Pastebin offline??

bsd.to is down?

seems so, try using dpaste.org instead

thanks

just looking at the 14 release schedule and it just says "delayed". is there any resource with more info?

not at the moment, the openssl3 work may be near completion but it needs people to actually test it

see here: twitter.com/khorben/status/1665529570756382720

Title: Pierre Pronchery (@khorben⊙mso) on Twitter: "User of #FreeBSD, you’d like to test 14-CURRENT ahead of the release with #OpenSSL 3 in base? Give the Pull-Request at t.co/8UlXX1WUxK a shot and let us know if anything fails or isn’t as expected!" / Twitter

man, with VimDiesel you don't even need to visit twitter if that's not your bag of tea. all of the interesting parts right there inline

t.co link -> freebsd/freebsd-src #760 if you're in that group and don't want to use a t.co link

Title: Update OpenSSL to version 3.0.9 in the base system by khorben · Pull Request #760 · freebsd/freebsd-src · GitHub

if I take 4 usb external drives and put them in zfs raid 0 will I get cmobined speed? usb 3 ports and compatible drives

will a saved fiel be broken into 4 sets of blocks and saved partiall o each of 4 drives?

so when reading it back it will be combined speed?

or no?

concrete_houses it depends on how you access the data, single threaded read from one application versus parallel reads from multiple applications. + block sizes for reads.

Hi, folks! Is there any possibility that chromium with pulseaudio support is being adopted in the ports (as flavor?) someday? There are a lot of info/threads out there pointing to "compile yourself" - which takes hours...

I mean, as far as I can see it's been there, but also got removed again.

And there's an open issue from 2020 with that topic -> bugs.freebsd.org/bugzilla//show_bug.cgi?id=246449

Title: 246449 – www/chromium: enable PulseAudio by default.

or - it is a little confusing to me the sndio/pulseaudio/oss/etc. topic... is there some high-level info on this from where I can start to dig deeper?

picolisp run on freebsd?

doesn't even compile

concrete_houses: the docs say it just needs a POSIX system but clearly the developer is wrong

hi all

tsoome: have you had a look at Hetzner's arm64 VPS yet? they are dirt cheap, but we don't get any output on efifb, and rather than fixing that, someone is working on reviews.freebsd.org/D40094

Title: ⚙ D40094 Add a virtio-gpu 2D driver

meena0 no output at all or no output when kernel is started?

I think no output at all

what firmware does it run?

no idea, hetzner.com/press-release/arm64-cloud

Title: Hetzner introduces arm64-based cloud serversHetzner introduces arm64-based cloud servers - Hetzner Online GmbH

at least some ampere servers have UEFI, but UEFI should not be an issue....

ah. if it has pure virtual gfx, then it is problem

current loader is excluding GOP instances without mapped framebuffer (because efifb in kernel needs it)

once this D40094 is usable, this requirement is not true and loader does not have to exclude such GOP

let's hope Hetzner will support FreeBSD officialy soon

mage: I'm working on cloud-init, and having better cloud-init makes it so there's less excuse to provide images

tsoome: that review might be usable, but it's not mergable

the limit in loader is set with this commit: 02732f945ed2ec2b4fd03421923720608b28a615 loader.efi: Do not use as frame buffer BLT-only GOPs.

basically we need to allow PixelBltOnly and for BS off case, we need to be sure we will not attempt to write directly to framebuffer (as there are none)

tsoome: how do we know for sure there's no FB? rather than we're trying to find it in the wrong place?

there is no "wrong" place in a sense that UEFI api does give us a list of protocol handles, and thats it.

of course, there may be the physical frame buffer is there, but not mapped for GOP

hello

I'm trying to set up a linux jail not a chroot and I've searched a lot on how to do jails but none of them detail a linux install. Is it almost the same as a bsd jail?

hyvoid: i successfully used bastille to setup ubuntu jails

using their instructions

Has anyone ever had the phenomenon that when creating a bridge interface and adding a tap member to it that after that operation the bridge interface disappeared?

never seen that happen

running FreeBSD 13.2-RELEASE

can I debug that somehow? /var/log/messages isn't very helpful in that case

Demosthenex: thanks, kind of want to try a different linux but ubuntu should do for now

modify syslog.conf add a line like ... "*.* /path/to/file"

when unsure ... log everything ;-)

then just hup or restart syslogd

thx - I'll try that

also if you have a network manager monitoring your interfaces it might just be removing them for you "maybe"

Hey. I have been trying to reconfigure my FreeBSD install to use a keyfile instead of the passphrase setup from the installer. I have added the keyfile to slot 1 on the geli devices, but haven't been able to get it to boot unattended. It still asks for the passphrase.

CmdLnKid: I cannot remember that I've installed something like that

hyvoid: maybe use linux compat and just use a freebsd jail?

stuff like kde or gnome do stuff like that so figured it would be worth mentioning.

so... what the logs now show is the following entry: "ntpd[31565]: Deleting interface #6 bridge0, 10.0.0.1#123, interface stats: received=0, sent=0, dropped=0, active_time=20 secs"

why is my ntpd deleting interfaces? XD

i don't think thats whats happening there, could be wrong but believe that message is that its just removing the interfcae from being listened on

it's deleting its internal knowlege of the IF, probably because the link went down

^^

hyvoid: should be the same, just put a linux system in the dir and have the linuxemu stuff loaded (but I never tested it)

nfs export / from another linux system onto /usr/compat/linux ;-)

well, here's what happened...: pastebin.com/axvRcr3K

Title: Jun 6 14:51:06 bsdbook sudo[70157]: userx : TTY=pts/0 ; PWD=/usr/home/userx ; - Pastebin.com

(the question was about a jail and not just linuxemu or a linux chroot)

and I can repeat it: add bridge, addm tap to bridge -> the same sequence

not sure why adding a tap interface to a bridge kills the dhclient for my other interface (lagg0)

and not sure if this has anything to do with it

ifconfig bridge187 create; ifconfig tap187 create; ifconfig bridge187 addm tap187

not getting the same thing here at all

not using dhclient though

and no lagg iterface either

FreeBSD 13.2-STABLE stable/13-n255450-0644746d5091

hello guys, one very basic question. How do I download freebsd sources ? I have to build virtualbox kernel module.. so I need only kernel sources..

handbook dude

CmdLnKid: thanks

git or svn knowlege needed

all releases usually also include the source

hmm... do you think that dhclient or lagg has something to do with this behavior? (I use lagg with failover for em0 and the wifi card which get IPs via DHCP)

unsure of that. i haven't seen this behavior before

okay... It's getting weirder

i believe there used to be a tunable sysctl for handling unused tap and tun interface where they would be removed when unused but its been a long time since i had to visit that

I have created bridge and tap as you did and the bridge still exists XD

only thing i can think of is if the address thats assigned to the lagg collides with the tap or bridge interface maybe the interfaces are being autodeletedd

so, when executing "ifconfig bridge0 create; ifconfig tap0 create; ifconfig bridge0 addm tap0" the bridge0 interface gets destroyed. but executing "ifconfig bridge1 create; ifconfig tap0 create; ifconfig bridge1 addm tap0" works...

the only difference here was the bridge interface number

interesting.

you have any definitions defined for those interfaces in your dhclient.conf or rc.conf

?

"0" or "ifconfig bridge0 create; ifconfig tap0 create; ifconfig bridge0 addm tap0" works fine in this setup

my dhclient is empty... but yeah, I create it in rc.conf with 'cloned_interfaces="lagg0 bridge1 tap0"' and setting it up with 'ifconfig_bridge0="inet 10.0.0.1 netmask 255.255.255.0 addm tap0 up"'

and I was curious because after the system start there was no bridge0 - although I could see it within dmesg

placing a pretty good bet it has something to do with dhclient. id say at least try to statically assign right now without dhclient and disable it temporily just to diagnose it.

sorry, the cloned_interfaces previously should have been 'cloned_interfaces="lagg0 bridge0 tap0"' (I've already changed it to try to work with "bridge1" XD

wait... gotta restart

lol

quick everyone /quit

;-) itll seem like a dream

ok, now I changed my rc.conf to create bridge1 and set it up (inet and addm) and it disappeared as well... but now I can create bridge0 and set it up (inet and addm) and it keeps existing

and when I setup a bridge1 manually (inet and addm tap1 - newly created) the bridge1 disappears

I don't get it

so, I can setup a bridge manually when the system has started, but not within rc.conf (at least on my system)

dhclient stuff or rc code seems to be destroying the interfaces

CmdLnKid: yeah, but only the bridge that is being setup with rc.conf... manually setting up another bridge (with another interface number) works

I'll try to set a static IP on the lagg interface and disable dhclient

(which wouldn't be the production usage for me, since I am switching wifis on a regular base for work)

sounds like you are in an admin role at $work

nope, I am an analyst with a few different customers. and when I'm working on-site I use their wifi infrastructure.

ah i gotya

or ethernet - whatever is available

(hence the lagg)

sys and net engineer here, usually my FE is macOS and my tools are within VB on FreeBSD

yeah, I am desperately trying to get rid of macOS - too much proprietary magic happening there

I mean, I am an admin for my own infrastructure (mostly proxmox cluster with OpenBSD server as VMs) - but I am trying FreeBSD as a daily-driver desktop

BSD are much more predictable to me why i don't rely on homebrew or macports as good as they are i don't like depending on apple providing what i can do with my networks or systems

i employ smartos for my home system/network needs ;-)

early sunos guy too

runs freebsd fine... nothing wrong with proxmox tho

so, restart to try static ip config

well, ... no success

now I can do the same thing - bridge disappears, but this time without the dhclient logs in the logfile

same situation: bridge1 is defined and setup in rc.conf, during the startup the interface disappears (but I could see it during the startup logging) - and when manually creating bridge1 disappears but bridgeN (where N is not the interface number from rc.conf) it works and the bridge keeps alive

interesting

i have instances where this would be a problem in development but i cannot repeat it for the life of me

is this on physical hardware or virtual ?

it's physical hardware - a T480

is it a problem that I define the bridge in the rc.conf within one line? -> 'ifconfig_bridge1="inet 10.0.0.1 netmask 255.255.255.0 addm tap0 up"'

when I ifconfig manually (inet and addm) only the member gets assigned but not the IP address is set

I mean, with ifconfig manually I have to 1.) addm the tap and in a dedicated ifconfig command set the inet

should the rc.conf reflect this as well (with multiple "ifconfig_bridge1=" entries)?

can you file a bug report at bugs.freebsd.org with a uname -a; cat a boot message log its somewhere is /var can't remember off the top of my head at the mmoment

/var/log/dmesg.today

ofcourse scrub that of anything you don't want in there

yeah, regarding dmesg... somehow I think in the middle of booting it stops logging into dmesg

it seems (to me) that there are some logs missing

I mean, I can see more logs during the boot (until the login manager starts) than what "dmesg" reports

compare the release syslog.conf and syslog.d directory for anything you suspect missing

but yeah, I'll file a bug

i have not read the whole backlog, so please ignore me, if i misread the problem; just to throw it in: i have the inet for my bridge as alias0 since forever (and no comment on why); maybe something to try if you are desperate

good idea "alias" seems there is a sidecase bug here and really starting to lean on the rc subsystem as the origin. could be wrong tho

maybe devd

ridcully_: you mean, having one line 'ifconfig_bridgeN="addm tap0 up"' and one line 'ifconfig_bridgeN_alias0="inet 10.0.0.1 netmask 255.255.255.0"' (does the _alias0 also needs "up"?)?

(where N is an arbitrary interface number)

RhDoc: correct. and the up is with the addms (not in the alias)

alright... let's give it a try --> rebooting

well... it was worth a try

same behavior

anyone know the trick to make chrome not show blank white screen? is there some cache or lock file that I can delete to make it work again?

uhm #chrome

um #freebsd

uhm #chrome

was workin fine 2 days ago

im on freebsd 13.2 amd64

what'd ya change ?

nuthin

try clearing your catche, history, resetting to defaults

can guarentee that in over 10 years of freebsd usage that a change in the source has not effected chrome displaying a white page

spelling mistakes intentional

okay, thank you for your support, but the alias didn't work either -> I'll file the bug and see what happens next

there is at least a manual workaround XD

try to keep that visible in #freebsd-bugs

alright, will do

Why does the passphrase prompt still appear during boot after configuring keyfile locations in loader.conf? I've had a look around on the forum & manpages, but haven't been able to figure out what I've missed :/

rhdoc, may be worth while emailing network⊙fo, freebsd-network⊙fo with the bug report as well

reboot fixed chrome yay

anyone ever try and make a browser based on nfs? not http?

webNFS kinda thing?

CmdLnKid: sorry if it's duplicate, but I haven't seen in the logs if I already asked you (since I rebooted again): what component would that bug be: conf, kern, bin? what would you suggest?

hi there, what is the optimal size of ccache for poudriere ?

just in case

CmdLnKid: alright - I've assigned it to "kern" component, since other bugs regarding bridges were also assigned to it, and I've sent it to freebsd-net⊙fo mailing list (the addresses mentioned by you didn't exist according to the mailer daemon)

ntpd[667]: ntpd: daemon failed to notify parent!

ok, what's that all about

(latest stable/13, which I notice has a new ntpd version)

whos got a high volume website with varnish and ha-proxy running? bonus if using smalltalk picolisp haskell forth werc.cat-v.org swi prolog c gnoga ada or anything besides java and .net????? and extra points if no database

concrete_houses: I've worked in a shop that developed in Haskell, but, honestly, mostly so they'd have their maintenance contract in the pocket

i actually wrote a blog entry on freebsd yesterday

puddinghead: link?

so the /bin/sh is the old bourne shell in freebsd?

no

it's a posix shell

I am starting to like that there is so much to choose from, ksh, oksh and mksh which I was using up until recently, now I am slowly replacing that because I see that development happens more on ksh

(what it _isn't_ is bash or ksh or zsh)

so /bin/sh on freebsd is something completely different from all others?

... no? it conforms pretty well to the POSIX spec

yes

it's not the same codebase as ksh or bash or zsh, mostly for licensing reasons

so FreeBSD had this own from the start and only NetBSD and OpenBSD did have ksh as standard?

I believe some other platforms have or had it under the name "ash" (Almquish shell)

see

*Almquist

yes, that may be the case

(after the original author)

ash was also used in this openwrt minimalistic busybox distros I think

it's much lighter weight than bash or zsh

well, that's what I am aiming for, I do not like heavyweight shells

I don't know when ksh became more freely licensed, it used to be proprietary

that's why I am more fan of those ksh stuff

(I used it a lot on commercial unixes back in the day)

pkdsh

or pdksh

pdksh was a rewrite, not the original, no? I remember trying it out in the early days and finding it lacked major features I was familiar with

so when I switched from commercial unixes to freebsd, I ended up mostly using bash for interactive shell

later I switched to zsh in mostly ksh-emulation mode

well yes

that's what I noticed today too that zsh is a continuation of ksh

what do you mean by commercial unixes? AIX and Solaris?

plus half a dozen others, yes

Almquist rewrote a Bourne-compatible shell because of the UNIX licensing, and the BSDs adopted it back in the 90s.

Early 90s, I mean.

in approximate order I have used umax, ultrix, aix, hp-ux, sco (sco unix not xenix), one or two random svr4 ports, solaris (2.5 through 9), and I'm probably forgetting some

Net/2 in '91 included it.

..although that version doesn't have history or line editing, so good luck.

ksh is a favorite of mine. Learned to use it on Fedora 5, a very long time ago, and have run across it in various old distributions on miscellaneous ancient hardware.

I think the only really widespread commercial versions I missed were irix and dec's later versions (tru64?)

I usually use ksh98, I do remember trying pdksh but don't remember much about it

true64, I think

ksh was where a bunch of the posix shell features started out, and most commercial versions include it specifically to provide a posix shell

I've yet to interact with Solaris. I've never really been fortunate enough to get my hands on SPARC hardware, only x86 and itanium stuff.

from sun, that is.

eh, you didn't miss much. (and sparc is, as far as I know, dead now)

If you can find SPARC hardware now, Illumos will be the thing to use on it.

There's still stock to buy, but there won't be any new versions iirc.

Yes, IIRC Oracle killed it after buying up Sun.

Not quite.

apparently fujitsu is still making them but is phasing them out

Fujitsu was the last hold-out.

RhodiumToad: like I said, they'll sell you what stock they have.

Oh I see. I may look into that, I've been meaning to play around with SPARC for some time. I have a nasty habit of collecting dusty old dead architectures.

The newest supercomputer Fujitsu did was ARM based, though - and that's several years old at this point.

I have a 9-track tape here somewhere that I'm due to write NetBSD/vax11 onto soon

things like branch delay seem ... quaint, these days

Not too long ago I passed on a SPARC server, because for the amount of power they suck down they aren't _that_ fast.

Hey, if you can't measure your computer's power draw in tens of amps, is it really a computer? :P

a long time ago when I was at Supernews, we ended up giving away a bunch of them, E4500s as I recall, 8 cpus each

fonks: if that's your benchmark, the way to go is a decommisioned mainframe.

I didn't realize Oracle killed Sun. I thought they bought Sun was because Oracle was predominant on Sun hardware. wow

..assuming you can get them, they're not easy to source even if you want one.

eh, you want a mainframe, just run hercules

kindred: Larry Elison had to find some way to be even more evil.

we're not bitter, us Ex-Sun people, oh no

RhodiumToad: but how do I get the actual hardware

do you have anywhere to put it? :-)

I had one mainframe interfacing job a long time ago, making some unix system talk to a mainframe via an emulated card reader/card punch

I'm not even an ex-Sun person and I'm bitter about it. Sun made some cool stuff and Oracle killed most of it, and ruined the rest :/

jgh: Your initals made me check one of my old references; did you work at BBN once?

RhodiumToad, that sounds like my kinda fun. What kinda architecture were you dealing with?

nope, sorry

Ah, fair - then I don't recognize who you are, sorry. :(

In my defense, I only remember the initials jgh because it came up today

heh. Were BBN in Boston? I did get there once, for about three months... back in the m88k days

Bolt, Beranek and Newman - responsible for TCP, but not for making it fast.

fonks: iirc, that was an ES/9000, at the time about the top end of IBM mainframes

aye

Now there's one I've not heard of in a while, jgh. I'm quite partial to m68k but there's basically no consumer-facing m88ks I've ever found

jgh: can I ask what you worked on at Sun? I'm always interested to pick peoples brains about what they did there.

it didn't do well, which was kind of a pity

Sun - originally FT systems, then networking

Wasn't that the CPU Data General Aviion (unsure of capitalization) workstations used?

yes

Yes. That's the only system I know of that did offhand

There was a UNIX workstation-y thing IIRC, but the name escapes me.

I worked on an SVr4 port for one, but I don't recall the client. The exception-handling was a real pain

When isn't exception-handling a real pain?