc, and then another list with everything else. I called pouderiere bulk with the first list, then the second.

Unless if I have another bad stick.

But I can watch vmstat and see out of swap in /var/log/messages.

I don't think so. Compiling llvm, rust, and ordinary ports at the same time is resource intensive.

Maybe I should just bump swap up a little

I have more ram than you by x2 and encounter the same problem, but sure, go for it

While only building rust?

It's not just rust. It can also happen with llvm, esp if you build multiple versions at the same time.

important note - i restrict poudriere to build 1 port at time.

Oh. That changes things quite a bit.

i found it is as fast as having multiple jobs ( building several ports at time )

For me, that was the solution

skered: yes, the last update to rust from some days ago just built fine.

I originally mentioned in #elixir as that was my primary use case, but I later added tracking FreeBSD systems which peopel here might find useful. To add a machine, define its version/arch/pkg branch so 15.1 amd64 latest and then it monitors your pkgs for updates/advistories. Alerts you if any of the pkgs you have installed have an advisory, or if there are updates. Free as in beer, if anyone

wants. Granted there is less value overall if you don't also use elixir, I could add other languages to track. hexorcist.net

advisories*

If there is a better way to do this in freebsd please let me know... For years I subscribed to the rss feed but often forget if I am using that one pkg of several hungded on each deployed machine

hundred* typos, need to clean this keyboard.

Today I learned that 10 million log entries (mostly nginx) takes up about 450MB within VictoriaLogs - bsd.network/web/@dvl/116799701807951970

That seems reasonable.

Assuming 10m logs / week is, reasonable

skered you lost a stick of ram due to packaging rust?

this prolly doesn't help but i install rust by running the curl ...sh line on the rust install page. i don't use pkg's for that

i'm on an updated 14.4, why doesn't pkg search iridium or pkg search iridium-browser return anything?

kerneldove: I don't know what causes the stick to go bad but memtest86 doesn't like it and I was having random kernel panics.

But I'm sure it wasn't poudriere.

great time to lose sticks of ram. enjoy paying 5x last year's price ffs

OEM replaced it for free.

nice!

It will be here this weekend.

kerneldove: However, I think I can no longer build rust due the missing 16G. But I think I've built it at least once or twice since I moved down to 48G.

So I don't know if this something new with the port or something else (moving to 15.1).

Before I was 14.4.

i'm on an updated 14.4, why doesn't pkg search iridium or pkg search iridium-browser return anything?

kerneldove: for some reason it is explicitly not built on the official builds.

is google paying someone off for that? like why not? ungoogled-chromium stopped getting built into pkgs, now iridium? doesn't add up

it's not a conspiracy, we disable some stuff because it uses too many resources or is unbuildable due to resource scheduling

yeah both were disabled for the reason, they became too large to extract and build

sure

kerneldove: about your python question earlier, the version of the python port is of no consequence, it is a metaport that depends on the default version of python and provides the python symlink, but as it contains PORTVERSION=${PYTHON_DEFAULT} the version should be the same as the default version of python, so currently 3.11, where did you get that 3.9 from? changing to a higher default version

of python is a bit of a problem currently, but hopefully we will be on 3.12 as default soon, see also bug 285957

nimaje freshports.org/lang/python says runtime dep python3.9 : lang/python39 does that not mean it installs 3.9?

nimaje: lang/python is special.

kerneldove: ^ I mean

how so?

kerneldove: FreshPorts says it's 3.9 but that reflects the default version of python when that port was last updated. It's a rabbit hold to get it right.

Also open to any tips or advice for new homeowners that you wish you knew when you started.

ok well i just install python3.11 pkg and create my own symlink from /usr/local/bin/python3 to /usr/local/bin/python3.11

^ bad paste, sorry

grats on new house

kerneldove: Yes, you can, but I prefer to install the helper port

[15:30 mydev dvl ~] % pkg info -x lang/python

python-3.12_3,2

python3-3_4

python312-3.12.13_2

^ That's what I have. What you get should be conistent.

what do you have for pip?

kerneldove: Nothing, I didn't install it. but:

[15:31 mydev dvl ~] % pkg search pip

py312-pip-23.3.2_4 Tool for installing and managing Python packages

hm ok so if i install the "python" and "pip" ports, i'll get 3.12 and the symlink?

^ keep in mind, I'm using my own personal pkg repo. Created with poudriere.

kerneldove: freshports has a bit of a problem with dynamic data in ports, like that PORTVERSION=${PYTHON_DEFAULT} I mentioned earlier, as the page gets generated when freshports sees a change for that port, but dynamic data in ports means, that it can change due to changes elsewhere

kerneldove: Try installing it, find out.

oh

ok i'll try good idea

no pkg named pip in pkg repo?

usually pyXXX-pip (py311-pip)

if there is a "python" why not a "pip" to match it?

I don't work here :-)

All the python pkgs are prefixed with the version they belong to

so python installs python 3.12, but then i install py311-pip?

python should install python311 for you, as that is currently the default version

ok ty, but meh i'll just stick with my method because then i can explicitly match the python and pip versions installed

sorry i d/

c

kerneldove: You can use the builtin venv?

When does the default python version get changeD?

skered venv?

kerneldove: python312 -m venv myvenv; . myvenv/bin/activiate;

You can use pip from there. Just activate before

meh, ty but what i have is simpler and already works

mns: not often

I'm running FBSD 15.1. I created a host Bhyve server and using Sylve. I would like a way to use sparse file system for virtual machines so they use less storage than initially assigned.

and what do you hope to get from that? having space for more VMs? what should happen when all your VMs require the assigned space?

mary751: i know vm-bhyve will create them sparse, but i don't know how sylve does it

nimaje: the plus is obvious no? you save disk space you aren't using.

but how is it saving space? you can't really use it for anything else, as you could end up with not enought space for the VMs

You could create a 500GB sparse image with the intention of maybe using all of it, or maybe not. It's the flexible option. Might use it. Might not. If not, it's better spent on other things.

robert12: they didn't ask to be convinced to not

rtprio: what?

also, "what is the right size" is a tricky discovery, depending on how much of a pain it is to grow the volume

yeah, that's why I asked "what should happen when all your VMs require the assigned space?"

I'm not convincing anyone of anything. I'm replying to nimaje. Sparse images are useful.

then it uses all the assigned space. but until then it doesn't. they didn't say it was overprovisioned

A sparse image lets you grow as demands require it, you can grow disk space they live on too, but consuming the full 500GB is not desirable (for me anyway), it just gives you room to be flexible.

thanks guys - it seem I need to look into sylve --- i have other server on which virual machines occupy only as much as it is used --- but this new one

is there a jail equivalent of vm-bhyve? just a real basic minimal overlay

if it isn't overprovisioned, then why does it matter if the file system is sparse or not? and if it is overprovisioned, then you obviously have to think about failure conditions

That's their preference, and its a valid preference

i do it, most of mine are 100gb. do thet usually exceed 50? no. is it handy to not have to do anything when they do exceed 50? yes

exactly, there are *many* reasons why you might do that, and launching an inquisution into that is not helpful.

and I don't see how it would be helpful to answer that question, without making sure possible failure conditions of the use case where throught about, but as I will not list all failure conditions for all the possible use cases, I first have to know the use case and maybe the answer will be not to use any sparse file systems, like when you aren't even try to fit more data on your storage then all

VMs using 100% of their assigned storage would need

I'm trying to setup Postgres as a jail with Bastille. I've been trying to "jail" the dataset, works as expected until I try "unjailing" the dataset and trying to see the data/files from the host afterwards? should I not be jailing the dataset?

You seem to be of the mindset that their approach is fundamentally wrong, but it's not, it can be valid for a wide range of reasons, and usually an answer that amounts to "youre doing it wrong" is not helpful.

The only reason I'm unjailing it is to reassure that the data is still in that dataset after something went wrong.. but I'm failing to be reassured since I can't seem to see the data after all that.

im running a bhyve vm on a server and i vnc to the vm. works great. any way to make the text size bigger tho?

if I thought it was fundamentally wrong, I would have explained why it is wrong, not asking for their reason and about some failure condition they hopefully already throught when comming up with that approach

ah looks like alacritty config does it

you only use freebsd as server? not as desktop?

it's also possible to use on the desktop, and many people do.

I intend to use as desktop

I am thinking about guix vs freebsd

i dunno guix. i know freebsd and i highly recommend it.

Also using FreeBSD, both backend and desktop

Stick to supported hardware

heston76 my hardware is normal lenovo laptop

amd64 and all

jan0t: this might be a good read : wiki.freebsd.org/Laptops

jan0t i tried guix. if you have minimal hardware i think you'll find it's too resource hungry

freebsd is lighter on hw usage

i like guix ideas tho

If you have a Lenovo Thinkpad you're in a good place for both FreeBSD and OpenBSD. I would consider OpenBSD seriously. It works *very well* as a desktop computer. The only requirement is matching hardware, and you have it.

Anyone know if Intel Graphics (ADL-N) - Alder Lake-N Vendor 0x8086 device 0x46d3 is supported by i915kms in a recent-ish drm-kmod?

robert12: They don't do SMAPI do they?

I'm not sure what SMAPI is

robert12: In this case, it lets you say things like "don't charge this battery past foo%"

robert12: Helps your batteries live longer.

Ah I see. I'm not sure is the answer :) My laptop stays connected to a power source always, for better or worse. I can vouch for its ability to wake from a suspend though.

robert12: That's the use case right there - if you're plugged in all the time, without the ability to cap charge, your battery is topped off all the time, and will die fairly quickly.

Indeed

ive got a bridge and tap setup for a vm on my machine, works as expected i can ssh into the vm from other place on the lan BUT i cant ssh into it from the desktop running it

running bhyve

you have a firewall set up?

give some more details. networking etc