Did anybody have any success running FreeBSD on a MacBook Pro 2015 ?

I got a newer latptop for my better half, so this one is lying around. It is a solid piece of hardware, still runs pretty nicely, 8 years in, and going strong

I have no need for OSX, Linux runs OK on it

asmc(4) is the only "secret" sauce I know of, but I don't know if that works specifically with that revision.

thanks debdrup, I didn't know that!

Running on Macs can be a bit hit-and-miss. Depending what's inside, for example if it's got a Broadcom wifi chip then you'll need a USB dongle for wifi.

good point, I need to check what wifi chip is in it. I think it should be Intel..

it has dual GPU, radeon and intel

is hard to disable a power-hungry descrete card?

argh, Broadcom BCM43602

hw.pci.do_power_nodriver=3 should ensure that devices that don't have a driver attached will be powered off.

You can also do this manually via devctl.

Well, maybe devctl can do it. I don't remember.

Title: bwn(4)

That OID goes in /boot/loader.conf by the way.

this suggests that there is hope for broadcom wifi

That's the softmac, there's a fullmac too, iirc.

Title: Landon Fuller: FreeBSD Broadcom Wi-Fi Improvements

That has a full list of what's supported

the article suggests that I might have more luck running Open/Net BSD

Finding a used ThinkPad was my solution, and I haven't regretted it.

hello anyone ? I have some problem to configure snort... my problem concern the "reputation.blacklist" and "reputation.whitelist". Simply snort cant start because cannot find the files.. I tried many differnt conf but nothing was changed

drobban: tried to deploy to; didn't use flyctl on freebsd

mystic: you installed the port and now are trying to configure it?

rtprio: no, I Installed the package because my pc is very old and building a package is to hard. I read an howto and I configured the /usr/local/etc/snort/snort.lua file. I have the following lines ---> "blacklist = BLACK_LIST_PATH .. '/ip-blacklist' - The same (but changed) is for whitelist. The file snort_defaults.lua has the following variables: WHITE_LIST_PATH = '../../intel' ---- and the same for

BLACK_LIST_PATH

the ip-blacklist file is under /usr/local/etc/snort/intel/ directory

but as I said snort cannot find the file..

how does freebsd do regarding power saving features for laptops?

meena: tbh the guidance on swap is a bit cloudy these days. 128GB RAM is p. common. Saying all that, i have a server in a datacentre with 4GB swap, 512GB ram. later had to add a 32GB swapfile. it only has one swap as a partition.

hi, this is the sixth time i've tried to install this OS on my laptop and every single time, i can't seem to boot it. install goes fine, i remove the usb stick, dead in the water. is the bootloader broken/not installing? i've looked at several threads in the forum but the outlined solutions have yet to work for me. i'd appreciate some help

deadguy: is your laptop disk nvme or m.2 ssd ?

also which installer

f451: no. this is a dell laptop from 2012 with an msata disk. tried both bootonly and dvd1

which version

13.1

snapshot or release?

release

also in the dell bios, is there anything in there like 'secure boot' or anything specific to windows

cos if thats not turned off then booting wont work on some systems

there's a secure boot option but it's disabled. this machine hasn't seen windows in a while, used to run one flavor of linux or another in it

i'd try a 13-stable snapshot in that case, lor 14-current

apart from that, if it doesn't work then freebsd mailing lists

on reboot, whats the error?

i suppose i could. i think i've tried every combination of zfs, ufs, gpt, and mbr i could think of. don't think this machine is new enough for uefi

it's "operation system not found"

so its not writing the boot partition possibly

try latest -stable and/or -current

i boot back into the installer and try gpart show. and sure enough there's a freebsd-boot partition. i've tried a number of commands from the forums but still no luck

sounds like bios or installer prob to me, other than that i dunno

def an installer problem. i guess i'lll try current

is the boot part set bootable, active

it's marked as bootme

huh

mailing list may help id -current doesnt

i have a little bit of time, i'll try current

good luck

im trying to make ungoogled-chromium my default browser but the "set as" button in settings doesn't work. xdg-settings get default-web-browser returns ungoogled-chromium.desktop even tho ungoogled-software/ungoogled-chromium #1929 implies it should be different. when i try xdg-settings set default-web-browser xdg-settings

set default-web-browser com.github.Eloston.UngoogledChromium.desktop like the issue says, xdg-settings get ... still returns ungoogled-chromium.desktop! how can i fix it?

Title: I can't set Ungoogled-Chromium as default browser · Issue #1929 · ungoogled-software/ungoogled-chromium · GitHub

1929 – HP C1536A DAT drive errors bugs.freebsd.org/bugzilla/show_bug.cgi?id=1929

wtf paste error sorry. fixed it...

im trying to make ungoogled-chromium my default browser but the "set as" button in settings doesn't work. xdg-settings get default-web-browser returns ungoogled-chromium.desktop even tho ungoogled-software/ungoogled-chromium #1929 implies it should be different. when i try xdg-settings set default-web-browser

com.github.Eloston.UngoogledChromium.desktop like the issue says, xdg-settings get ... still returns ungoogled-chromium.desktop! how can i fix it?

Title: I can't set Ungoogled-Chromium as default browser · Issue #1929 · ungoogled-software/ungoogled-chromium · GitHub

1929 – HP C1536A DAT drive errors bugs.freebsd.org/bugzilla/show_bug.cgi?id=1929

f451: the image for 14-current tells me "distribution checksum failed". so much for that i guess

Is there a recommended way to run daemons or other long running processes as the current user? (this is for a desktop system running XFCE or KDE)

Both of those DEs have an "autostart" feature, but there isn't a clear way to restart things that were autostarted on login, for example

f451: looks like i got the proper packages the second time around. and i've even booted to a working system, so here's to you

question, daily pkg check is complaining that all python .pyc files have mismatched checksums. This started sometime this month after I upgraded ports. What's the deal? Did pkg mix up the order of operations?

Perhaps you had updated python?

I mean, I just rebuilt devel/meson and then reran pkg check -s meson, and it's still complaining. It's like it's not recording the checksums of the freshly compiled scripts during install

oh hm. there's always script.pyc, script.opt-1.pyc, and script.opt-2.pyc. It's complaining about the first two. Hm. Is this tracking 3 different, older versions for older python?

one of the build's steps said it was removing previous stale versions <_<

(this is the first time in 15+ years that something like this has come up)

What does "pkg info -x python" say?

no, all 3 versions have current timestamp, so they were all freshly built, so they can't be older

Ok

python3-3_3 python-3.9_3,2 python39-3.9.16_1 huh.

should I ditch the metaports?

I don't think they do anything, although the fact that each shows a different version number is concerning

You could; have not seen any ill effect here

Currently I am using 3.11 as the default version (when installed via ports); in /etc/make.conf: DEFAULT_VERSIONS+= python=3.11 python3=3.11 ; USE_PYTHON= concurrent

oh they actually are needed. python39 installs bin/python39, python installs the 'python' symlink

python=3.9 python3=3.9

What is the purpose of "python3-3_3"?

installs symlink /usr/local/bin/python3 and several other alternatives

seems like it's a middle step when you want python 3 but don't care about specific subversion

I am not sure, but maybe the version number on the metaport is just from the last time the maintainer had to touch that metaport

anyways, one guide I found just sidesteps the whole issue and re-registers the new pyc files... but I feel like that just covers up a defect in the build system

Very plausible

I'm not sure what the deal is, I just ran portmaster -a like usual. some recent ports system change might be causing this problem

ah, python39 port changelog and discussion on the recent commits explains it

a dev is working on a structural change, the checksum mismatches were a todo for later, yesterday the change was reverted due to additionally discovered issues.

ultramage, Much appreciate the update

.oO( Ooh. "uname" says last world install was c 202211; time to update seems )

that's not so bad. I'd wait for march/april freebsd point release

I suppose I could wait to update -STABLE. On -CURRENT, the dreaded "IGNORE_OSVERSION" prompt from "pkg upgrade" makes me take a wild guess at relatively safe commit to reinstall world.

Instead of being hounded by "pkg upgrade", I should likely plan, switch to -STABLE

Some time ago, I had installed a 13 world (on 14 host) in a new boot environment but there were issues with booting. (Just remembered that)

bsd.to/dV8n - trying to reinstall py-setuptools makes it crawl into old python port subdirectories, which do not exist since it's a sparse checkout. there is explicit code saying it handles sparse checkouts.

Title: dpaste/dV8n (Plain Text)

eww

# Mk/Uses/python.mk includes this file, don't remove it!

what I don't get is that it starts crawling 3.8, but ports start at 3.7 (which I also don't have) and setuptools makefile USES python:3.7+

also ignoring that my forced default is 3.9

_PYTHON_VERSIONS= 3.9 3.8 3.7 3.10 3.11 2.7 # preferred first

this makes it seem like it's going to check all of them in this sequence, and it's erroring on the 2nd step. But I never had to have all these ports directories before.

sigh. well, I patched uses/python.mk to always check path before including, never require it to exist. hopefully that doesn't change any other logic

Anyone here can explain to me how to get started with NetGraph? I want to replicate my epair/bridge network using ng

antranigv: i'd start from the example scripts, study what they do, then move them slowly over to something more familiar

Title: freebsd-src/share/examples/jails at main · freebsd/freebsd-src · GitHub

pretty sure this can be removed github.com/freebsd/freebsd-src/blob/main/share/examples/jails/VIMAGE

Title: freebsd-src/VIMAGE at main · freebsd/freebsd-src · GitHub

all of that stuff is in GENERIC by now

meena yeah, they all are. altho we've been using VIMAGE before 10.0-RELEASE.

meena dteske's code is awesome, but it's not as simple as it feels :D

meena are you using epair/bridge or netgraph?

it's really dense

epair

your code mostly ;)

if you find ng to be better, I'll jump ship

I need to change my code to always generate vnet0, vnet1 instead of epair126, epair156 etc device names

meena the only use for ng for me right now, is for bhyved jails. since bhyve supports netgraph, and netgraph interfaces can be vnet'd, it would be the ideal situation.

meena however, I am modifying jail.conf.d so support /etc/jail.conf.d/www0.conf to be run, while ignoring /etc/jail.conf.d/www_old.conf.disabled to be ignored. without the need to specify jail_list in rc.conf

s/to be ignored//

also, in Jailer, I'll be either using `ifconfig epair create` without the need to specify the epair ID, OR just set names and be done with it :D

i.e vnet.inetface = ${name}0b and ${name}0b stays at the host

while ignoring /etc/jail.conf.d/www_old.conf.disabled ⬅️ don't do that

only include /etc/jail.conf.d/*.conf to begin with

meena yeah! that's what the code does :D what I meant to say, that instead of playing around with jail_list, if jail_list == "", then it will boot everything in jail.conf and jail.conf.d/*.conf

hmmmm… re: "dteske's code is dense" — i wonder if that's why she favours forth / if forth influences her other programming style

antranigv: good.

is there a way to make the daily and weekly emails only be sent if there's a problem, like in openbsd?

What is or isn't "a problem" is best defined by the person reading the emails, I think.

But see -n in crontab(5).

Oh, wait. periodic, not cron. I need more caffeine, looks like.

See periodic(8) and specifically *_show_info and *_show_badconfig.

V_PauAmma_V: thanks

Does anyone here use `jail_parallel_start` ?

Most of my jails have some sort of dependency graph, so it doesn't help me. :/

love finding more bugs, while trying to debug one bug… or discovering a lot more bugs after fixing one, but especially, i love both of those things at the same time

isn't that the nature of software

debdrup you were the one who reported that depends doesn't work with jail.conf.d, right?

antranigv: as far as I remember, first time I heard of it was from you

ah, I see

well, I fixed the directories issue, so no need to specify jail_list anymore. have a final bit to fix as well

Oh, nice!

What about being able to specify common values in jail.conf and only per-jail values in the jail.conf.d?

As in, is that something anyone has brought up? Because I've noticed that that could simplify a lot of things.

debdrup yes, many people did bring that up. that is unfortunately part of jail(8), and not rc.d/jail. I'm trying to understand if it would be better to either "hack" that feature with rc.d/jail or actually patch jail(8)

antranigv: ah, okay - I had the thought that if it's jail(8) that needs the patch, maybe it could be solved by having jail.conf include the jail.conf.d?

So it essentially gets treated as one file, rather than several files that get loaded sequentially.

jail can read from stdin, so that might be an option, lemme check

I'm just throwing stuff at a wall to see what sticks, so don't worry too much about it :)

yeah, `cat /etc/jail.conf.d/* /etc/jail.conf | jail -f - -c` seems to be a thing ;;)

Now the question is, will that cat make it so that the common values in jail.conf also work for the jails in jail.conf.d ?

mwl defines them as "default settings" in this slide: youtu.be/S3u8OtjfGFE?t=811

Title: 20 Years of FreeBSD Jails (2019) - YouTube

the variables outside of the brackets are what i'm talking about

yes it does.

oooo

I think I will rewrite rc.d/jail to support pipe

it will make everything easier

debdrup thanks for the tip <3

it absolutely will

being able to define a jail just by including the name, its ip address, and optionally its depdends variable would be absolutely fantastic

path and hostname variables in the mwl example can also be made into globally scoped variables by using $name if memory serves

oh right, he covers that very thing a few seconds later: youtu.be/S3u8OtjfGFE?t=863

Title: 20 Years of FreeBSD Jails (2019) - YouTube

debdrup run `jail -f /etc/jail.conf.d/j20.conf.disabled -e '%' | tr '%' '\n'` and watch the magic :D

4 lines of change

I wonder if this fixes depends

Well, it does.

for like 6 months there's been a bug in the ansible zfs module on freebsd, termbin.com/vlffa anyone else run into that?

For those who want to test the latest changes in rc.d/jail that fixes many issues. The complete script at reviews.freebsd.org/P562 and the patch/diff at reviews.freebsd.org/P561

Title: ✎ P562 rc.d/jail by antranigv

omfg ty for working on jails!! i'm not using them yet but reading everything i can find and trying to learn

VVD: hey is that you responding in QW-Group/mvdsv #120 ? If so, thanks for the support. I imagine a solution would have to be cobbled together for the FreeBSD port as well as the OpenBSD port I maintain as well.

Title: Source tarball missing protocol.h · Issue #120 · QW-Group/mvdsv · GitHub

pertho, yes, it's me

antranigv: woot!

antranigv: I love the notion of inheriting base configs. Good deal.

mason: it's a thing that's been percolating in my brain for quite a while

VVD: ahh.. the FreeBSD port for games/mvdsv looks a little old..

I tried getting some people interested in it during lockdown thinking someone might have time for it, but no dice

So it's nice that I finally got someones attention :P

polyex your welcome :) least I can do

pertho, yes, but maintainer is danfe

antranigv: those are pastes, not reviews though

antranigv: do you know about git arc? freshports.org/devel/freebsd-git-devtools and freebsdfoundation.org/wp-content/up…reeBSD-Code-Review-with-git-arc.pdf

Title: FreshPorts -- devel/freebsd-git-devtools: Simple Git tools from the FreeBSD src tree

VVD: git submodules are horrible :/

pertho, ye, I create several patches for other ports with sources from gh with submodules

hardcoded in Makefile 10+ lines with "GH_TUPLE="

antranigv: I'd suggest getting a git patch (or a review, or a pull request on github - just something more than diff(1)) and sending a message about it to freebsd-hackers@

antranigv: that's also much easier to signal boost than a couple of pastes.

Ideally most of the testing instructions you need done should go into the body of the email rather than the diff, along with an explanation of what everything you're trying to accomplish

github.com is still ipv4 only *grumble*

means that if theres something pkg wants to fetch and it's not in freebsd distiles, it tries to fetch it from github then fails, on an ipv6-only poudriere

is there a way to have vm-bhyve wait for nfsd to come up before starting vms?

ah, boom, it's right there

vm_delay="5"

maabye

actually i dont think that's it

that's time between starting vms

a rc.d script that has a REQUIRE: NFSD

well, with appropriate caps

i think i'm talking nonsense actually

its the guest that needs to wait before starting docker images

wrong channel, wrong everything :)

Hi, has anyone tried running a kind of NAS on a raspberry pi (3b+)? I am interested in hosting a small server for my family using ZFS, but I have doubts on whether it is powerful enough or not.

will usb attached disks be too slow?

Too slow? ...yeah I guess an hdd passed through usb is pretty slow

Having had general bad experiences with USB attached disks I now avoid them. USB disk adaptors just has not been reliable for me. Always a source of problems.

rwp: How bad were your experiences? Is your deployment something large like a company or just a home lab?

luca: nextcloud on freebsd runs well with zfs on external usb3-connected disk with rpi4(8GB)

by 'runs well' i mean also clocked to 2.1GHz and postgresql and nginx

f451: Nextcloud sounds heavy. Could I get away with setting up everything by hand with the cli? Also my rpi3(1GB) kinda pales in comparission to yours ;)

things requiring fast maths like ssl and encryption are going to be slower because theres no aesni in hardware, but it's fine for file storage

id not run it on a pi3 not unless you like waiting, though there is a nextcloud for the pi, think it's linux based tho

but yes on the pi4 everything was set up in cli

Thanks, I'll try it and hopefully everything will work fine

nextcloud will be slow on a pi3 cos of lower overclock, less ram and usb2-only external hd

if you google for picloud or similar im sure you'll find it

I have one last worry though. I don't run FreeBSD on my main machine. If the rpi dies will I get locked out of my files (at least until I get another FreeBSD device)=

if the rpi dies then swap out the microsd

not sure if you can mount ufs on linux, it depends i guess on what distro

But if I cared most about portability I probably shouldn't use ZFS, right? Maybe NTFS or FAT32 would be better

for freebsd?

Portability of the HDD

So I can use it on other devices adhoc

yeah fat32 or ntfs

so - boot from microsd, access the ext device via usb2

it'd be fine for that. i have pi2b+ configured that way

no nextcloud ;)

Glad to hear it's been tried in the field :)

you might want to run nfs server on the pi. not tried that, always used either nfs client or sftp for low-end rpis running freebsd as servers

How is nfs/sftp support for windows hosts?

no idea really. sftp you can get something like winscp or putty. but it's been over 20 years since ive used windows

That's fair. It seems like samba works really well on FreeBSD though

for windows access no idea about nfs, samba server on the fbsd host would be used for that. no idea of overhead though

also will it be internal behind a firewall or exposed to the internet cos if the latter you'll want to configure a firewall

Internal. Only inside my house

Title: Samba fileserver on FreeBSD (Update FreeBSD 12) - davd.io

says nothing of requirements. still, if it's only running sshd and samba server id guess it's enough

yeah, I've followed that. Got Samba working really well actually. I haven't connected any HDD to the pi yet though

(side note, webmin is a good tool)

yeah. after all this time im mmore comfy with the cli. also dont need to worry about a webserver

after a binary is compiled, what makes it "decide" to choose, at runtime, a particular libstdc++.so.6 version?

nb not ports, just a pre-built binary

Thanks for the help everyone. I'll head off now o/

somebody reports that the wrong gcc version is being chosen

I am guessing something like this -Wl,-rpath=/usr/local/lib/gcc10 is missing?

Drat! I see I just missed getting back before luca left. :-(

My problems were that about monthly I would find USB storage would need a full power cycle. On two different bits of hardware over time.

So... Things worked. But infrequently the USB adaptor or drive would glitch out and need a power cycle.

On the other side of USB I have found USB network adaptors to be rock solid. Go figure. Though I expected problems there they have worked great.

I haven't been running on RPis though I was running x86 systems.

For the USB network adaptors those were in random airport office locations around the state with a service product.

So just sitting there and periodically USB storage would fritz out and need a power cycle.

For the USB storage that was in my home on the server shelf untouched by me and I don't have cats to get into them either.

rwp: whats your make of power adapter?

also rpi2/3/4?

The USB storage power adaptor? Those were all no-name chinese manufacture random enclosures.

f451, Sorry but I was using x86 hardware not RPis.

are you using a UPS?

No UPS but our power here is all underground and very reliable and none of my other systems noticed anything.

id be looking at the power adaptor 1st as it's a noname version

swap it out for something known that has a guarantee

I'd be looking at the USB adaptor as those are all no-name versions. And USB is a terrible non-standard, standard.

heh

so many standards

One of the USB enclosures had a funky PS2 style round DIN connector for power. And one must be cautious of those because the pin out is non-standard there too.

I burned one up by mixing up two that had the same power connector but had power and ground flipped on the pins.

i guess the only way of proving it would be to monitor output under load. ive known 12v adaptosupplies to decide to either go bang and completely fail (good) or just not regulate the voltage enough and it breaks what is being supplied, like a scanner (bad)

If you are using an RPi then for storage of course you must use USB storage. And then it is what it is.

But when I mentioned my experience there hadn't been a mention of RPis in the conversation yet.

sorry was talking with luca before hence rpis

Strangely I have used USB network adaptors of different vendors very reliably. I expected trouble there but instead they have been rock solid over years in random service locations.

but in my experience the most likely cause for yr usb disk issues would be power

yeah NICs are better

i have ext disks on 2x rpi4s but they are powered by the pis. made sure to get the recommended psu

on x86 ive lost 5 disks due to power issues (brownouts and powercuts) which is why i was asking if you had a ups

i put off getting a ups way too long lol

We literally go for years without power interruptions here. When I had UPSes the batteries would not last as long. Had more UPS failures than power events. So at the house I just don't bother.

On the other hand at the airport offices that I mentioned those are always located out on rural electric power and if those go a month without a power outage then it has been a good month. Those always get a UPS.

hi, i have some ssd in my server. to get out full performance on those drives do i need to reconfigure anything in fbsd ? or it should work out of the box ?

i do not run any zfs or simular

n30, I think everything should work out of the box. No need to do anything.

rwp: i thought so... then i probably need to order home some new drives then ... mine is probably not healthy anymore

Are you having problems with an SSD at this time?

Some vendors are better than others. Backblaze is a useful resource of public data on different devices. backblaze.com/b2/hard-drive-test-data.html

Title: Backblaze Hard Drive Stats

They mostly have data on spinning drives but I recall SSDs being in there too.

Yes. And here is a Tom's Hardware review of their data tomshardware.com/news/backblaze-confirms-ssds-more-reliable-than-hdds

Title: New Backblaze Data Confirms SSDs Are More Reliable Than Hard Drives | Tom's Hardware

I have never had a good Intel or Samsung SSD fail for me but just in the last month a laptop running, ahem, windows, on a no-name SSD failed to boot.

Lacking other options I booted a live-boot image and ran ddrescue across the entire device until it read successfully.

Then used dd to rewrite the image back onto itself. And then it was able to boot again and all seems to be fast now.

Since I have a full image backup and therefore no risk, this is used for a flight training program only, I put the laptop back to do that task with the original SSD and it is still running.

The accepted wisdom is that SSD firmware is important and sometimes poor quality firmware won't refresh marginal bit storage on its own but we can trick poor quality firmware to refresh marginal storage back to health.