happy owner of intel i9 - 32 core cpu, ruct compile time - 2h 11m, llvm - 3h gcc12 - 3h 10m; total bild time - 4h

poudriere 8 parallel builds

notice, that rust builds also including of special version of llvm ( bundled )

^^ exactly

on my older machine it takes usulally 2 days for entire port tree

btw can I limit cpu core count used by the compiler ?

something like setting affinity?

several ways, depending

usually i'm doing nice -n 9

RhodiumToad, heard about cpuset but looks like it doing something else ...

poudriere's default is to ask individual port builds to use only one cpu, but some build scripts use more anyway

that is what i'm trying to avoid

what exactly do you want then?

limit cpu usage say only 4

4 per individual worker, or 4 total?

4 cpu core window for everything poudriere's doing

maybe with cpuset?

cpuset ( if i understood correctly ) reserve cores for specific task

not quite

never worked, read only manual and has 0 experience

it restricts a process and descendents to a specific set of cores

so it is affinity right ?

other processes can still use those cores

is it work same way as nice do ?

any child process takes same rules ?

the main problem is that at least historically, most things that detected the number of cpus to decide how many jobs to run didn't pay attention to cpuset

yes, child processes inherit the limitation

basically, if you run poudriere under a cpuset limiting it to 4 cpus, you might also need to specify -J 4 and also set MAKE_JOBS_NUMBER_LIMIT=4 in a make.conf file

my worries is that while compiling some ports, when all cpu cores are used cpu temperature ( core temperatures reported by coretemp ) is ~80 c)

I say "might" because I've not tried recently

RhodiumToad, thank you for detailed answer, I will try and reply results here

probably on this weekends

Title: Ksnip Screenshot - Imgur

hi Guys, can you please share your knowledeg, please ! Hardware. Mac Mini Late 2011 i7-2nd Gen, Network issues. Network, Wlan I have issues broadcom B4331, ethernet is Broadcom tigon. TG3 . The Goal. To use this as a home media server. I have Linux installed a gentoo flavour ( Redcore), drivers worked out of the box. The issue is using the linux drivers I have droputs, regardless if I disable each of the network device,

even tried usb wifi, still the same issues. My question is if switch to Freebsd, is bsd drivers the same as linux, & will likely to have the same issues. I'm just wondering that Freebsd code is closer to Mac, that it may work better with bsd than linux? Your thoughts would be appreciated, thanks kindly !

nerozero: is it physical 32 core cpu, or hyperthreading?

me 64 bit

baremetal

Mystified, under bsd try ifconfig <interfacename> -tso

bsd doesn't supports tso well, so in my case it was the cause of loosing packets and random disconnects

I'haven't tried freebsd yet, I do Have a copy of live nomad, I could try that.

angry_vincent, intel.com/content/www/us/en/product…-up-to-5-80-ghz/specifications.html

Title: Intel® Core™ i9-13900K Processor

that is the cpu

Mystified, you are asking question in FreeBSD channel, I would suggest to find appropriate channel to your system

nerozero: exactly the same, I'm looking to purchase, usb ethernet port to see ifit will work.

Mystified, try to look into this thread: forums.freebsd.org/threads/broadcom-bcm4313.74040

Title: Solved - Broadcom BCM4313 | The FreeBSD Forums

maybe u find something useful

nerozero: I'm going install freebsd, hopefully see if it works, but you suggest an appropiate channel, are suggesting there may be other channels. ie. broadcom network or mac devices repurposed channels. If so, I'm happy to hear your suggest ions

thanks nerozero: :)

rwp: for what it's worth I just tried booting the 13.2 release raw VM image, and all the expected files have their schg flags set

rwp: and I also tried a fresh 13.2 install from the release image and that has the schg bits too

RhodiumToad, what is "schg bit" about ? I had just updated the system to 13.2

nerozero: this is from a discussion last night

any special I should be worry about ?

nothing to do with you

:) ok thanks :)

Hey guys :) We're facing a kernel panic with FreeBSD 12.4 and have reported a bug report: bugs.freebsd.org/bugzilla/show_bug.cgi?id=270816

Title: 270816 – kernel panic in sym_hipd.c

Sadly there wasn't a reaction until now and we're kinda stuck without further help.. Is there any way to push this issues or provide further information?

that looks like a fairly old driver?

danel1: is it vmware vm with "lsilogic" adapter?

Nope. It's a VM running on OpenStack (based on Ubuntu).

and you don't have an option to use another emulated adapter?

surely you could be using virtio-blk or virtio-scsi

as i understand correct, this would lead to an other driver that would be used instead of the LSI driver which "uses" the code from "sym_hipd.c"?

sys/dev/sym is the symbios / LSI scsi driver, its sole purpose is to interact with LSI hardware; if the emulated hardware doesn't behave exactly like physical hardware it could easily cause issues.

if you can have the VM present disks as virtio devices instead, then the LSI driver would never be used

I'm assuming that this is emulated hardware rather than, say, passthrough of a physical device

Yeah that kinda makes sense.. good catch.. let me take a look and verify

hi, could somebody give me a hint, what is a proper way to create tap interface in rc.conf ?

nerozero: cloned_interfaces="tap0 tap1" (for as many as you want)

RhodiumToad, I thought that was only for bridges ....

thanks!

nerozero: and then ifconfig_tap0="..." to configure the interface

the rest is clear, thank you

remember that you can only have one cloned_interfaces line, so it needs to have everything in it

RhodiumToad, :) yes

Is it a good idea to rename interface to give them a better context ?

like ifconfig_re1_name="if_public0"

in case of big numbers of interfaces it is difficult to remember which interface doing what...

nerozero, I think that's squarely in the opinion catagory.

I do not, but I could see scenarios where it might be advantageous.

CrtxReavr, on one machine i have a bhyve virtualization and it cause quite a big mess, so i'm thinking to make some order by give a proper name to interface

is there some better naming way for that ?

nerozero, not up to me to dictate a naming policy for you.

watching 3 pages of interface listing with standard names quite difficult to digest ;)

CrtxReavr, just asking for recommendations

I mean. . . one thing I've always liked about FreeBSD is NIC names conveyed the type of device. . . em0, xl0, sf0, etc. . . those names had meaning, as opposed to linux, where everything was ethX.

in my case I'm having qiote a lot of tap0..tap15 ...

If renaming those interfaces looks like an advantage to you, then do it. . . just think about who else may be sharing in the admin of this implementation and relevant support/documentation issues.

thank you !

what is the right way to get always last in rcorder when booting and always first when shutting down, I see securelevel uses "REQUIRE: adjkerntz ipfw pf sysctl_lastload" so I was just going to go with "REQUIRE: securelevel" - thoughts?

What is the latest RELEASE 13.1 and how to check?

nmz: what do you mean by "latest"?

You mean P8 or P9?

nmz: 13.1-RELEASE-p7. but the latest production release is 13.2

freebsd-version?

It seems I'm on 13.1 p6, didn't know there was a p7 until I got a warning that my version was deprecated

nmz: freebsd-version -kru

the P's

you can check the current patch level in e.g. cgit.freebsd.org/src/tree/sys/conf/newvers.sh?h=releng/13.1#n57

Title: newvers.sh « conf « sys - src - FreeBSD source tree

Since 13.2 is out, you'll get warnings to upgrade

ok thanks

so upgrade straight to 13.2 and not p7?

I'm waiting a few more months before upgrading

nmz: yes, the upgrade directly to 13.2.

otis: Will you be able to join the GSoC meeting tomorrow?

jrm: yes, i'm counting with it.

Nice. It should be short. Google requests that we begin with such a meeting.

RhodiumToad, Hmm... "Curiouser and curiouser", said Alice. I'll install a pristine VM here and then compare that against my system. In a bit...

Curiousity, it kills more than just cats.

Only fair if dogs are also killed

curiousity used to be a leading cause of death of humans too :P

Humans, eh

look at their experiments with the atom

otis: Some GSoC contributors and mentors hang out in #freebsd-soc here on Libera.Chat.

Curiosity is why I will install it in a VM. That way if there is a problem it can only be a virtual problem. :-)

It seems my desktop has drifted somehow from what happens with a fresh install. Maybe it was the fresh install of 12 and then upgrade to 13. But that's the issue RhodiumToad and I are discussing. My system does not have schg (immutable file) flags set on files that a pristine 13 install has set.

rwp: i'm using freebsd in a vm right now!

Shh! We don't want it to know it is living in a simulation. :-)

rwp: I'm pretty, and sure, that 12 has schg